20051690 Ver 1_Closeout Request_20120710

10 July 2012
Todd Tugwell
Special Project Manager
Wilmington District
US Army Corps of Engineers
11405 Falls of the Neuse Road
Wake Forest, NC 27587
Dear Mr Tugwell,
Introduction and Background
Carbonton Dam (the Project) was removed in late 2005 and early 2006 to provide stream
mitigation for the NCEEP As indicated by the results of the Project monitoring program —
from 2006 - 2010 —the removal has provided numerous ecological benefits to the Deep
River and its tributaries Among the most remarkable improvements is the re- colonization
of the Cape Fear Shiner (Notropis mek►stocholas), which has now been observed in
numerous places within the Deep River as well as the two largest tributaries within the
former impoundment— McClendons and Big Governors creeks Additional benefits within
the former impoundment include (but are not limited to) increased dissolved oxygen
concentrations and improved habitat for lotic communities
As you know, the close out process for this particular project has been somewhat
protracted During the initial close out site visit on October 13, 2011, the various
participating agencies expressed several concerns about the monitoring program
Specifically, there were concerns the Project's monitoring program design focused too
much on the main stem of the Deep River and its major tributaries and not enough on the
minor tributaries We need to reiterate for the record, although I'm certain you are aware,
this monitoring program was designed, with agency input, several years ago (dating back
to 2002 when it was proposed as a mitigation bank) While we do not disagree with the
agencies concerns regarding minor tributaries, we hope you recognize the monitoring
program was implemented as designed and approved
One of the major objectives of the October site visit was the agency investigation into the
suitability of the Project's major tributary habitat to support rare species (the Cape Fear
P 0 Bov 29921 1u tin l X 78755 512 2 113779 www i ivei b inkeLosystenis coin

Shiner and various mussel species) The NCWRC and USFWS visited reaches along Big
Governors and McClendon creeks during the site visit No rare species were observed, and
WRC's conclusions regarding the development of appropriate habitat were summarized in
Travis Wilson's email to you (copy enclosed) This email communicates a position that the
Cape Fear Shiner is not likely to use McClendons Creek, however, since appropriate habitat
was available for rare mussel species, future colonization for rare mussels may occur
Regarding Big Governors Creek, Mr Wilson concluded that suitable substrate was not
observed during the field visit He suggested that as silt /clay sediments are transported
out of the system, habitat would continue to improve, supporting rare species
Other concerns were raised regarding tributaries for one, Line Creek is currently
impounded by beaver dams and therefore not free - flowing, as it has been for several years,
and second, the hydrological statuses of the project's smaller tributaries may have shifted
from perennial to intermittent following dam removal, thereby presenting credit
implications
Setting the tributaries issue aside for a moment, you also raised three other issues I'd like
to summarize here 1) a NCWRC boat ramp within the former impoundment was
"stranded" following the dam's removal, the loss of which has not yet been offset, 2) the
park was supposed to have informational signs, which were not posted as of the October
site visit, and you wondered whether work had actually been performed at the park that
would qualify the effort for bonus credits, and 3) you were unaware of what was provided
and produced as part of the project's academic research bonus criterion
Accompanying this letter are various materials documenting the work accomplished
relating to the project's bonus criteria and to the stranded NCWRC boat ramp In addition,
we propose a close out approach that will eliminate the need to further investigate ALL
minor tributaries, the crediting of which was raised as a potential complication during the
October site visit
Recent Project Developments
Since the October 2011 site visit two separate observations of Cape Fear Shiners have been
made and documented by Tim Savidge of The Catena Group (02 December 2011 and 23
April 2012, enclosed) The first observation was of a single individual identified by Neil
Medlin of NCDOT in McClendons Creek in the vicinity of SR 1006 on 22 November 2011
The Catena Group at the same location on 12 April 2012 later observed eleven Cape Fear
Shiner individuals That same day, The Catena Group observed 5 Cape Fear Shiner in Big
Governors Creek below the SR 1625/1318 crossing Therefore, since the October 2011 site
visit, 17 Cape Fear Shiner observations have been made in McClendon and Big Governors
creeks within the former Carbonton Impoundment, indicating these tributaries are
providing habitat for at one protected species
With the help of Mr Travis Wilson of the NCWRC, Restoration Systems was able to figure
out what was expected of it in order to replace the lost recreational benefits of the stranded
boat ramp It appears, based on Shannon Deaton's letter (29 November 2011, enclosed),
that the issue was left unresolved for several years, but that this was not the fault of
2 CARBONTON DAM REMOVAL CLOSE OUT

Restoration Systems Ms Deaton concluded that a contribution of $20,000 to NCWRC
would be adequate to offset its expenses to establish an alternative boat launch at a nearby
Deep River Park Association park
Regarding the second issue you and the agencies raised outside of protected species habitat
during the October site visit (park signs and work performed), Restoration Systems has
created and installed two informational signs (pictures enclosed) One discusses the dam
removal project and the realized ecological benefits, while the second discusses history in
the Carbonton area To provide you with some concept of the work performed to construct
the park, we've provided you with the park's site plans (enclosed) As you can see
extensive work was performed at the park, which required an additional land purchase
outside of Restoration System's commitment to the NCEEP We've also enclosed a copy of a
letter from the Deep River Park Association (owner and manager of Carbonton Park) to the
NCEEP as testament to the recreational benefits the park has thus far provided
One final issue to discuss before addressing the outstanding tributary concerns is that of
academic research Restoration Systems and the USF'WS provided unrestricted gifts
(totaling $170,000) to UNC's Department of Environmental Sciences and Engineering
While the gifts were unrestricted, it was asked that the funds be used to study the
environmental implications of dam removal, studying Carbonton and Lowell dam removals
specifically These funds were used, along with other grants, to support the research
activities of Dr Riggsbee and Dr Julian as well as numerous undergraduates and high
school students seeking field research experience In all, relating to Carbonton, these funds
resulted in 2 dissertations and 4 peer - reviewed publications involving the Carbonton reach
of the Deep River The peer - reviewed publications have appeared in some of the more
prestigious scientific, engineering and water resources journals (Science, Journal
Geophysical Research and Water Resources Research) Each of these articles (enclosed) not
only expanded the understanding of dam removals, they also provide insight into broader
fluvial geomorphic and biogeochemical processes as well national environmental policy In
all, we are proud of this public- private partnership and its resulting scientific contributions
Now to the tributary concerns while we acknowledge the IRT's concerns on this particular
issue are legitimate, we do not believe this needs to occupy any more time and energy We
propose here what we believe to be the most practical solution to rapidly close this
successful mitigation project out Our proposed approach is fairly simple credit the Deep
River, McClendons Creek, Big Governors Creek and Little Governors Creek portions of the
project to their maximum potential per the 2004 Dam Removal Guidance (69,301 SMUs),
and combine with the maximum bonus criteria credit potential (25,334 SMUs)
Collectively, these two sources of credits are adequate to compensate for Restoration
Systems' commitment to NCEEP of 90,494 SMUs Meaning, if this solution is accepted by
the IRT, nothing more needs to be done regarding the project's smaller tributaries (Line
Creek, Lick Creek and Tributaries 1 -11) This approach will result in a surplus of 4,141
SMUs, and RS is willing to forego any rights they may have to sell these credits
CARBONTON DAM REMOVAL CLOSE OUT , 3

Conclusion
We trust the information contained in this letter and its enclosures will be adequate
support for you to close out the Carbonton project with relative expedience As most
agency representatives said during the October site visit, this project has been remarkably
successful on several levels Therefore, there is no need to further delay the Project's
closure The most practical way forward for all parties at this point is to focus the credit
accounting efforts on the most successful portions of the project (Deep River, major
tributaries and bonus criteria), while removing the minor tributaries from any further
credit analysis This approach will allow for a nearly immediate close out of Carbonton,
giving the agencies and Restoration Systems much needed time and resources to dedicate
to other projects
As always, please feel free to contact me if I can provide you with any additional
information or to answer any questions you may have It has been a pleasure working with
you, and I'm hopeful we can come to a swift resolution to the project's outstanding
questions and concerns
Sincerely,
)AII,
J Adam Riggsbee
Enclosures (11)
4 CARBONTON DAM REMOVAL CLOSE OUT

From John Preyer <jpreyer@restorationsystems com>
Subject FW Carbonton field review October 13
Date October 14 2011 12 32 49 PM CDT
To George Howard <george @restorationsystems com> Randy Turner <randy @restorationsystems com> Worth Creech
<worth @restorationsystems com> Adam Riggsbee <adam @riverbankecosystems com>
Just got this
Original Message
From Wilson Travis W [mailto travis wilson @ncwildlife org]
Sent Friday October 14 2011 1 26 PM
To Tugwell Todd SAW
Cc John Preyer Ellison Michael Haupt Mac Pearce Guy Russell Penann Fritz Rohde McCorcle Justin P SAW McLendon Scott C SAW
Baumgartner Tim Eaton Larry Mike Wicker @fws gov tsavidge @thecatenagroup com Matt Cusack (Matthew Cusack @atkinsglobal com)
(Matthew Cusack @atkinsglobal com) Heise Ryan J Cox David R
Subject Carbonton field review October 13
The comments below are in response to WRC observations from the field assessment on October 13
Big Governors Creek We reviewed this site from the crossing of Steel Bridge Road to the confluence with the Deep This area should give an adequate
representation of the Big Governors and Little Governors stream reaches It also encompasses the monitoring site referenced in the reports
Restoration of appropriate aquatic community (Big /Little Governors Creeks) Data collected during monitoring year 4 identifies multiple species
associated with lotic communities Although suitable lotic habitat is sparse nothing was identified during our field visit to indicate further degradation of
this area or conditions that may prevent the continued colonization of an appropriate lotic community
Restoration of Rare and Protected Aquatic Species (Big/Little Governors Creeks) Presence of T &E species have not been documented in this reach by
the Catena Group and it is not expected to be utilized by Cape Fear Shiner therefore the assessment as it pertains to T &E species focused on available
habitat for mussel species Existing habitat is very sparse and marginal in this reach due primarily to the lack of suitable substrate There are small
patches of gravel and course sand however the reach is characterized by a silt clay substrate As sediment is transported out of this system it is
probable that habitat will continue to improve however at this time it is unlikely that sensitive mussel species would recolonize these tributaries with the
limited habitat available
McLendons Creek We reviewed this site from the crossing of Glendon Carthage Road downstream also encompassing the monitoring site referenced
in the reports
Restoration of appropriate aquatic community (McLendons Creek) Data collected during monitoring year 4 identifies multiple species associated with
lotic communities and suitable habitat was present throughout the reach
Restoration of Rare and Protected Aquatic Species (McLendons Creek) Presence of T &E species have not been documented in this reach by the
Catena Group and utilization by Cape Fear Shiner would likely be limited to the extreme lower portions of this tributary therefore the assessment as it
pertains to T &E species focused on available habitat for mussel species Mussels (Elliptio sp ) were observed during the field visit and suitable habitat
was present throughout the reach Therefore habitat is present for future colonization of sensitive mussel species
Line Creek We reviewed this site from the confluence with the Deep River to the beaver impoundment This site did not exhibit habitat suitable for the
restoration of Rare and Protected Aquatic Species Furthermore due the degraded condition of this reach it is unlikely that an appropriate lotic
community will reestablish in this system
If you have any questions please let me know Travis
Email correspondence to and from this sender is subject to the N C Public Records Law and may be disclosed to third parties

The
Catena
roup
Mr John Preyer
Restoration Systems, L L C
1101 Haynes Street
Suite 211, Pilot Mill
Raleigh, NC 27604
410 B Millstone Drive
Hillsborough NC 27278
[phone]
December 02, 2011
Subject Recent Discovery of the Cape Fear Shiner (Notropis mekastocholas) in
McLendon's Creek
References 1) Carbonton Dam -Deep River Watershed Restoration Site (2010
Annual Monitoring Report -Year 5)
2) October 14, 2011 E -mail correspondence (attached) from Travis
Wilson of the NCWRC to Todd Tugwell of the USACOE, regarding
appropriate aquatic fauna, and rare and protected aquatic species
3) November 29, 2011 E -mail correspondence (attached) from Neil
Medlin of NCDOT Biological Surveys Unit to Tim Savidge of Catena
Group, regarding discovery of Cape Fear Shiner in McLendon's
Creek in Moore County, North Carolina
Dear Mr Preyer,
On November 29, 2011, The Catena Group, Inc (Catena) was made aware of a recent
discovery of the federally endangered Cape Fear Shiner (Notropis mekastocholas) in
McLendon's Creek within the former Carbonton Dam impoundment A summary of the
details of this discovery as well as other post- removal Cape Fear Shiner occurrences
within the former impoundment is included below
If you have any questions or require additional information, please contact me at 919-
732 -1300 As always, we appreciate the opportunity to assist Restoration Systems
Sincerely,
The Catena Group, Inc
W6��a
Michael G Wood, LSS
President
Restoration Systems Carbonton Dam Removal Project December 2 2011
Cape Fear Shiner Update

CAPE FEAR SHINER OCCURRENCES IN FORMER IMPOUNDMENT
One of the goals of the Carbonton Dam Removal Project which was1completed on
February 03, 2006 was to re- establish the federally endangered Cape Fear Shiner
(Notropis mekistochals) within the formerly impounded reaches of the Deep River and its
larger tributaries (McLendon's Creek, Big Governor's Creek) As noted in the final
monitoring report (Ref 1) lotic habitats were restored shortly following dam removal in
the mainstem of the Deep River, however, the transition was much slower in the
tributaries A component of the approved five year monitoring plan involved conducting
fish surveys targeting the Cape Fear Shiner at various stations and time intervals in the
Deep River, McLendon's Creek and Big Governor's Creek The results of these surveys
are summarized in the Year -5 monitoring report (Reference 1)
Deep River The Cape Fear Shiner was not found at any of the sampling stations in
Year -1 post removal, however, it was found at eight stations in the Deep River spaced
throughout the former impoundment in Year -2 Further surveys for the Cape Fear Shiner
in the Deep River were not conducted during subsequent monitoring years, although
numerous individuals were observed throughout the impoundment while conducting
other faunal surveys
Tributaries The Cape Fear Shiner was not found in McLendon's Creek or Big
Governor's Creek during any of years sampled (Year -1 and Year -4), however, increases
in species diversity of lotic- adapted fish were evident between Year -1 and Year -4 It was
concluded that as nffle habitats and overall habitat complexity continues to develop, the
Cape Fear Shiner may utilize these creeks in the future, primarily during higher flow
events (Reference 1) It was also concluded that McLendon's Creek was more likely
than Big Governor's Creek to support the Cape Fear Shiner
As part of the project closeout process, a site evaluation was conducted on October 13,
2011 by personnel with various natural resource agencies (Ref 2) Based on this
assessment, it was concluded that the Cape Fear Shiner would likely only utilize the
extreme lower portions of McLendon's Creek, and was unlikely to utilize Big Governor's
Creek (Ref 2)
On November 22, 2011 biologists with the NCDOT were conducting fish surveys in the
vicinity of the SR 1006 (Glendon- Carthage Road) crossing of McLendon's Creek and
captured a single Cape Fear Shiner Surveys were then halted, and the specimen was
kept as a voucher and deposited in the North Carolina State Museum of Natural Sciences
(NCSM Catalog # 66730) This information was relayed to Catena on November 29,
2011 (Ref 3)
CONCLUSIONS
The utilization of tributary habitats by the Cape Fear Shiner is poorly understood It is
been hypothesized that the species utilizes the lower portions of tributaries during high
flow events However, the recent discovery of this species in McLendon's Creek, which
Restoration Systems Carbonton Dam Removal Project December 2 2011
Cape Fear Shiner Update

is located one mile from the confluence with the Deep River (Figure 1), indicates that
utilization of tributaries may not be limited to the extreme lower portions Furthermore,
the habitat where it was discovered is described as having large amounts of silt, leaf
deposits and woody debris (Ref 3), which are typically not associated habitats that
support this species This would suggest there may be greater potential for this species to
utilize Big Governor's Creek and other smaller tributaries than previously concluded
(Ref 2) In addition, the species was captured in the Deep River near the mouth of Big
Governor's Creek (Figure 1)
Restoration Systems Carbonton Dam Removal Project December 2, 2011
Cape Fear Shiner Update

N
CHATHAM 13
'= L E E
J �
MOOR \ 12
Ro
1 1.5 DEtY R/S ER \
2
f
9
JL Catena 2007 CFS Location
i
NCDOT 2011 CFS Location
1 21 (�rf Site Impoundment
— Stream
Primary Road
—_= Secondary Road
_ Chatham County
} Lee County
Moore County
The Cape Fear Shiner
Occurrences within Former
Catena Carbonton Dam Impoundment
Group Chatham. Lee, and Moore
Counties, North Carolina
Data
November 2071
Figure
Scale
0 1,125 2,250 Feat
I 1
Job No,
3280
Restoration Systems Carbonton Dam Removal Project December 2, 2011
Cape Fear Shiner Update

ATTACHMENTS
1 Correspondence From Travis Wilson (NCWRC) to Todd Tugwell (USACOE)
From Wilson, Travis W
Sent Friday, October 14, 2011 1 26 PM
To ' Tugwell, Todd SAW'
Cc John Preyer Ellison, Michael Haupt, Mac Pearce, Guy Russell,
Periann Fritz Rohde McCorcle, Justin P SAW McLendon, Scott C SAW
Baumgartner, Tim Eaton, Larry Mike Wicker @fws gov
'tsavidge @thecatenagroup com Matt Cusack
(Matthew Cusack @atkinsglobal com) (Matthew Cusack @atkinsglobal com)
ryan heese@ncwildlife org Cox, David R ( david cox @ncwildlife org)
Subject Carbonton field review October 13
The comments below are in response to WRC observations from the field
assessment on October 13
Big Governors Creek We reviewed this site from the crossing of Steel
Bridge Road to the confluence with the Deep This area should give an
adequate representation of the Big Governors and Little Governors
stream reaches It also encompasses the monitoring site referenced in
the reports
Restoration of appropriate aquatic community (Big /Little Governors
Creeks) Data collected during monitoring year 4 identifies multiple
species associated with lotic communities Although suitable lotic
habitat is sparse, nothing was identified during our field visit to
indicate further degradation of this area or conditions that may
prevent the continued colonization of an appropriate lotic community
Restoration of Rare and Protected Aquatic Species (Big /Little Governors
Creeks) Presence of T &E species have not been documented in this
reach by the Catena Group and it is not expected to be utilized by Cape
Fear Shiner therefore the assessment as it pertains to T &E species
focused on available habitat for mussel species Existing habitat is
very sparse and marginal in this reach, due primarily to the lack of
suitable substrate There are small patches of gravel and course sand,
however the reach is characterized by a silt clay substrate As
sediment is transported out of this system it is probable that habitat
will continue to improve however at this time it is unlikely that
sensitive mussel species would recolonize these tributaries with the
limited habitat available
McLendons Creek We reviewed this site from the crossing of Glendon -
Carthage Road downstream, also encompassing the monitoring site
referenced in the reports
Restoration of appropriate aquatic
collected during monitoring year 4
associated with lotic communities,
throughout the reach
community (McLendons Creek) Data
identifies multiple species
and suitable habitat was present
Restoration of Rare and Protected Aquatic Species (McLendons Creek)
Presence of T &E species have not been documented in this reach by the
Catena Group and utilization by Cape Fear Shiner would likely be
limited to the extreme lower portions of this tributary therefore the
Restoration Systems Carbonton Dam Removal Project December 2 2011
Cape Fear Shiner Update

assessment as it pertains to T &E species focused on available habitat
for mussel species Mussels (Elliptio sp ) were observed during the
field visit, and suitable habitat was present throughout the reach
Therefore, habitat is present for future colonization of sensitive
mussel species
Line Creek We reviewed this site from the confluence with the Deep
River to the beaver impoundment This site did not exhibit habitat
suitable for the restoration of Rare and Protected Aquatic Species
Furthermore due the degraded condition of this reach it is unlikely
that an appropriate lotic community will reestablish in this system
If you have any questions please let me know, Travis
2 Correspondence From Neil Medlin (NCDOT) to Tim Savidge (Catena)
From Medlin, Kenneth N [mailto knmedlm @ncdot gov]
Sent Tuesday, November 29, 20114 23 PM
To Tim Savidge
Cc Gray, Jared S
Subject Cape Fear shiner
Tim,
Here is the information concerning the new Cape Fear shiner (Notropis mekistocholos) record
we discussed earlier
McLendon's Creek at SR 1006 (Glendon Carthage Road), Moore County, NC,
• 1 individual collected — 500 feet downstream of the bridge
• Substrate a mix of silt, sand (dominant) clay (subdominant) gravel detritus and muck
• Large amount of logs and other woody debris
• Large amount of leaves in slow and non flowing areas
• Survey reach 5 50 feet
• Shocking time 5 100 seconds
• Tannin stained water
The survey was on November 22, 2011
Let me know if you need any more information
Neil Medlin
Biological Surveys Group
Natural Environment Section NCDOT
[phone]
Mailing Address
1598 Mail Service Center
Raleigh, NC 27699
Physical Address
1020 Birch Ridge Dr , CCB
Raleigh NC 27610
Restoration Systems Carbonton Dam Removal Project December 2 2011
Cape Fear Shiner Update

Restoration Systems Carbonton Dam Removal Project December 2 2011
Cape Fear Shiner Update

Via
The
Catena
roup
Mr John Preyer
Restoration Systems, L L C
1101 Haynes Street
Suite 211, Pilot Mill
Raleigh, NC 27604
410 B Millstone Drive
Hillsborough NC 27278
(919) 732 -1300
April 23, 2012
Subject Recent Discovery of the Cape Fear Shiner (Notropcs mekistocholas) in
McLendons Creek and Big Governors Creek
References 1) Carbonton Dam -Deep River Watershed Restoration Site (2010
Annual Monitoring Report -Year 5)
2) October 14, 2011 E -mail correspondence from Travis Wilson of the
NCWRC to Todd Tugwell of the USACOE, regarding appropriate
aquatic fauna, and rare and protected aquatic species
3) December 02, 2011 Memorandum from Tim Savidge of Catena
Group to Restoration Systems, regarding discovery of Cape Fear
Shiner in McLendon's Creek in Moore County, North Carolina
Dear Mr Preyer,
On April 12, 2012, The Catena Group, Inc (Catena) was conducting surveys for the
federally endangered Cape Fear Shiner (Notropcs mekistocholas) at the SR 1006
(Glendon - Carthage Road) crossing of McLendon's Creek, to gather data for the
preparation of a Biological Assessment for the NCDOT replacement of the bridge
crossing In addition, Catena conducted surveys at the Steel Bridge Road crossing of Big
Governor's Creek The Cape Fear Shiner was captured at both of these locations, which
were within the former Carbonton Dam impoundment A summary of the details of these
surveys as well as other post - removal Cape Fear Shiner occurrences within the former
impoundment is included below If you have any questions or require additional
information, please contact me at 919- 732 -1300 As always, we appreciate the
opportunity to assist Restoration Systems
Sincerely,
The Catena Group, Inc
�&W G 0a
Michael G Wood, LSS
President
Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update

CAPE FEAR SHINER OCCURRENCES IN FORMER IMPOUNDMENT
One of the goals of the Carbonton Dam Removal Project which was completed on
February 03, 2006 was to re- establish the federally endangered Cape Fear Shiner
(Notropis mekistochals) within the formerly impounded reaches of the Deep River and its
larger tributaries (McLendon's Creek, Big Governor's Creek) As noted in the final
monitoring report (Ref 1), lotic habitats were restored shortly following dam removal in
the mainstem of the Deep River, however, the transition was much slower in the
tributaries A component of the approved five year monitoring plan involved conducting
fish surveys targeting the Cape Fear Shiner at various stations and time intervals in the
Deep River, McLendon's Creek and Big Governor's Creek The results of these surveys
are summarized in the Year -5 monitoring report (Reference 1)
Deep River The Cape Fear Shiner was not found at any of the sampling stations in
Year -1 post removal, however, it was found at eight stations in the Deep River spaced
throughout the former impoundment in Year -2 Further surveys for the Cape Fear Shiner
in the Deep River were not conducted during subsequent monitoring years, although
numerous individuals were observed throughout the impoundment while conducting
other faunal surveys
Tributaries The Cape Fear Shiner was not found in McLendon's Creek or Big
Governor's Creek during any of years sampled (Year -1 and Year -4), however, increases
in species diversity of lotic- adapted fish were evident between Year -1 and Year -4 It was
concluded that as riffle habitats and overall habitat complexity continues to develop, the
Cape Fear Shiner may utilize these creeks in the future, primarily during higher flow
events (Reference 1) It was also concluded that McLendon's Creek was more likely
than Big Governor's Creek to support the Cape Fear Shiner
As part of the project closeout process, a site evaluation was conducted on October 13,
2011 by personnel with various natural resource agencies (Ref 2) Based on this
assessment, it was concluded that the Cape Fear Shiner would likely only utilize the
extreme lower portions of McLendon's Creek, and was unlikely to utilize Big Governor's
Creek (Ref 2)
On November 22, 2011 biologists with the NCDOT were conducting fish surveys in the
vicinity of the SR 1006 (Glendon - Carthage Road) crossing of McLendon's Creek and
captured a single Cape Fear Shiner Surveys were then halted, and the specimen was
kept as a voucher and deposited in the North Carolina State Museum of Natural Sciences
(NCSM Catalog # 66730) This information was relayed to Catena on November 29,
2011, and then summarized in the December 02, 2011 memo from Catena to Restoration
systems (Ref 3)
On April 12, 2012, Catena conducted additional surveys in McLendons Creek in the
vicinity of SR 1006 and captured 11 Cape Fear Shiner in a pool dust below a riffle Big
Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update

Governors Creek was also sampled that day and five Cape Fear Shiner were captured in
similar habitat below the SR 1625/1318 (Steel Bridge Road Crossing) These survey
reaches are depicted in Figure 1 and Figure 2 respectively
CONCLUSIONS
Since the removal of the Carbonton Dam in 2005, the Cape Fear Shiner has been found at
multiple locations, within the former impoundment, including mam -stem river, and
tributary locations (Figure 3) The utilization of tributary habitats by the Cape Fear
Shiner is poorly understood It has been hypothesized that the species utilizes the lower
portions of tributaries during high flow events Given the distances of these two
locations from the Deep River (1 0 mile for McLendons Creek and 0 3 mile for Big
Governors Creek) and the fact that this species has been found in tributary habitats on
separate occasions in relatively large numbers, indicates that utilization of tributaries is
not limited to the extreme lower portions Additional studies on habitat conditions and
utilization by Cape Fear Shiner in McLendons Creek will be conducted by Catena to
gather data for the preparation of a Biological Assessment for the NCDOT replacement
of the SR 1006 (Glendon - Carthage Road) crossing of McLendon's Creek This
information will further the knowledge of tributary utilization of Cape Fear Shiner
However, it can be concluded that the Cape Fear Shiner is actively utilizing the newly
restored lotic habitats in McLendons and Big Governors Creek Given these new
discoveries, it is also possible that the species is utilizing the lower reaches of smaller
tributaries in the former impoundment
Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update

Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update
2012 Survey Reach
The Cape Fear Shiner Surveys Deft: AW 2012
Fiwtr
Cotena
Catena: April 12, 2012 9-.
Mctendons Creek ° '0° 1,000 Net
Group
ClMtlnm, I.w. and Moon .ne No..
CauMMs, NoM Cuolw 3280
Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update

Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update
` *Y
2012 Survey Reach
The
Coteno
Cape Fear Shiner Surveys WI 2012
Catena: April 12, 2012
Figure
/1
Big Governors Creek o rao am ft"
L
Group
up
Lee. and Moon dm No,
Counaee, NOM CaroWd 32W
Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update

Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update
N
C A HAM _ 13 LEE
MOORE 2
�g
1. Sreer Rtvte
2
a�
v
C
Qa
e A Catena 2007 CFS Location
+A NCDOT 2011 CFS Location
(� Catena 2012 CFS Location
Site Impoundment
Stream
- Pnmary Road
Secondary Road
+ Chatham County
Lee County
Moore County
The
Catena
Cape Fear Shiner
Occurrences within Former
Carbonton Dam Impoundment
° "' � 2012
Figure
w
3=6 0 x. ♦OOOFN
Group
Chatham. Lee, and k$oor•
Courfta, NOM Caroiirw
,bp No,
32W
Restoration Systems Carbonton Dam Removal Project April 23, 2012
Cape Fear Shiner Update

® North Carolina Wildlife Resources Commission
Gordon Myers, Executive Director
November 29, 2011
Restoration Systems LLC
Attn George A Howard Member /Mgr
1101 Haynes Street Suite 211
Raleigh, North Carolina 27604
Subject Loss of public access due to removal of Carbonton Dam Motor Boating Access
Dear Mr Howard
The North Carolina Wildlife Resources Commission (Commission) is replying to your letter dated August 2, 2005
regarding the loss of public access on Deep River due to the removal of Carbonton Dam As you stated Deep River has
been known historically for its important recreational activity and public access was missed once the dam was removed
in late 2005 In this 2005 letter, Restoration Systems, Inc (RS) offered to search for restoration of motor boating access
in other, nearby reaches of Deep River within 10 miles of the Carbonton Dam including the area around Lockville "
Additionally, this letter indicated that 'RS is committed to contributing up to $20 000 pursuant to the goal of offsetting
the loss of motorboating access upstream of Carbonton
As you recall the search for an appropriate replacement public access site was not successful in 2005 Since 2005, there
has been no further resolution to the loss of public access on Deep River with regards to RS and Carbonton Dam The
Commission understands that the prerequisite monitoring period and success criteria are about to be closed out and the
public access component of the project remains incomplete
Please accept this letter as our support for a $20 000 monetary compensation for the loss of public access in lieu of
locating an appropriate nearby site Since 2005 the Commission did receive numerous local complaints and we
ultimately partnered with Deep River Park Association to construct a public access area in 2009 at Deep River Park near
Cumnock called Camelback Bridge Landing This cooperative project cost in excess of $40 000 to develop the site and
your funds would be compensation for those costs
Thank you for following up with us regarding this loss of public access for North Carolina sportsmen Please contact Erik
Christofferson, Engineering Services, regarding methods of payment at [phone] or me for additional questions at
[phone]
Sincerely,
S annon Deaton Habitat Conservation Program Manager
Cc Erik Christofferson and David Cox NC WRC
Todd Tugwell USCOE
Eric Kulz NC DWQ
Mailing Address Division of Inland Fisheries 1721 Mail Service Center Raleigh, NC 27699 1721
Telephone [phone] Fax [phone]

Deep River Project costs Federal and State Expenditures by Cost Center from 07/01/2008 to 11/30/2011 (as of
Totals $000 817 5 $22,337 12 11,561 0 $5,808 05 $15,204 39 $43,349 56
11/28/20114 02 45 PM)
Cost
Fund NonHighway
Hours Payroll Mlles Mileage
NCAS
Total
Center
216181015512 2161
817 5 $22,337 12 11,561 0 $5,808 05
$15,204 39
$43,349 56
Show
Fund Source
Employee (or NCAS acct#)
Amount
Cost
Details
2161 MILEAGE
PADGETT, LEX E
3530
$17650
2161 MILEAGE
BENTEEN, MICHAEL L
20400
$1,02000
2161 MILEAGE
TURNER, ELLIS M
3530
$17650
2161 MILEAGE
ROBERTSON, RONALD D
27170
$1,35850
2161 MILEAGE
BARBEE, DEAN A
43910
$2,19550
2161 MILEAGE
LEATH, WILLIAM E
5470
$27350
2161 MILEAGE
PAGE, LARRY K
3330
$16650
2161 MILEAGE
YARBORO, PATRICK J
8270
$44105
2161 NCAS
532185
00
$19000
2161 NCAS
532199
00
$31500
2161 NCAS
533240
00
$28389
2161 NCAS
533260
0 0 $11,011 11
2161 NCAS
533280
00
$46550
2161 NCAS
533290
00
$2,63262
2161 NCAS
533900
00
$30627
2161 Hours
PADGETT, LEX E
100
$22460
2161 Hours
BENTEEN, MICHAEL L
1015
$3,67067
2161 Hours
TURNER, ELLIS M
200
$40080
2161 Hours
ROBERTSON, RONALD D
2890
$7,51416
2161 Hours
BARBEE, DEAN A
2020
$6,45222
2161 Hours
LEATH, WILLIAM E
590
$1,27546
2161 Hours
PAGE, LARRY K
250
$54746
2161 Hours
YARBORO, PATRICK J
540
$1,11406
2161 Hours
PARSONS, TRACY L
570
$1,13769
Totals $000 817 5 $22,337 12 11,561 0 $5,808 05 $15,204 39 $43,349 56

I
4
�
� I r, 79 i -'1 09 1 1 * YIv
on
Goode
e ev 2891t Jid 200", Eve all 16415 it
44 a!
If
I
4
�
� I r, 79 i -'1 09 1 1 * YIv
on
Goode
e ev 2891t Jid 200", Eve all 16415 it

i
W�
0

01 9''1
Natural Rmw E,
RL- slo';1bon &C onseivatkmt
August 2, 2005
David Cox, Technical Guidance Supv.
N. C. Wildlife Resources Commission
1142 I -85 Service Road
Crccdmorc, NC 27522
SUBJECT: Proposed Removal of Carbonton Dam: Motor Boating Access
Dear David:
Restoration Systems, LLC (RS) recognizes that a historically important recreational
activity will no longer be feasible in the portion of the Deep River impounded by the
Carbonton Dam when the dam is removed during the fall of 2005. The formerly
impounded river reach will return to a natural (lotie) pattern of flow, characterized by
riffle and pool areas, but without sufficient water levels to routinely support the operation
of motorboats by fisherman and others.
To mitigate for this loss, RS will investigate the opportunity for restoration of motor
boating access in other, nearby reaches of the Deep River. Reconnaissance of areas with
I mpounded waters within 10 miles of the Carbonton Dam will be made, including the
area around Lockville. Any area identified by RS as a candidate site must fulfill the
overall North Carolina Wildlife Resources Commission (WRC) mission germane to
public boating access, however, such an identification will not foreclose on the WRC's
right to decide if the potential access would satisfy its requirements for an alternative
recreational access facility. RS is committed to contributing up to $20,000 pursuant to
the goal of offsetting the loss of motorboating access upstream of Carbonton.
Finally, RS wants to reiterate its interest in partnering with the WRC and other relevant
agencies throughout the planning, design and implementation phases of work on the
Carbonton project. RS makes the commitment to pro - actively engage the WRC in the
review of dam demolition and restoration activities, including design and constructability
discussions. RS appreciates the WRC's interest and looks forward to productive
discussions about the project.
Sincerely,
George A. I coward, Member /Mgr.
Pilnt R /lilt • IInt Nwnrc Ct Giitr 1117 . RA PIA NC �714U . ww vrevtnratinncvctPms rnm • Phone- 919 - 755 -94% *Fay - 919 - 751 -9492

00-
you are standing in the exact goographic center of
North Carolina, at the site of the former Carbonton Dam.
This $8.2 million dam removal project was successfully
completed by Restoration Systems of Raleigh for the
NC Ecosystem Enhancement Program in the Fell of
2005. The first and largest of its kind in North Carolina,
this project was undertaken as mitigation for unavoidable
development of watercourses elsewhere in the
deep River portion of the Cape Fear watershed.
iHere, 20 miles of the Deep River has been restored to
its natural state and me former site of the dam is now
a beautiful green space that still retains Its history.
Remnants of the dam remain, lining the bank, and the
steep powerhouse still towers over the bank opposite.
r,
I,
Dam removal and careful monitoring
The first demolition phase saw the step -by -step lowering
of the impoundment behind the spillway during October.
By ensuring the gradual release of colder water, this
method minimized the impact on the river, wildlife and
dissolved oxygen levels. Continuous monitoring showed
this was a success, creating far less impact than a
typical rainstorm
For the next 2 weeks, water was released through
the powerhouse's turbine shaft and mud gates under
constant evaluation. Huge concrete gates in the
spillway were gradually opened to carefully control
sediment release. The final step saw 70 -feet of the
dam wall removed.
For five years, 54 monitoring stations throughout the
1000- square -mile watershed measured water quality
and species diversity. Freshwater mussels and fish
were regutwly counted. All measures showed improved
e koninental health of the river since the dam's
removal. Improvement that will continue in the future.
River regeneratitm
With the removal of the Carbonton Dam the Deep River
has returned to a more natural stale, recreating lost
habitats and uncovering the river's past
First came the discovery of a f mbar, crib dam dating
from the 1800s that had been totally submerged. Further
upstream. 7 new riffles and a v- shaped Native American
fishing weir were revealed for the first time in over 150
years. New shallow cobble bars in the riverbed now
provide critical habitat for reestablishing endangered
mussels and shallow -water organisms.
Biologists have been astonished by the successful
repopulation by the federally - listed Threatened and
Endangered Fish, the Cape Fear Shiner. This small
silver minnow Is found in no other watershed anywhere
else in the world. The many dams on the Deep River
caused numbers to dwindle for decades and the return
of the Shiner represents one of the largest species
restoration of its kind in US history
e
r

k
i

4
DEEP RIVER
LEE & CHATHAM COUNTY
CARBONTON NORTH CAROLINA
MAY, 2006
OP
O
DEEP RIVER
PARK
O
O QO
PROJECT SITE VICINITY MAP
n
acFLq , 1
LIST OF DRAWING&
LI SITE PLAN LAYOUT
G1 SITE PLAN GRADING
SE1 SITE PLAN SEDIMENT & EROSION CONTROL
SD1 SD2 SITE DETAILS
i
MILONE�� CBROOM®
Leadwape AroWteotore
And Environmenw Sdenoe
707 B P W Suoe
ar—M SC 29601
o�rx7i ss9e
.ay.m buomsm
N
�0
'r
_ t
�r
Y
Y H P
-r I
wM'
F �
r �
I
i
F
7
l
r�
s
r r
I

N m °n iar�
SIDE Vim
NOTES
1 UTILIZE 7DOG COIR FIBER MATTING
2 SECURE AT 1 It INTERVALS (OR AS
.a '� RECOMMENDED BY MANUFACTURER BASED ON
S.OPE) 8ACKFlLL AND COMPACT SOIL
qan N.n
X11) BE USED AT TOP OF ALL MATTING
INITIAL MATTING ANCHOR TRENCH
STANDARD TEMPORARY SILT FENCE I 1
AM
Sam
ERASION AND SEDIMENT CONTROL SHALL PROCEED IN THE FOLLOWING GUNNER
THESE GUIDELINES SHALL APPLY To ALL WORK CONSISTING OF ANY AND ALL )
1 PERMANENT VEGETATIVE COVER SHALL BE ESTABLISHED AS VARIOUS I
ey
>TJPO dID /OR PERIiWENf YFASLf�.S 1D CONfR0. WATER POLLUTION Me
�
`-
1 PRIM TO MMMLNCEMEMT QF WON A PRECOHSTRLICTIW YEETD4 SHALL
SOIL A3 WT BE IEOUA® OlI16NG 7HE COSTRlIC110N Of THE
PROJECT
L i GOKdIICION 01 TK WIE AIE SI01 1Ml
BE HELD WITH THE EIgNEER AIW REPPESEIITATNES OP TXE CONTRACTOR, UTLRES
�_ �.`
vti `r
ALL CONSTRUCTION ARM SUBJECT TO EROSION W OM FINAL GRADING HAS
AND OWNER AT THIS MEETING THE SEDIMENT AND EROSION CONTROL
NNE W Nm DIDITD M 1K 1010L q 1a
IR U114 tNC 41oL.
A01Ue1
d s�Ri TD i ®ti
»S�[ 01 WAND
vcu �• •' r'�.L
'+��
MANNER 50 AS MDT W AN( WETLANDS WATERCOL* E WATERNOY
m UN X111 6
Wnt1v e T
aFAlot
® �tNU v TNF
ryS blDlwo lrt
rmNP A6r6
'mom
CONDUIT CAWVIWr WATER ETC THE c6NTRACTM SHALL LIMIT INSOFAR AS
POSSIBLE THE SURFACE AREA OF FARM MATERIALS EXPOSED BY CONSTRUCTION
T \
� M�
BE RETAINED NO DISTUIRBANCE IS TO TAKE RACE BEYOND THE LIONS STAKED
METHODS AND IMMEDIATELY PROVIDE PERMANENT AND TEMPORARY POLLUTION
] CDNIDIILI l0 cOMA01BRiA d t1E YCOM
AM �WOrt COkIILL M1JNC'HN MO 0001
CMORNCTM SHALL TAKE SPEOAL PRECAUTD16 TO PROTECT TREES AND
�3 ��
1
2 REMOVE (DOSE ILOtx STONE AND CQNSTRIA0710N DEERS FROM
AND EHOSiN6 IMPROVEMENTS ro REMAIN
3 CONTRACTOR TO COORDINATE WON SCHEDULE WITH IMPACTED PROPERTY OWNERS
NRUM uTm mm4
AREA
3 POTION ALL PLANTING OPERATIONS PARALLEL TO THE CONTOURS OF THE
• rI01Q RAWm "rt 1aANn1uR cum
mmnNwaN DITUI¢ ANP I6fIND AMOK
ELM
SLOPE.
LRCAW ANA
FIRM
rl
C
� 14ID1 f10E
CROSS SECTION
I CONSTRUCTION ENTRANCE 1 7 1
121;1 M 1214 4 WAKII1111 1 ':• IM\
NOTES
1 UTILIZE 70DG COIR FIBER MATTING
2 SECURE AT 1 H INTERVALS (OR AS
RECOMMEINbED BY MANUFACTURER BASED ON
SLOPE) BACKF7LL AND COMPACT SOL
3 ANCHOR STAPLE AND OVERLAP IN
ACCORDANCE WITH MANUFACTURERS
INSTRUCTIONS
4 TO BE USED ON ALL MATTING AREAS AT
INTERVALS AS RECOMMENDED BY THE
MANUFACTURER
INTERMITTANT MATTING
CHECK SLOT
CONSTRUCTION SEQUENCE /
SEDIMENT & EROSION CONTROL NOTES
I
SEDIMENT & EROSION CONTROL SPECIFICATIONS VEGETATIVE COVER
GENERAL. GENERAL
ERASION AND SEDIMENT CONTROL SHALL PROCEED IN THE FOLLOWING GUNNER
THESE GUIDELINES SHALL APPLY To ALL WORK CONSISTING OF ANY AND ALL )
1 PERMANENT VEGETATIVE COVER SHALL BE ESTABLISHED AS VARIOUS I
>TJPO dID /OR PERIiWENf YFASLf�.S 1D CONfR0. WATER POLLUTION Me
SECTIONS OF PROJECT ME COMPLETED 0) ORDER TL STABILIZE THE
1 PRIM TO MMMLNCEMEMT QF WON A PRECOHSTRLICTIW YEETD4 SHALL
SOIL A3 WT BE IEOUA® OlI16NG 7HE COSTRlIC110N Of THE
PROJECT
SOU. REDUCE DO�/SiRFAW O"CE FROM SEDIMENT AND RUNOFF AND TO
ENHANCE THE AE T ETIC NATURE OF THE SITE R WILL BE APPLJED ITo
BE HELD WITH THE EIgNEER AIW REPPESEIITATNES OP TXE CONTRACTOR, UTLRES
ALL CONSTRUCTION ARM SUBJECT TO EROSION W OM FINAL GRADING HAS
AND OWNER AT THIS MEETING THE SEDIMENT AND EROSION CONTROL
N GENERAL, ALL CONSTRUCTIO I ACTMITES SHALL, PROCEED N SUCH A
BEEN CUYPILiED AND A PERMANENT COVEIR IS NEEDED
PTA" Wu BE DISCUSSED
MANNER 50 AS MDT W AN( WETLANDS WATERCOL* E WATERNOY
SITE PREPARATION
2 coNTRACTOR To sE KE our uw of DNSI1xmucE AND VEGETATION To
CONDUIT CAWVIWr WATER ETC THE c6NTRACTM SHALL LIMIT INSOFAR AS
POSSIBLE THE SURFACE AREA OF FARM MATERIALS EXPOSED BY CONSTRUCTION
BE RETAINED NO DISTUIRBANCE IS TO TAKE RACE BEYOND THE LIONS STAKED
METHODS AND IMMEDIATELY PROVIDE PERMANENT AND TEMPORARY POLLUTION
1 INSTALL REOUTED SURFACE RATER CONTROL ME/SRRES
CMORNCTM SHALL TAKE SPEOAL PRECAUTD16 TO PROTECT TREES AND
C0N(RpL MEASLES TO PREVENT gTI1SMOHATGH OF ADJACENT WETLANDS
1
2 REMOVE (DOSE ILOtx STONE AND CQNSTRIA0710N DEERS FROM
AND EHOSiN6 IMPROVEMENTS ro REMAIN
3 CONTRACTOR TO COORDINATE WON SCHEDULE WITH IMPACTED PROPERTY OWNERS
IIAYERCOURSE5 WATERDOOES AMC TO PREVENT INSOFAR As POSSIBLE
mosoK ON YHE SITE.
AREA
3 POTION ALL PLANTING OPERATIONS PARALLEL TO THE CONTOURS OF THE
TO MAINTAIN SATE VEHICLE AID PEDESTRIAN ACCESS AND PARKING.
j.,
LAND GRADING
SLOPE.
CONTRACTOR To LA wAn DISRUPTION TO THE GRFAYESY EXTENT PRACTcvu
GENERAL.
4 APPL TOPSOIL AS INDICATED ELSEWHERE HERDII
4 CONTRACTOR TO (STALL SEDIMENT AND EROSDN COMROLS PRIOR TO
5 APPLY FFFTILIZER ACCORDING 70 SOIL TEST OR
CLEARING AND GRUBBING INSTALL CONSTRUCTION ENTRANCE PADS AS DEPICTED
ON THE PLANS
1 THE FN34APYNG OF TIC GRAND SIIRF/ICE BY EXCAVATION AND FLUNG OR
A COMBOA71W OF BOTH TO OBTAIN PLMNEYI GRADES SAIL PROCEED N
L��_� �
OF 10 10-10 kWWER PER ACRE (23 LBS PER 1 BOO SO FT ) THEN SO(
ACCORDANCE WITH THE FOLLOWING CRITEIM
TO MIT WEEKS LATER APPLY ON THE SURFACE AN ADDITIONAL 300
UI MAINTAINED
3 711ROIXi1 ROI'I O RATERCOIXZSE SHALL BE YNIDIDEO WRING CONSTRUCT" 50 AS
HALL
THE CUT FACE OF EARTH EXCAVATDN $HAUL NOT BE STEEPER THAN TRIO
LB& O' 10- 0 10 FERTILIZER PER ACRE. AFTER SEPTEMBER 1 TDANA RY
VEGETATIVE COVER SHALL BE APPLIED
ME
NOT SUSPEND SEDIMENT FROM
HORZONTAL TO ONE VERTICAL (2 1)
\
6 IRRITATE E RTwm 6PERAYGR6 AFTER ALL SLTAMENT AND EROSION
a THE PERMANENT E7EAgD FACES O FEU SMALL NOT BE STEEPER THAN
In
CONTROLS ARE N P LACE'
RCI
TWO HORIZONTAL TO ONE VFRL (2 1)
VEGETATIVE COVER SELECTION &MULCHING
7 AREAS OF ACTIVITY AND MIOSED AREAS ARE TO BE WMWQM STABILIZE
NOT BE STEEPER THAN ONE
y1
TEMPORARY
TEMPORARY VEGETA H IVE OVER
ALL SLOPES IMMEDIATELY AFTER THEIR ESI BI-SMA NT
KOOWCZOONTALL TO F" �� 4)
5 ESTABLISH ALL SLOPES TO GRAVE N ARE15 OF DISTURBANCE AS SOON AS
4 PROJSNON 511WLD BE MADE TO CONDUCT suwA`E WATER SAFOLY TO
'� 1 APRIL 15
TEMPORARY SEED AND MULCH N ACGDIRDANCE WITH THE LANDSCAPE
PREVENT SURFACE RUNOFF FROM DAMAGING CUT FACCS
=
120 � /AC �NENAMENTS FOR AMDPOARYI
SPECIFICATIONS T6
SLOPES
B THE SEDIMENT AND EROSION CONTROL PLAN MAY BE MODIFIED BY
TOPSOILING
APRIL /5 AUGUST 5
GERMAN MILLET
THE SITE ENGINEER AS NECESSITATED BY CHANGING SITE CONDITIONS ADDTIONL
LfERWEST 2000LB5/AC
WNTROL DEVICES BESIDES WHAT Is SHOWN N PLANS WILL BE ADDED BY THE
GENERAL
1 1TIRAL
ENGINEER IF NEEDED 10 SkILOEM CONTROL YFASIR6 SHALL BE INSPELTED AT LEAST ONLE
1 70PSOL SHALL BE SPREAD OVER A E7(POSED AREAS N ORDER TO
pp�V� A SOIL MEDIUM HAMNG FAVORABLE CHARACTERISTICS FOR THE
AUGUST ,800 LBSrAD
RYE GRIN 30
120 LBS/AC
THAN SEVER CALENDAR OF P DAPS Ino AFTER sEANY GREATER
THAN DS INCHES OF PRE OUIIRE ANY REPLACE PERIOD DAMAGED
E,SMBL9/MQNT txRaWrH AND MANTET4AILE OF VEGETATION
COVER
INEFFECTIVE DEVICES HALL BE REPASSED OR REDUCED ARf
2 UPON FOAL SLIBCRADES SCARIFY SURFACE ro PROVIDE A 6000
PERMANENT VEGETATIVE
AL FEATURES S
ALL SEDIMENT COMTRO FEBTAI ES SHALL BE MMNANED IAITL FINAL
IL FINAL
STABILIZATION 1415 BEEN OBTAINED
BONG WITH TOPSOIL_
WITH TO
PERMANENT GR/55, "RIPMIAN NO BUFFER SEED Nil( AT DBS /TIRE FROM LELLOW
WISH PMML SyER OTT lDtiM GVOLQM M APPROVED EOIYrL
11 INSPECTION OF THE 511E FOR EROSION SHAW. CONTINUE FOR PEA00 OF
3 ROXW ALL LARGE STONES, TUTEE LINES ROOTS AND C01STRUCMN
TEMPORARY NiiDuc
THIEE MONTHS AFTER COMPLETION WHEN RAINFALLS OF ONE INCH OR
STRAY OR HAY 70 RIO IBS /1000 SOFT
MORE OCCIIL
4 APPLY LIME ACCORDING M 50L TEST OR AT THE RATE OF TWO (2) TONS
PER ACRE.
(TEMPORARY VEGETATIVE AREAS)
12 ALL EROSION CONTROL DEVICES SHALL HE PROPERLY MAINTAINED DURING ALL PHASES
MATERIAL
WOOD FIBER IN HYDR0M1ALH MMM T.�50 MIT WO SO FT
OF CONSTRUCTION UTIL THE COMPLETION OF ALL COIISTRIICTDH ACTMIIES AND Al
DSRNHD AREAS HAVE BEEN STABILIZED All 7ENPOARY CONTROL DEVICES
1 TOPSOIL SHWLD HAVE PIASICN- CHEMICAL AM &O.OGIGL
ESTABLISHMENT
SHALL BE REMDAD WIEN COISIRAXIVN IS COMPLETE AND THE SITE IS STAAA7ED
CHARACTERISTICS FAVORABLE TO THE GROWTH OF PLANT
13 THE CONTRACTOR MUST TAKE NECESSARY ACTIN TO MINIMIZE THE TRACKING OF MUD ONTO
2 TOPSOIL SHOULD HAVE A SkRDY OR LOW TEXTURE
I SMOOTH AND FIRM SEEDBED WITH CULTPACIIER OR OTHER SIMILAR
EQUIPMENT PRigR TO SEEDING (EXCEPT WHEN HTOROSEEOING)
THE PAVED ROW"Y FROM CONSTRIXTION AREAS THE CONTRACTOR SHALL REMOVE
MUD /SOU. FROM PAYEIENT DAILY AS MAY BE REOUNED
3 TOPSOIL SH BE RpA]IVELY FREE O' SUBSOIL MATERIAL RICA MST EE
L SELECT ADAP'TED SEED MIXTURE FOR THE SPECIFIC MOTE
TOLD
FIFE OF STIES (OVAL 1 fN GIA DE) LUMPS SH SOIL, ROUTS. TREE
THE G O T OVEIR SELECTION
RATES AND ONE SEEDING DATES (SEE VEGETATIVE HAVER S73ECITON Q
14 THE SITE SHOULD BE KEPT CLEAN OF LOOSE DEEMS AND SURD NO WTERIALS
ERE
IJ/BS AND TRAASII OR C015TRI1CTDN DEBRIS R SHOULD BE FREE O ROOTS
E
OR RHIZOMES SUCH AS THISTLE NUTCR SS AND O ACKYRASS
MULCHING SPEMICATON BELOW)
SLOE THAT NONE O" THE ABOVE ENTER STORMWATER FACILITIES RONDWAYS
WATERCOURSES OR WETLANDS
4 AN ORGANIC MATTER CONTENT OF SDL PERCENT (SIT) S REQUIRED AVOID
3 APPLY SEED UNIFORMLY AACCORWIC TO RATE INDICATED BY
BROADCASTING &BLIND OR HYMAJLIC APPLICATION
15 A COPY OF ALL PANS AND REVISIONS AND HE SEDIMENT AND EROSION
LIGHT COLORED SUBSOIL. MMUML
CONTROL PAN SHALL BE MAINTAINED ON SITE AT ALL TDE3 DURING
5 SOLUBLE SALT CONTENT W OVER SOD PARR PER MOLJON (POM) S LESS
4 COVER Glt SS AND LEGUME SEED WITH MDT MOTE THAN 1/4 NCH OF SOIL
CONSTRUCTION
SUITABLE VOID TAIL MARSH SOBS BECAUSE OF HIGH SALT CONTEXT
WITH SUITABLE EOIAPYENT (EXCEPT WHEN HYDROSEEDND)
16 A COPY OF ALL INSPECTION LOGS SHALL BE RETAINED FOR THE DURATION
AND SULFUR ACIDITY
5 MACH IMMEDIATELY AFTER SEEDING F REQUIRED ACCORDING TO
OF THE PROJECT
6 THE PH SHOULD BE MORE THAN 60 F LESS ADD LAVE 70 INCREASE PH
TEMPORARY k MCHINGEG SPECIFICATIONS (SEE
ELOW) (ETATNE COVER
17 ALL SEDWNT ANT, EROSION CONTROL MEASURES SHALL BE RONWED ONLY
ro AN ACCEPTABLE LEVEL.
6 ALL F'O"`S USE FOUR TIES
IRON sABRQATION O ALL LJPGRADIEI+r AREAS
APPLICATION
MOOl41 RATES W1O`URL H,oORRa NN`EDG11ME
AVOID SPREADING WHEN TOPSOIL S WET OR FROZEN
7 URSE SOD WHERE THERE S A HFAYY CITNCENTRAIGM O WATER AND N
CHBTFAL AREAS WHERE R S BmRRTANT To DIET A OU1O( VEGETATIVE
2 SPREAD TDP90L UNIFORMLY TO A DEPTH OF AT LEAST SA INC HR (67
OR TO THE DEPTH SHOWN ON THE LANOSGAPDIG PLANS
COVER TO PREVENT EROSION
TEMPORARY VEGETATIVE COVER
MAINTENANCE
I TEMPORARY VEGETATIVE C06M SMLL BE ESTABLAED ON AL
1 TESJ FOR 501E ACUITY EVERY THREE YENS AND LIE AS REORRED
UNPROTECTED BCEM THAT
CR \ONC COMPLETmPR00PC AREAS wow AREAS � TED PERIOD
L ON SITES WHERE OW556 PRIDOIOHATEA BROADCAST ANNUALLY 50) POUNDS
OF WIRE SOL DI OSUURE 5 LESS THAN 12 MONTHS TEMPORARY
O 10 10 10 fEITITL¢FA PER AME (12 LBS. PER 1 000 SO R) OR AS
SEED AND MULCH ALL DISTURBED AREAS ACCORDING TO NCOOT ST WARD
NEEDED ACCORDING TO AW41AL SOIL TEST
SPECIFICATIONS
SITE PREPARATION
3 � MI AS wo��T�m By SOL TEST 3 P OF O-m 20 OR
EQUIVALEKT PER ACRE (S LBS PER 1GDO SO FT)
1 INSTALL FW"tED SURFACE WATER CONTROL MEASURES
2 RDOVE LOOSE ROCK STONE AND CONSIRIDIGN DEBRIS FROM AREA
3 APPLY LIME ACCORDING TO SOIL TEST OR T A RATE OF ONE 45 LBS OF
GROUND DOOOIC LIMESTONE PER SF
4 APPLY FERTILIZER ACCORDING TO SOIL TEST OR AT THE RATE OR 1000
LBS OF 10 10 10 PER ACRE (23 LOS PER 1 BOO SO FT j ANA SECOND
KTN
APPLICXT OF 200 LOS OF 10 1G-10 PER ICRE(5 LOS PEER 1000 SO FT)
WHEN GRASS B FOUR INCHES (4-) TO S X INCHES (6-) HIGH APPLY
ONLY WHEN GRASS IS DRY
5 UNLESS HYDROSEEDED WORK N LIME AM FERTILIZER TO A DEPTH OF
FOUR (4) INCHES USING A DISK OR ANY SUITABLE EOIPMEKT
6 TILDE SHOULD ACHIEVE A REASONABLY UNIFORM LOOSE S® NON ON
CONTOUR F SITE S SLOPING
ESTABLISHMENT
1 SELECT APPROPRIATE SPECIES FOR THE SITUATION NOTE RATES AA
EM 13AY) (SEE VEGETATIVE COVER SELECTION t "CITING
SEED
2. APPLY SEED UEOPORWLY ACCORDING TO THE RATE INDICATED BY
BROADCASTING DRILLING OR HYDRAULIC APPLICATION
3 UNLESS HFDROSEEDED GER RYE GRAIN IN WITH NOT MO THAN /
INCH OF SOIL USING SIVUTABLE EQUIPMENT
4 MULCH IWEDATELY AFTER SEEDING IF ROARED (SRI VEGETATIVE
COVER SELECTW t M1LOHING SPECIFICATION BELOW) APPLY STRAW OR
RAY MULCH AND ANCHOR TO SLOPES GREATER THAN JX OR WHERE
CONCENTRATED FLAW WILL OCCUR
_
d
to
O
8
u
G
D
a�
N
�w
D J
Z
Z
a0
Z
TIN[
D� MAY 25 2006
I¢r n
2681 01 05
s .......In-
SD1
J
. TT
7 ..
I
1`
F
1
}
I
1
\
In
iV
d
to
O
8
u
G
D
a�
N
�w
D J
Z
Z
a0
Z
TIN[
D� MAY 25 2006
I¢r n
2681 01 05
s .......In-
SD1
J
. TT
7 ..

t
l
3X8 TREATED
HARDWOOD LENGTH
PER PLAN
(r r, -(� I�f Irirl4I
I
NUT
W /WASHER
6
LAG BOLT
W /WASHER
WOOD STEP
3
EXISTING
4 X8 BRICK
24 RCP
MANHOLE
PROPOSED
�48 RCP
1
2 X3 FRAME AND GRATE
SET FLUSH WITH TOP
EXISIINC
24 RCP
_TF 2345
24 RCP IN INVERT IN
24 RCP IN 232 55
48 RCP OUT
23055 58LF - 48 RCP S- 9%
I NVERT IN
237 98
SLOPE
CONCRETE
BRICK MANHOLE BASE W /FLAT SLAB TOP
n
24 DIA
3/4 HOLE
9 FROM END
DRIVE i4 REBAR INTO
SUBGRADE TO ANCHOR
..8 PRESSURE
TREATED WOOD
6 WOOD WHEELSTOP
MFER ALL AROUND
DIAMETER HOLE
PRESSURE TREATED
POST
WOOD BOLLARD W/WIRE STRAND
Ks
>1
8
lilt "Irl � pp
w
q�oDo
aU a
a �
�= N
w
W UU
� J Z
PBS
wi
MAY 25 2006
��" 2891-01.06
s aas.�spmu.imr
SD2
{ 1
i
iy
a

r...:

t
4 ' f
L

\ ` 1
I� ,
ail � r � t r — •ti \
a
,
+ + rl 1 1
,
r \ ,
1
, 1 O • � 7 / � I I r i � � , \
I
I r Ir
, I
I
O
I O
J
I
�
1
r
r a —
a
r .
l
r
1 r
I
1
I
1
t ,
f r
1
C
I
Z
I
ti
t
rn
!
■
PARK LAYOUT
REVISIONS
DEEP RIVER
e
�
MIIANS & MACBROObt
I
CARBON DAM RESTORATION SITE
$
4
3078 r u sU
Z I s me mim s,sol
CARBONTON NORTH CAROLINA
tt6�1 71.�F 1 -^ 7�1 X135

7
' r
fill
I
III
Ill�i l
�� VIII I II
IIII, I �I
I� I!I ilil
I�
f
I
I6
,
I
k \\,
= 1
—�\ %\,
�\ \ -7111 � 1
IIjj1,llllrl 11 l i f I \` \1Ill!lllll / / /j1 / /W)I J.l II
M
i�1i1►1 ,,.ts�:
I' /� Ifll
III! F I
`t
II I
IIII I�f I /
fit iIl/
fill /IIII ;
Illii I/ i li ..
II I II
I I
I I IJ/
I
\\ \\ \ \ \1 \111 \\
� \1
al
�I`\
1111\:1411 1 11, 1�Ipl
lift
0
" IN
1
1 ! r'
V, ,\
/•ter —i — '- - -_.., � � \ � �����s \ \ �
.,���_
5 � � 1 _
/1- I r li /iii
I till
(1111
77ti�
/— \l t%
i
1
E
p
�
RT III\�\��\� \\
/N1111 \ \ \ \ \ \ \ \ \ \ \\ \ \\
IIjj1,llllrl 11 l i f I \` \1Ill!lllll / / /j1 / /W)I J.l II
M
i�1i1►1 ,,.ts�:
I' /� Ifll
III! F I
`t
II I
IIII I�f I /
fit iIl/
fill /IIII ;
Illii I/ i li ..
II I II
I I
I I IJ/
I
\\ \\ \ \ \1 \111 \\
� \1
al
�I`\
1111\:1411 1 11, 1�Ipl
lift
0
" IN
1
1 ! r'
V, ,\
/•ter —i — '- - -_.., � � \ � �����s \ \ �
.,���_
5 � � 1 _
/1- I r li /iii
I till
(1111
77ti�
/— \l t%
i
., ✓� .riz � v -.
l
1
E
p
q
GRADING PLAN
REVISIONS
pmt
_
w b.,�mad ..rte.
DEEP RIVER PARK
Q, MILONE &IMACBROOM.
F�
LEE & CHATHAM COUNTY
307eFdbsrm
NC HWY 42
OrtmnOq SOUM Caoimi 29501
WA 271-M Fu W4) 2711133
..
CARBONTON NORTH CAROLINA
j
+ Y
F
\ ;
., ✓� .riz � v -.
l

6
t
r �
it f

Dear Mr Ellison
Deep River Park
Association
May 23, 2012
I heard your organization's direct role in the Carbonton Dam project is nearing an end
As the long -term steward of the six acres that now comprise of Carbonton Park, I am
writing to express appreciation for the protect and ajob well done
The Deep River is a very special place to us down here and removing the dam has
unproved the river immensely New rapids, ancient Indian fishing weirs and a clear
passage down the river through our region for the first time in 150 years has made the
nyer more popular than ever In fact, it is cause to rejoice — baptisms are a regular use of
the new park?
The land donation and endowment provided to the Deep River Park Association to create
Carbonton Park has made a great river project even better Since the removal in 2006 the
park property is used frequently, safely and to the great enjoyment of the community
We stage regular community picnics and paddling events at the Park Local craftsman
have made the powerhouse safe for the first time in years, with new safety railings and
secure closure of the structure from trespassing We are currently in talks with a
company that is interested in developing a climbing, zip -lme and water sports facility
with the powerhouse at its centerpiece As your contractor Restoration Systems knows, I
personally appealed to have the powerhouse remain and I am grateful it was spared It is
an important part of our history and our future
We appreciate the installation of the interpretive signs in the near future Six years ago I
was skeptical that the signs would survive the vandalism But now that the park has
become such a local focal point, I think it is tune your organization took a "bow" with the
signs and tell the story of the good work done by the Ecosystem Enhancement Program
In closing, if there is a way the Deep River Park Association can assist the NCEEP in
future projects along our river corridor please do contact us We have a number of
projects that could benefit from future collaboration with you
We All Live Down River,
Dick Hamson
Executive Director
deepnvemc @gmail corn
Deep River Park Association

JOURNAL OF GEOPHYSICAL RESEARCH VOL 113 G03019 dot 10 1029/2007JG000654 2008
Hel�
Full
Article
Suspended sediments in river ecosystems Photochemical sources of
dissolved organic carbon, dissolved organic nitrogen, and adsorptive
removal of dissolved iron
J Adam Riggsbee 12 Cailin H Orr 36 Dina M Leech 4 Martin W Doyle 5
and Robert G Wetzel'
Received 19 November 2007 revised 14 March 2008 accepted 31 March 2008 published 8 August 2008
[t] We generated suspended sediment solutions using river sediments and river water at
concentrations similar to those observed during 1 5 year floods (QI 5) and a dam removal
( -325 mg L -1) on the Deep River, North Carolina Suspended sediment solutions were
exposed to simulated solar radiation equivalent to one clear, summer day at the study site
(35 °N) Concentrations of dissolved organic carbon (DOC) dissolved inorganic carbon
(DIC) total dissolved nitrogen (TDN), dissolved inorganic nitrogen (DIN) dissolved
organic nitrogen (DON) soluble reactive phosphorus (SRP) and total dissolved iron (Fed)
were measured before and after exposure Total dissolved carbon (TDC) budgets for each
experiment were produced using DOC and DIC data Sediment suspensions in the
presence of simulated solar radiation were significant sources of dissolved C (119 f
11 fcmol C L -1 d -1, f values indicate 1 standard error) and DON (1 7± 0 5 µmol N L -1
d- ) but not DIN or SRP Extrapolations through the Deep River water column suggest
that suspended sediments in the presence of light represent dissolved organic matter
fluxes of 3 92 mmol C m -2 d -1 and 40 µmol N m -2 d -1 Additionally sediment
suspensions lowered river water Fed concentrations Immediately ( -24 %) and
progressively (-40 -90 %) in both light and dark treatments Our research suggests
suspended sediments in river ecosystems are potential sources of dissolved organic C and
dissolved organic N while effectively removing Fed from the water column
Citation Riggsbee J A C H Orr D M Leech M W Doyle and R G Wetzel (2008) Suspended sediments in river ecosystems
Photochemical sources of dissolved organic carbon dissolved organic nitrogen and adsorptive removal of dissolved iron J Geophys
Res 113 G03019 dot 10 1029/2007JG000654
1 Introduction
[ ,7] On broad spatial and temporal scales rivers owe their
origins to erosional processes which create channels that
transport htllslope materials to the world s oceans Howev
er on finer scales nvers are actually mosaics of aggrading
and degrading reaches routing materials through a series of
erosional and depositional zones Biogeochemical processes
within river ecosystems are governed by these smaller scale
hydrogeomorphic conditions making rivers important sites
for global biogeochemical transport and transformation
Much scientific effort has been invested exploring links
'Department of Environmental Sciences and Engineering University of
North Carolina Chapel Hill North Carolina USA
2Restoration Systems Raleigh North Carolina USA
3Carolina Environmental Program University of North Carolina
Chaf,el Hill North Carolina USA
Institute of Marine Science University of North Carolina Morehead
City North Carolina USA
SDepartment of Geography University of North Carolina Chapel Hill
North Carolina USA
6Now at National Center for Earth Surface Dynamics Saint Anthony
Falls Laboratory University of Minnesota Minneapolis Minnesota USA
Copyright 2008 by the American Geophysical Union
0148 0227/08/2007JG000654$09 00
among hydrology geomorphology and biogeochemistry in
river ecosystems including the River Continuum Concept
[Vannote et al 19801 as well as organic matter and nutnent
spiraling [Webster and Patten 1979 Minshall et al 1983
1992 Newbold et al 1982 19921 However the biogeo
chemical role of suspended sediments which are also
controlled by hydrogeomorphic factors in watersheds is
less well understood
[3] Materials conveyed by rivers are largely attributed to
external or allochthonous sources i e watershed hill
slopes For example dissolved organic carbon (DOC)
flushing from htllslope soils was elicited to explain DOC
hysteresis during snowmelt dnven floods [Hornberger et
al 1994 Boyer et al 1997 2000] Watershed subsurface
flow paths and near stream sources in particular are repeat
edly cited as variable source areas providing DOC enrich
ment to channels during floods [Meyer and Tate 1983 Tate
and Meyer 1983 McDowell and Likens 1988 Buffam et
al 2001] Most recently DOC quality has been used as a
hydrological tracer to infer htllslope source areas during
various stages of flood hydrographs [Hood et al 2006]
Nitrogen flushing from watershed source areas has also
been used to explain similar trends in dissolved organic
nitrogen (DON) and dissolved inorganic nitrogen (DIN)
G03019 1 of 12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
flood dynamics [Creed et al 1996 Creed and Band
1998a 1998b Buffam et al 2001]
[4] While allochthonous sources of organic and inorganic
materials are of obvious importance to river channel bio
geochemistry and metabolism autochthonous (i a in chan
nel) sources may also provide significant contributions of
organic and inorganic materials to the water column during
episodic transport events (i e floods and dam removals)
Once fine hillslope soils enter channel networks the fre
quency and magnitude of movement is dictated by hydro
geomorphic controls that produce and maintain erosional
and depositional features along the river continuum These
fine materials are subjected to multiple cycles of suspension
and deposition as they are transported from hillslope to
ocean This cycle of suspension and deposition in rivers
represents an important link among hydrology geomorphol
ogy and biogeochemistry We assert that this cycle driven
by flow conditions and channel geomorphology results in the
biogeochemical processing of watershed materials during
downstream transport and thus represents an in channel or
autocthonous influence on water column b►ogeochemistry
[5] Deposited fine sediments in aquatic ecosystems serve
as benthic substrate with high denrtrif►cation potential
[Pmay et al 2000 Wetzel 2001] and also as adsorptive
sinks for dissolved organic matter (DOM) [McDowell and
Wood 1984 Nelson et al 1993 Aufdenkampe et al 2001 ]
NH4 [Triska et al 1994 Schlesinger 1997] and SRP
[Meyer 1979 Klotz 1988 Mulholland 1992] During
interflood periods in which quiescent conditions dominate
benthic sediments may become anoxic producing strong
redox gradients which lead to the accumulation of DOM
inorganic N P and vanous reduced tenninal electron
acceptors such as Fe 2+ in interstitial waters [Wetzel 2001 ]
However during periods of high discharge these sediments
are subject to resuspension and may become an internal load
(i a in channel source areas) of dissolved inorganic and
organic forms of C N P and Fe to the water column via
two pathways interstitial water release and desorption from
sediment mineral surfaces
[6] The concept of internal loading from anoxic hypo
lemma and pore waters in lake ecosystems is well estab
lished and represents an appreciable source of P and N in
mictic systems [Wetzel 2001] In rivers however it is
unlikely that pore water release could produce a measurable
increase in water column biogeochemistry because of dilu
tion effects On the other hand sediment desorption may
represent an important source of C N P and/or Fe to the
water column
[7] There is some experimental evidence that suggests
suspended sediments could be a considerable source of
DOM to aquatic ecosystems Reagent grade clay mineral
surfaces sorbed appreciable quantities of DOM from leach
ate solutions and simulated solar radiation facilitated the
desorption of previously accumulated DOM [Tieyen et al
2005] Additionally recent experimental results involving
Mississippi River deltaic suspended solids in distilled water
and artificial seawater solutions demonstrated photod►sso
lution of POC [Mayer et al 2006] Finally photoproduc
tion of DOC in sediment laden estuarine waters was also
demonstrated in recently published work [Kieber et al
2006] These studies suggest that suspended materials in
the presence of light can provide a measurable supplement
of DOM to river ecosystems
[8] In rivers the resuspension of fine sediments likely
represents a source of DOM to the water column during
transport events such as floods or dam removals As
previously shown the desorption or dissolution of DOM
from sediment surfaces is accelerated in the presence of
light Further the photochemical mineralization of desorbed
DOM may release inorganic forms of C N P and Fe as has
been demonstrated in nvers lakes and estuaries [McKnight
et al 1988 White et al 2003 Vahatalo and Wetzel 2004
Vahatalo and Zepp 20051 We propose that photoassisted
desorption coupled with photochemical mineralization
represents an internal load of dissolved organic and ►nor
gan►c matter from sediment surfaces to the water column
during floods While it is likely that watershed contributions
(► a external loads from watershed flushing) to flood
biogeochem►stry are more important from a total load or
flux perspective the concept of internal loading from
sediment suspensions offers important insight into channel
biogeochem►cal processing (transfonnation) of hillslope
materials routed to coastal ecosystems
[9] Laboratory experiments replicating the resuspension
of river sediments were conducted in the presence of
simulated solar radiation to determine if photoassisted
sediment desorption of DOM could contribute measurable
fractions of DOC to the water column during floods
Additionally organic and inorganic forms of N P as well
as total dissolved iron (Fed) were measured to determine if
the photochemical mineralization of desorbed DOM would
further enrich the water column Thus we tested whether the
resuspension of fine sediments within river ecosystems
represents an internal source of dissolved C N P and Fe
2 Methods
2 1 Overview of Approach
[►o] In laboratory experiments we replicated total sus
pended solids (TSS) concentrations observed during floods
with recurrence intervals of 15 years [Qi 5 Simon et al
2004] on the Deep River NC DOC dissolved inorganic
carbon (DIC) total dissolved nitrogen (TDN) DIN DON
soluble reactive phosphorus (SRP) and Fed concentrations
were measured before and after exposure to simulated solar
radiation and relative to dark controls to determine whether
photoassisted sediment surface desorption and photochem
►cal mineralization of desorbed DOM contributes to water
column biogeochem►stry dynamics during sediment suspen
sion events (i a floods and dam removals)
22 Site Description
[n] The Deep River is a 6th order system draining
—2 770 kmz of the Central Piedmont in North Carolina
(Figure 1) land uses in the watershed are predominantly
forest (72 %) agriculture (25 %) and urban (3 %) Within the
Deep River watershed is a Level IV ecoregion (as desig
nated by the US Environmental Protection Agency) known
as the Triassic Basin characterized by wider valleys with
lower relief than other Piedmont areas Soils within this
ecoregion are clay rich because of Lower Mezozo►c sed►
mentary parent material consisting of unmetamorphosed
shale sandstone mudstone siltstone and conglomerates
2of12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
M
800 w 79 4JV w Ts 3jV w »1 Dw
IL
z z
► o 75 130 M0 Iowmews
z z
a
Ramseur Gage 'Sr
ma tore Gag
S
M �
J 7 f-,I 8arbonton,Gam= I R
X _ z
m m
n
wrornereis
so 010 W 79 400 W rs 309"w 79 in w
Figure 1 Deep River watershed map detailing location of USGS stream gages and the former
Carbonton Dam (Latitude 35 °31'N Longitude 79021'W)
[Horton and Zullo 1991 ] These particular clays exhibit
high shrink swell capacities [see Velde 1992] thus they are
capable of adsorbing considerable quantities of organic
material
[12] The sites used for sediment and water collections are
located within the Triassic Basin and were impounded by a
run of river dam ( Carbonton Dam Figure 1) that was
removed following our study We chose these particular
sites because of their clay rich geology and the abundance
of accumulated fine sediments that were partially mobilized
during dam removal In addition this section of the Deep
River often exhibits considerable turbidity well after high
discharges have subsided This is attributed to the character
of fine clay particles in transport which are easily mobilized
and remain in suspension regardless of hydraulic condition —
a process sometimes referred to as washload [Leopold
et al 1964] Washload sediments may therefore be sub
jected to solar radiation for several days during and after
floods
2 3 Hydrogeomorphic and Environmental Scaling
[13] Sediment concentrations used in our experiments
were scaled to common transport events using USGS data
from two gages that envelope the study reach (Figure 1)
The upstream gage (Ramseur NC USGS # 02100500) is
located approximately 35 km from the upstream extent of
3of12

G03019
1000
800
600
E
N 400
200
J
E
to
600
500
400
300
200
100
RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS
TSS vs Q
Deep River at Ramseur NC
0 50 100 150 200 250 300 350
Q (MIS')
TSS vs Q
Deep River at Moncure NC
0 100 200 300 400 500 600 700
Q (m's')
Figure 2 TSS versus Q (mean daily values) curves from
two gages surrounding sampling sites on the Deep River
NC Dashed vertical lines mark Q values representative of
flows with a 15 year recurrence interval (Q, 5)
the former impoundment The downstream gage (Moncure
NC USGS # 02102000) is located approximately 35 km
downstream of the former dam Gage data were analyzed
using standard recurrence interval (RI) analyses [Knighton
1998] and all available TSS measurements were plotted
against discharge (Q mean daily flow) for the TSS collec
tion dates (Figure 2) The upstream gage has an 82 year
record with 46 TSS measurements the downstream gage
has a 74 year record with 126 TSS measurements Plots of
TSS versus Q were generated to determine the appropriate
TSS concentrations at 15 year recurrence intervals (Q, 5)
We determined that the system exhibited TSS concentra
tions of 200 -400 mg L -1 during Q, 5 events
[14] Solar radiation exposure was delivered to each
treatment using an Atlas Suntest XLS+ solar simulator
equipped with an arc xenon lamp The lamp was call
brated to deliver radiation equivalent to the amount
received by the study reach (Latitude 35 °31'N Longitude
79 °21'W Figure 1) during one clear summer day The
solar simulator supplies a light intensity of 650 W /m2
which delivered a total light dose of 14 000 kJ /m2 over a
course of 6 h A forced air cooling system kept water
solutions at 25 f 0 7 °C (n = 9 f represents 1 standard
deviation) during the exposure process Quartz tubes (5 cm
diameter x 33 cm length oriented in a flat position relative
to the light source) were used for all treatments as quartz
transmits full spectrum sunlight
2 4 Sediment and Water Collections
G03019
[15] Sediment cores and river water were collected on
four separate occasions during September and October 2005
from the impounded reach of the Deep River Sample
collections occurred before the reach was affected by dam
removal (20 October 2005) Significant sediment accumu
lations are often associated with run of river impoundments
Such sediment accumulations are typically undisturbed for
long periods of time allowing for the development of strong
redox gradients Particle size distribution of collected sedi
ments was determined to be 9% sand 49% silt and 42%
clay (using methods from Dane and Topp [2002])
[16] During all sampling trips one core was collected
from each of six randomly selected sites within the impound
ment Cores were collected by inserting 30 cm by 5 cm
diameter polycarbonate sleeves into the soft submerged
sediment deposits along channel margins Each sleeve was
pushed into the sediments until the top was flush with the
sediment surface The top opening was capped and the
sleeves were removed from the sediment at which time
the bottom of the sleeve was also capped Tape was used to
seal the caps to the core sleeves and the cores were placed
in a light proof cooler packed with ice and transported to the
laboratory Upon arrival cores were stored overnight in a
light proof container at 4 °C
[17] On each sampling date approximately 15 L of river
water was collected in acid washed HDPE containers at a
location central to the six coring sites Water samples were
packed on ice in dark coolers for transport to the laboratory
Collected river water was stored at 4 °C and filtered within
12 h of collection using precombusted 0 7 µm glass fiber
filters (Whatman GF/F filters were used in all filtration
unless otherwise mentioned)
2 5 Experimental Sediment Preparation
[18] In the lab overlying water was poured off each core
and the sediment was carefully pushed out of its sleeve and
into a plastic bag purged with N2 gas to minimize air
exposure When all six cores were added the bag was then
further purged with N2 gas and sealed Sediment homoge
nization was accomplished with vigorous hand kneading for
several minutes A grab sample of the homogenized sedi
ments was then collected for addition to each sediment
treatment Remaining homogenized sediments were again
purged with N2 gas sealed and stored for future expen
ments in a light proof container and kept at 4 °C for no more
than 48 h Experiments were run for three consecutive days
using the same sediment stock
[i9] To deliver the appropriate mass of sediment to each
treatment reliably and repeatedly we developed a sediment
volume to dry mass curve Two modified syringes (BD
20 mL and 3 mL) were used as small conng devices which
allowed for reliable volumetric delivery to predned and
preweighed crucibles Crucibles were dried at 100 °C for
24 h and reweighed It was determined each 1 mL of wet
homogenized sediment was approximately equivalent to
600 mg of dried mass (R2 = 0 99 n = 6) Thus selection
of 0 5 mL of wet sediment was determined as the appro
pnate volume to add 310 f 0 06 mg of TSS (n = 83 t
represents 1 standard deviation) to each treatment
4of12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
Table 1 Experiments Data Collected and Key Calculations
Rate of DOC
Experimental Constituents Measured,
Sampling Frequency
Experiment Treatments b
Control(s) Calculated
Key Calculations
Photoassisted I RW w /sediments light
Dark and DI treatments DOC COz DIC
Two sediment/water (RW and DI)
desorption 2 RW w/o sediments light
TDN DIN DON
solutions were mixed and destructively
3 RW w /sediments dark
and SRP
sampled to measure initial concentrations
4 RW w/o sediments dark
before incubation Following
5 DI w /sediments light
incubation all treatments
6 DI w/o sediments light
were destructively sampled to measure final
7 DI w /sediments dark
1 RW w /sediments light
condition [Desorption] = [Final] — [Initial]
8 DI w/o sediments dark
production
2 RW w/o sediments light
Rate of DOC
1 RW w /sediments light
Dark treatments
DOC One quartz tube from each treatment
desorption
five identical solutions
was destructively sampled at each of
prepared in quartz tubes
the following times during incubation
2 RW w/o sediments dark
(hours) 0 0 5 1 2 and 6
five identical solutions
prepared 1 in quartz tubes
Rate CO2
1 RW w /sediments light
dark and w/o
COz The headspace of each treatment was sampled
production
2 RW w/o sediments light
sediments treatments
at the following times of incubation
3 RW w /sediments dark
(hours) 0 2 4 and 6
Fed removal
1 RW w /sediments light
dark treatments
Fed [Fed] were determined for RW (no sediments)
2 RW w /sediments dark
and a mixture of sediment/RW
3 RW w/o sediments light
(destructively sampled)
4 RW w/o sediments dark
before incubation
Reported as percent removal
%Removal = ([Initial] — [Final]) /[imtial]
Rate of Fed I RW w /sediments light NA Fed One replicate was destructively sampled at each
removal five identical solutions of the following times during incubation
prepared in quartz tubes (hours) 0 0 5 1 2 and 6
2 RW w/o sediments dark
five identical solutions
oreuared in ouartz tubes
RW denotes filtered Deep River water
bDI denotes deionized water
The source of Fed was river water no DI treatments were used
2 6 Experimental Design
[2o] A series of experiments were performed under lab
oratory conditions (see Table 1 for summary) In all expert
ments except those measuring the photoproduction of DIC
(discussed below) homogenized sediment subsamples of
0 5 mL were added to acid washed quartz tubes using a
modified 3 mL syringe followed by the addition of 950 mL
of either river water (RW) or deionized water (DI) The
resulting sediment/water solutions were generated at the
target concentration of —325 mg L- i that completely filled
the volume of the quartz tubes so that no head space
remained All tubes were gently mixed at the beginning of
the expenments to suspend all sediment and incubated
under the appropnate conditions for 6 h Once adequately
mixed the majority of the sediments stayed suspended for
the duration of the experiments in all treatments (i a light
and dark) as fine clay particles remain suspended irrespec
tive of hydraulic condition (i a wash load [Leopold et al
1964])
2 6 1 Photoassisted Desorption of C, N and P
[21] To quantify the photoassisted desorption of dissolved
C N and P from sediment surfaces a serves of expenmental
runs were performed as follows Four quartz tubes contain
mg sediment suspensions (in RW and/or DI) were placed
directly in the solar simulator (referred to hereafter as
Light) In order to account for microbial mediated process
es another four suspensions (in RW and /or DI) were
wrapped in aluminum foil and incubated on the lab bench
directly adjacent to the solar simulator (referred to hereafter
as Dark) Before incubation two additional sediment water
solutions were destructively sampled to represent the pre
exposure (Initial) concentrations of DIN TDN SRP and
DOC This approach was chosen over sampling each tube
before exposure to maintain controlled volume conditions
for mass balance calculations and to avoid possible exper
imental error associated with altered diffusive gradients
Following a 6h exposure the solutions were sampled for
DIN TDN SRP and DOC representing the final condition
(Final) Desorption of C N and P in all treatments were
quantified by subtracting Final from Initial concentrations
Experimental runs were repeated until each treatment was
represented by nine samples
[22] DOC and TDN samples were filtered into glass TOC
vials and acidified to pH 2 with 2 M ultrapure HCl Samples
were stored at 4 °C until analyzed using a Shimadzu TOC V
CPH total organic carbon analyzer coupled with a TNM 1
total nitrogen measuring unit This particular instrument
utilizes the high temperature combustion method when
quantifying DOC concentrations TDN is measured using
the chemilummesence method Samples were analyzed
within 2 d of collection
[23] DIN and SRP samples were filtered into acid washed
125 mL amber HDPE bottles Samples were frozen at
—20 °C until analyzed for NH4 N NO3 NO, N and SRP
by the Analytical Services Laboratory at North Carolina
State University in Raleigh NC using an FIA autoanalyzer
5of12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
DIN measurements from each treatment were subtracted
from TDN measurements to determine DON content
[24] Statistical analyses were conducted using analysis of
variance adjusted for multiple observations within days
Mayor effects analyzed included light treatment (light
versus dark) and water treatment (DI versus RW) All
statistical results were analyzed using a significance level
of 0 0250
2 6 2 Photoproduchon of DIC
[25] CO2 accumulation in the head space of the gas tight
quartz tubes was measured at the end of 6h incubations for
three experimental runs These experiments were prepared
as detailed above with two exceptions (1) sediment and
water volumes were reduced and (2) an additional control
was used to account for photomineralization of DOC in RW
without sediments The volume of sediment was reduced by
one half (0 25 mL) as was the water volume (475 mL) This
maintained the same concentration of sediment (325 mg
L -i) but allowed enough head space for gas sample
collection CO2 concentration measurements were used to
determine differences in photochemical mineralization rates
(i a change in DIC concentrations) among treatments
[26] Headspace gas samples were taken from each treat
ment using 1 mL plastic syringes and measured within 1 h of
sampling for CO2 concentration on a Shimadzu GC 14A gas
chromatograph equipped with a Supelco 80/100 parapet
Q column (6 ft x 1/8 in) a methanizer (set at 500 °C) and a
flame ionization detector Oven temperature was set at 35 °C
and the runtime was 7 min Helium was used as the carrier
gas
[27] Corrections for CO2 concentrations were needed to
determine the photochemical mineralization of C from
sediment sources only In the case of DI treatments the
only source of CO2 was the sediments themselves To
account for biological respiration dark control CO2 (PPM)
measurements (sediments and water) were subtracted from
light treatments For RW treatments sediments are a CO2
source but DOC contained within river water also contrib
uted to CO2 production via photomineralization Therefore
CO, production from sediments in RW treatments was
calculated by subtracting both RW light/without sediment
and RW dark (with sediment) treatments from RW light
sediment treatments
[28] Changes in CO, concentrations within the head space
of each treatment were used to calculate changes in DIC
concentrations Measurements of CO2 (g) were converted to
mmol C using the Ideal Gas Law DIC was then calculated as
H2CO3 and HCO3 using an assumed solubility constant
(Kc02) of 3 38 x 10 -2 mol L -i atm i [Pankow 1991
Stumm and Morgan 1996] For both RW and DI treatments
calculations were based on an assumed pressure of 1 atm a
measured temperature of 25 °C and a measured pH of 7
2 6 3 Fed Dynamics
[29] Fed dynamics were also quantified during the expo
sure process Fed is reported for RW treatments only as
river water was the source of Fed in our experiments Fed
was determined before and after light exposure in two
experimental runs using three treatments light with sedi
ments and dark with and without sediments Fed data are
presented as percent difference because of considerable
variability in Fed concentrations within stock river water
between experimental runs (2 -18 µmol L-1)
[3o] Fed was determined using the ferrozme colonmetnc
method with hydroxylamme hydrochloride as a reducing
agent [Stockley 1970] Samples were immediately filtered
upon collection using sterile 0 22 µm nylon filters and
analyzed within 3 h of collection using a Beckman DU
650 UVNis spectrophotometer
[31] Statistical analyses were conducted using analysis of
variance adjusted for multiple observations within days
Mayor effects analyzed included light treatment (light
versus dark) and presence of sediments (with or without)
All statistical results were analyzed using a significance
level of 0 0250
2 6 4 Rate Experiments
[32] Using the appropriate analytical techniques described
above we performed additional experiments to characterize
the kinetics of photoassisted DOC desorption DIC photo
production and Fed removal in RW sediment solutions
2 6 4 1 Rate of DOC Desorption
[33] River water and sediments were infixed in five quartz
tubes as detailed in the photoassisted desorption section
above This rate experiment used a destructive sampling
design that permanently removed a quartz tube from expo
sure at each sampling interval Sampling consisted of five
measurements throughout the 6 h exposure process 0 0 5
1 2 and 6 h Water samples were filtered and analyzed for
DOC to establish a relationship between DOC concentration
and exposure duration
2 6 4 2 Rate of DIC Photoproduchon
[34] Two quartz tubes were filled with a mixture of river
water and sediments (light treatment and dark control) as
detailed in the photoproduction of DIC section above and a
third tube was designated as the RW light control —thus
contained no sediments Headspace gas sampling consisted
of four measurements throughout the 6 h exposure process
0 2 4 and 6 h CO2 concentration measurements were used
to determine differences in photochemical mineralization
rates (i a change in DIC concentrations) among treatments
2 6 4 3 Fed Removal
[35] A Fed removal rate experiment was performed using
the same destructive sampling design of the DOC rate
expenment described above The source of Fed in this
experiment was filtered Deep River water Data are plotted
relative to exposure which includes a negative time repre
senting RW Fed concentrations before sediment addition
3 Results
[36] Exposing sediment water solutions of concentrations
relevant to floods in the Deep River system with recurrence
intervals of 1 5 years to simulated solar radiation increased
DOC DIC and TDN concentrations while scouring Fed
from the water column changes in SRP concentrations were
negligible Reported C values were corrected by subtracting
dark controls (respiration) and where appropriate light
controls without sediments (i e DIC production expen
ments) All C N and P values are reported with a range of f
1 standard error (SE)
31 Photoassisted Desorption of C, N and P
[37] DOC desorption from suspended sediment surfaces
was significantly higher in both DI and RW light treatments
compared to dark controls (p < 0 001 n = 56 Figure 3a)
6of12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
There were no statistically significant differences between
RW and DI treatments (p = 0 665 n = 56) DOC increased
in DI light treatments by 414 t 6 µmol C L -1 d -i while
DOC increased in RW light treatments by 37 7 t 4 µmol C
L -1 d -1 (Figure 3a)
50
40
J
E 30
Z
a 20
0
m 10
w
c
r 0
U
10
120
100
J
6 80
E
c 60
U
o 40
a
N
U 20
180
160
'e 140
J
c 120
E
100
v 80
S
1& 60
r-
40
20
DI Light DI Dark RW Light RW Dark
DI RW
DI RW
[38] Reported CO, (and DIC) values were corrected by
subtracting dark controls (for both DI and RW series) and
in the case of RW light sediment treatments an additional
correction was made by subtracting CO2 production from a
second control (RW without sediments) This final correc
tion accounted for CO2 production associated with photo
mineralization of DOC within Deep River water not
associated with sediment desorption
[39] Changes in corrected CO2 concentrations did not
show significant differences between water treatments (i e
DI versus RW) That is CO2 production associated with DI
light sediment treatments were not significantly different
than those of RW light sediment treatments (p < 0 03 n = 9
Figure 3b) Both light sediment treatments resulted in
significantly higher CO2 production than dark controls
(p < 0 001 n = 9) The additional control treatment (RW
without sediments) accounted for 54 f 9 µmot C L -1 d -1 of
CO2 production while the CO2 production in RW light
sediment treatments was 92 f 17 µmol C L -1 d- 1(Figure 3b)
Thus sediments in Deep River water increased photoproduc
tion of CO2 by 38 ttmol C L -1 d -1 or 41%
[4o] Based on our calculations using the corrected CO2
production values reported above suspended sediments in
DI and RW treatments increased DIC concentrations by 116
and 81 µmot C L -1 d -1 respectively (Figure 3c) This DIC
pool includes both headspace and water column inorganic
C DIC and DOC pools were added together to quantify the
total dissolved carbon (TDC) contributed from suspended
sediment surfaces during exposure to simulated solar radi
ation 158 and 119 µmol C L -1 d -1 in DI and RW treat
ments respectively (Figure 3c)
[41] As was the case with DOC sediments in both RW
and DI light treatments were a statistically significant
source of TDN enrichment compared to dark controls
(p < 0 001 n = 52 Figure 4) Water type (i e RW versus
DI) did not generate significant differences in TDN desorp
tion from sediment surfaces (p = 0 15 n = 52) Mean water
column enrichment of TDN in DI light treatments was 2 7 f
0 6 µmol N L -1 d -1 while mean enrichment in RW light
treatments was 1 7± 0 7 µmol N L -1 d -1 In both water
treatments nearly all TDN enrichment was identified as
DON (99 %)
[42] Sediments in both water treatments in the presence of
simulated solar radiation resulted in slightly higher concen
trations of SRP (0 32 t 0 3 µmot L -1) however these
values are at or near instrumentation detection limits of
Figure 3 (a) DOC desorption from suspended sediments
in light and dark treatments Error bars represent tl SE n =
9 per treatment (b) CO2 production for DI with sediments
was corrected by subtracting CO2 production of dark
controls River water with sediment CO2 production was
corrected by subtracting both river water /no sediments in
light and river water /sediment dark controls River water /no
sediment light controls are also shown for comparison
purposes Error bars represent fl SE n = 3 per treatment
(c) Total dissolved carbon (TDC) desorption in the presence
of light TDC was calculated as the sum of changes in [DIC]
and [DOC] resulting from exposure to simulated solar
radiation Error bars represent f1 SE
7of12

G03019
350
v
250
0
E
1 50
z
0
t
5 050
0
tLi
050
1 50
RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
DI Light DI Dark RW Light RW Dark
Figure 4 TDN desorption from suspended sediments in
light and dark treatments DON accounted for the vast
majority (99 %) of TDN desorption Error bars represent
fl SE n = 9 per treatment
0 3 µmol L-' Thus the increase in SRP concentrations was
neglected for further analyses
3 2 Fed Removal
[43] Sediments in both light and dark treatments removed
Fed from river water (p < 0 005 n = 9 p < 0 0250 n = 9
respectively) relative to dark controls without sediments
(Figure 5) There were no significant differences between
light and dark sediment treatments (p = 0 21 n = 9) Thus
the presence of sediments independent of light controlled
Fed dynamics in our experiments
3 3 Rate Experiments
3 3 1 DOC Desorption and Photoproduction of DIC
[44] Even though the DOC desorption rate expenment
produced considerably greater DOC enrichment than any of
the experimental values reported above (150 µmol C L -i
compared to 41 µcool C L -1) the kinetics demonstrated by
these experiments are still considered representative of the
general process of photoassisted desorption As seen in
Figure 6a most of the desorption activity (67 %) took place
in the first 2 h of exposure which produced a DOC
concentration increase of 100 µmol C L -i Sampling during
the fourth and sixth hours of exposure showed contributions
of an additional 34 and 14 µmol C L -1 respectively
[45] Photoproduction of DIC rate experiments show sim
ilar results to those of DOC desorption kinetics That is
70% of the COz produced by photochemical processes
occurred in the first 2 h of exposure (Figure 6b)
3 3 2 Fed Removal
[46] The Fe rate experiments showed that once sediments
were suspended in RW there was an immediate 24%
removal on Fed in both light and dark treatments (Figure 6c)
Following the first 2 It of incubation sediments in light
treatments removed 60% of Fed while sediments in dark
treatments removed 48% The dark treatment did not
remove any more Fed over the course of the remaining
incubation However Fed concentrations continued to drop
in the light treatment throughout the exposure process
eventually removing 86% of Fed (Figure 6c) Fed removal
in both treatments was within ±1 SE of the respective mean
removal values reported above (section 3 3 and Figure 5)
4 Discussion
41 Experimental Discussion
[47] Our experimental results provide insight into the
biogeochemical roles of suspended sediments during floods
and dam removals When exposed to simulated solar
radiation TSS concentrations representative of Q1 5 events
contributed dissolved organic carbon and nitrogen to the
water column (Figures 3a 3b 3c and 4) while effectively
removing Fed (Figures 5 and 6c) TDN and TDC loads
in RW treatments equaled 1 7± 0 7 µmol N L- 1 d- 1 and
119 pmol C L -1 d -i (Figures 4 and 3c) In the following
section our results were extrapolated to provide context at
environmentally relevant scales and compared to other
reported values of C and N fluxes in aquatic ecosystems
[48] Removal of Fed in sediment treatments (Figure 5)
can be explained by sediment surface adsorption as fine
sediments have an affinity for Fe 3+ oxides Previous re
search has suggested this relationship in an Eastern Shore
aquifer in Virginia where extractable Fe from oxic aquifer
sediment surfaces was an order of magnitude greater than
anoxic sediments from the same system [Knapp et al
2002] Additionally Fe3+ has been demonstrated to alter
surface charges on clay particles inducing coagulation and
sedimentation of Fe clay complexes [Pierre 1997 Ma and
Pierre 1999]
4 2 Extrapolation and Comparison
[49] While the photochemical processes described here
are not as significant to flood biogeochemistry as has been
previously demonstrated for watershed hillslope source
areas (i a external loading) they do represent pathways
of internal OM loading For similar results to be expected in
natural systems there are certain conditions which must be
met particularly a combination of adequate solar radiation
and elevated concentrations of fine suspended sediments
100%
80%
�v
0 60%
m
40%
m
LL
20%
0%
Light w /sediments Dark w /sediments Dark no sediments
Figure 5 Fed removal efficiency in treatments containing
sediments was not significantly different in the presence or
absence of light Error bars represent f1 SE n = 3 per
treatment
8of12

G03019
200
150
J
a
100
RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS 603019
Time of exposure (hours)
E 2000 —4— River water wiseds W t
IL • Rner water notseds 6gM
1600 - River water Wseds dark
U
X111;
m
v
° 800
° 400 • —
a
- - -�"-
J
v
r
Time of exposure (hours)
1 0 1 2 3 4 5 B
Tkm of exposure (hours)
Figure 6 (a) DOC rate of desorption in river water
sediment mixtures light versus dark Note that the total
desorption of DOC during this experiment (- 150 µmol L -' )
was higher than mean desorption values reported in the text
(38 timol L -' d -') (b) Results from CO2 production rate
experiment Photommeralization occurred predominantly
during the first 2 h of exposure to simulated solar radiation
but continued through the course of the experiment (c) Fe
removal rates from sediment and river water mixtures in
light and dark conditions Data are plotted relative to
exposure which includes a negative time representing RW
Fed concentrations before sediment addition
Such conditions are often observed during floods as flood
waves are routed through reaches for hours to days follow
mg precipitation events Dam removals and the actions of
natural ecosystem engineers such as carp cow and hippo
potamus may create sediment suspensions as well
[5o] There are challenges in applying our results directly
to natural systems because they simulate processes occur
ring within the top 5 cm of the river water column only
(depth of quartz tubes used for experimentation) Under the
assumption that all DIC pools presented in our results were
first desorbed as DOC and based on mean desorption
values in RW treatments suspended sediment loads within
the upper 5 cm of the water column could contribute
119 mmol DOC m -3 d -' and 1 7 mmol DON m -3 d -'
It is important to note that in our experiments most C (65%
or 77 mmol C m -3 d -1) was mineralized during the course
of exposure and was therefore DIC (Figure 3c) Such
photochemical mineralization rates are comparable to pre
vious studies in humic lake ecosystems which report C
photochemical mineralization rates in the surface layer of
the water column to be 19 to 57 mmol C m -3 d -' [ Pahatalo
et al 2000 Anesto and Graneh 2003] These systems are
located in northern latitudes of Europe and thus photo
chemical mineralization was limited by latitudinal controls
[51] Since our data represent photochemical reactions
with suspended sediment in the top 5 cm of the water
column only other data are required to calculate this
contribution throughout the water column including extmc
tion coefficients and degree of npanan shading Since these
data are not currently available for the Deep River similar
data from the nearby Neuse River NC were used [from
Vahatalo et al 2005] so the following extrapolations are
limited in this respect We assumed that UV A (320 -400 nm)
was responsible for the reactions seen in our experiments
based on previous research which showed that 68% of the
photochemical mineralization of DOM in a humic lake was
accomplished by UV A [Vahatalo et al 2000] For the
purposes of this analysis we used 360 nm to represent the
wavelength of light responsible for the photoassisted de
sorption and photochemical mineralization of DOM from
suspended sediments The extinction coefficient for 360 rim
on the Neuse River at Goldsboro NC is 17 6 m -' and
npanan shading at this same site reduces light availability
to the channel by 58% [Vahatalo et al 2005 personal
communication] The average channel width of the Neuse
River at this site is 36 in while the average channel width of
the Deep River at Carbonton is 40 in Thus we assume that
riparian cover effects on light availability are similar
Incorporating these variables suspended sediments account
for 3 92 mmol C m -2 d -' and 40 µmol N m -2 d -'
[52] Values allowing direct comparisons of our DON data
were not readily available in the literature because most
efforts involving photochemistry are focused on mmerali
zation, specifically We were unable to detect N minerah
zation in our experiments (i e DIN concentrations did not
increase during exposure) However our N ennchment (as
DON) is comparable to the photochemical ammonification
of DON in the Baltic Sea which was found to generate
53 µmol of NH4 M-2 d -' [Vahatalo and Zepp 2005]
[53] Photoassisted desorption of DOC from suspended
sediments as reported here is comparable to phytoplankton
productivity at the Goldsboro NC site on the Neuse River
9of12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
5 33 mmol C m -2 d -i [Yahatalo et al 2005] It is not
comparable however to the mean daily load of DOC in the
Deep River near Carbonton At a mean DOC concentration
of 595 µmot L-1 (calculated during these experiments n =
9) and a mean annual Q of 113 m3 s- i (USGS gage #
02100500) the Deep River transports 5 81 x 108 mmol C
d-' Once this is normalized to mean channel width (40 m)
assuming flow across the channel is evenly distributed this
equals 145 x 107 mmol C m -2 d-1 Therefore photo
assisted sediment surface desorption of DOC during floods
is minimal compared to Deep River mean daily loads of
DOC However this mechanism of DOC enrichment may
supplement organic C losses caused by reduced phytoplank
ton productivity during floods
[54] Dams are effective sediment traps within channel
networks and therefore likely suppress sediment processing
in rivers However their removals may present the proper
conditions for photochemically mediated sediment desorp
tion of DOM Accumulated sediments behmd Carbonton
Dam were composed of predominately silt (49 %) and clay
(42 %) fractions Following dam removal fine sediments
were observed exiting the impoundment during floods in
concentrations greater than 350 mg L -t (J A Riggsbee
unpublished data 2005 and 2006) During reservoir dew
atenng (20 October 2005) impounded water was released
from the structures sluice gates and fine sediments were
exported from the reservoir in concentrations greater than
200 mg L -t (J A Riggsbee unpublished data 2005) Our
results suggest that dam removal activities are capable of
affecting downstream water quality through the photoas
sisted desorption of DOM from suspended sediment surfa
ces in transport
4 3 Photochemical Processing of Terrestrial C
[55] Clay and silt soils are responsible for halting the
movement of DOC from soil pools to streams In particular
polysacchandes f ilvic and humic materials are associated
with clay soil surfaces [Tisdall and Oades 1982 Oades
1988 Tiessen and Stewart 1988] On the basis of these
relationships clay and silt minerals exert strong controls on
stream biogeochemistry by limiting DOC transport from
hillslopes to channels In combination with the photochem
ical processes described by our data the movement of clay
minerals from hillslopes to river channels represents an
appreciable source of DOM to river systems Desorption
and mineralization of terrestrial C associated with sediment
surfaces is a gradual process driven by hydrogeomorphic
features within fluvial systems as sediments are slowly
routed through channel networks
4 4 Suspended Sediment Controls on Dissolved Iron
Dynamics in River Ecosystems
[56] Iron oxides in the presence of light can mediate
the oxidation of DOM in natural aquatic ecosystems
through metal ligand surface complexation reactions
[McKnight et al 1992] In acidic surface waters the
photochemical mineralization of DOM is coupled with
the photoreduction of Fe 3+ to the more soluble Fe 2+
which can increase Fed [McKnight et al 19881 In less
acidic systems photo Fenton reactions oxidize reduced
iron [Cooper et al 1988 Zuo and Hoigne 1993]
maintaining the availability of metal ligand complexes
for continued DOM mineralization [Brinkmann et al
2003] Results from our experiments suggest fine river
sediments suspended during and following floods can
rapidly scour the water column of Fed (Figures 5 and 6c)
Thus sediment suspensions reduce the importance of metal
ligand complexation reactions during photochemical miner
alization
[57] There are interesting implications for Fed removal
from the water column during floods beyond that of photo
chemical mineralization The ability of clay minerals to
adsorb Fed from the water column during suspension events
offers a source of Fe 3+ to benthic microbial communities
responsible for Fe reduction reactions Knapp et al [2002]
suggest that Fe reduction can be fueled by Fe 3+ found on
mineral surfaces in aquatic ecosystems This ability of fine
sediment to remove dissolved Fe from river water also affects
the quality of Fe delivered to coastal ecosystems That is as
the world s rivers are subjected to greater sediment budgets
resulting from urbanization more Fe will be delivered to the
world s oceans in particulate forms
Conclusions
[58] Hydrogeomorphic controls on the suspension and
deposition of fine sediments within river corridors have an
often overlooked role in the regulation of OM processing in
rivers via two mechanisms (1) photoassisted desorption of
DOM which transfers and oxidizes C from suspended
sediments and (2) Fed adsorption which diminishes the
importance of photo Fenton reactions in photochemical
mineralization of DOM delivers Fe 3+ (necessary for Fe
reducing bacteria) from the water column to the anoxic
benthos following deposition and alters the form of Fe
delivered to the world s oceans
[59] Altered land use has increased both erosion rates and
the frequency and magnitude of events that are capable of
suspending fine sediments in rivers [Wolman 19671 Con
sequently altered hydrology has increased the flux of C
containing materials from watershed to channel It is unclear
whether this hydrological control would increase or de
crease the efficacy of photochemical reactions on C pro
cessmg during suspension events However it is reasonable
that with increased frequency of suspension and increased
fine sediment loads appreciable quantities of nvenne dis
solved Fe may be transformed into particulate Fe This
relationship may influence global C cycling because fluxes
of particulate Fe are not thought to reach pelagic manne
phytoplankton communities which are often considered to
be Fe limited [Jtckells et al 2005]
[60] Acknowledgments We thank Anssi Vahatalo for his willingness
to discuss the project s design some preliminary data and reading an early
draft Much gratitude is owed to Emily Stanley for also reviewing an early
draft Jason Julian for his help in the field and Barrett Jenkins for his GIS
assistance This project was supported financially by the US Fish and
Wildlife Service and Restoration Systems LLC
References
Anesio A M and W Gmneli (2003) Increased photoreactivity by acid
ification Implications for the carbon cycle in humic lakes Limnol Ocea
nogr 48 735 -744
10 of 12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
Aufdenkampe A K J I Heges J E Richey A V Krusche and C A
Llerena (2001) Sorphve fractionation of dissolved organic nitrogen
and amino acids onto fine sediments with the Amazon Basin Limnol
Oceanogr 46 1921 -1935
Boyer E W G M Homberger K E Bencala and D M McKnight
(1997) Response charactenstic of DOC flushing in an alpine catchment
Hydrol Processes 11 1635 -1647
Boyer E W G M Homberger K E Bencala and D M McKnight
(2000) Effects of asynchronous snowmelt on flushing of dissolved or
game carbon A mixing model approach Hydrol Processes 14 3291-
3308
Bnnkmann T P Horsch D Sartonus and F H Fnmmel (2003) Photo
formation of low molecular weight organic acids from brown water or
game matter Environ Sci Technol 37 4190 -4198
Buffam I J N Galloway L K Blum and K J McGlathery (200 1) A
stromflow/baseflow comparison of dissolved organic matter concentra
tions and bioavailability in an Appalachian stream Biogeochemistry 53
269 -306
Cooper W J R G Zika R G Petasne and J M C Plane (1988)
Photochemical formation of H202 in natural waters exposed to sunlight
Environ Sci Technol 21 1156 -1160
Creed 1 F and L E Band (1998a) Exploring functional similarity in the
export of nitrate N from forested catchments A mechanistic modeling
approach Water Resour Res 34 3079 -3093
Creed I F and L E Band (1998b) Export of nitrogen from catchments
within a temperate forest Evidence of a unifying mechanism regulated by
variable source area dynamics Water Resour Res 34 3105 -3120
Creed I F L E Band N W Foster I K Morrison J A Nicholson R S
Semkin and D S Jeffenes (1996) Regulation of nitrate N release from
temperate forests A test of the N flushing hypothesis Water Resour
Res 32 3337 -3354
Dane J H and G C Topp (2002) Methods of Soil Analysis Part 4 —
Physical Methods 1692 pp Soil Sci Soc of Am Madison
Hood E M N Gooseff and S L Johnson (2006) Changes in stream
water dissolved organic carbon during flushing in three small watersheds
Oregon J Geophys Res 111 G01007 don 10 1029/2005JG000082
2006
Homberger G M K E Bencala, and D M McKnight (1994) Hydro
logical controls on dissolved organic carbon during snowmelt in the
Snake River near Montezuma Colorado Biogeochemistry 15 147 -165
Horton J W and V A Zullo (Eds ) (199 1) The Geology of the Carolinas
Univ of Tenn Press Knoxville Tenn
Jickells T D et a] (2005) Global iron connections between desert dust
ocean biogeochemistry and climate Science 308 67 -71
Kieber R J R F Whitehead and S A Skrabal (2006) Photochemical
production of dissolved organic carbon from suspended sediments Lim
not Oceanogr 51 2187 -2195
Klotz R L (1988) Sediment control on soluble reactive phosphorus in
Hoxie Gorge Creek New York Can J Fish Aquat Sci 45 2026-
2034
Knapp E P J S Herman A L Mills and G M Homberger (2002)
Changes in the sorption capacity of Coastal Plain sediments due to redox
alteration of mineral surfaces Appl Geochem 17 387 -398
Knighton D (1998) Fluvial Forms and Processes A New Perspective
383 pp Arnold London
Leopold L B M G Wolman and J P Miller (1964) Fluvial Processes in
Geomorphology 2nd Edition pp 522 Dover Pub] Inc New York
Ma K S and A C Pierre (1999) Colloidal behavior of montmonllonite
in the presence of Fe 3+ ions Colloids Surf A and B 155 359 -372
Mayer L M L L Schick and K Skorko (2006) Photodissolution of
particulate organic matter from sediments Limnol Oceanogr 51 1064-
1071
McDowell W H and G E Likens (1988) Origin composition and flux of
dissolved organic carbon in the Hubbard Brook Valley Ecol Monogr
58 177 -195
McDowell W H and T Wood (1984) Podzolization Soil processes
control dissolved organic carbon concentrations in stream water Soil
Sci 137 23 -32
McKnight D M B A Kimball and K E Bencala (1988) Iron photo
reduction and oxidation in an acidic mountain stream Science 240
637 -640
McKnight D M K E Bencala G W Zellweger G R Aiken G L
Feder and K A Thom (1992) Sorption of dissolved organic carbon by
hydrous aluminum and iron - oxides occurring at the confluence of Deer
Creek with the Snake River Summit County Colorado Environ Set
Technol 26 1388 -1396
Meyer J (1979) The role of sediments and bryophytes in phosphorus
dynamics in a headwater stream ecosystem Limnol Oceanogr 24
365 -375
Meyer J L and C M Tate (1983) The effects of watershed disturbance
on dissolved organic carbon dynamics of a stream Ecology 64 33 -44
Mmshall G W R C Petersen K W Cummins T L Bott J R Sedell
C E Cushing and R L Vannote (1983) Interbiome comparison of
stream ecosystem dynamics Ecol Monogr 53 1 -25
Mmshall G W R C Petersen T L Bott C E Cushing K W Cummins
R L Vannote and J R Sedell (1992) Stream ecosystem dynamics of the
Salmon River Idaho — An 8th order system J N Am Benthological
Soc 11 111 -137
Mulholland P J (1992) Regulation of nutrient concentrations in a tempe
rate forest stream Roles of upland riparian and instream processes
Limnol Oceanogr 37 1512 -1526
Nelson P N J A Baldcock and J M Oades (1993) Concentration and
composition of dissolved organic carbon in streams in relation to catch
ment soil properties Biogeochemistry 19 27 -50
Newbold J D (1992) Cycles and spirals of nutrients in The Rivers Hand
book I Hydrological and Ecological Principles edited by P Calow and
G E Petts pp 379 -408 Blackwell Sci Publ Oxford
Newbold J D P J Mulholland J W Elwood and R V O Neill (1982)
Organic carbon spiraling in stream ecosystems Oikos 38 266 -272
Oades J M (1988) The retention of organic matter in soils Biogeochem
istry 5 35 -70
Pankow J F (1991) Aquatic Chemistry Concepts 683 pp Lewis Publ
Chelsea Mich
Pierre A C (1997) Sedimentation behavior of kaolmite and montmorillo
rote mixed with iron additives as a function of their zeta potential
J Mater Sci 32 2937 -2947
Pmay G V J Black A M Planty Tabacchi B Gumiero and H Decamps
(2000) Geomorphic control on denitnfication in large river floodplam
soils Biogeochemistry 50 163 -182
Schlesinger W H (1997) Biogeochemistry An Analysis of Global Change
2nd ed 588 pp Elsevier San Diego Calif
Simon A W Dickerson and A Heins (2004) Suspended sediment [tans
port rates at the 1 5 year recurrence interval for ecoregions of the United
States Transport conditions at the bankfull and effective discharge?
Geomorphology 58 243 -262
Stockley L L (1970) Ferrozure A new spectrophotometnc reagent for
iron Anal Chem 42 779 -781
Stumm W and J J Morgan (1996) Aquatic Chemistry 3rd ed 1022 pp
John Wiley New York
Tate C M and J L Meyer (1983) The influences of hydrologic condi
pons and successional state on dissolved organic carbon export from
forested watersheds Ecology 64 25 -32
Tiessen H and J W B Stewart (1988) Light and electron microscopy of
stained microaggregates The role of organic matter and microbes in soil
aggregation Biogeochemistry 5 312 -322
TietJen T A V Vahatalo and R G Wetzel (2005) Effects of clay mineral
turbidity on dissolved organic carbon and bacterial production Aquat
Sci 67 51 -60
Tisdall J M and J M Oades (1982) Organic matter and water stable
aggregates in soils J Soil Sci 33 141 -163
Tnska, F J A J Jackman J H Duff and R J Avanzino (1994) Am
monium sorption to channel and riparian sediments A transient storage
pool for dissolved inorganic nitrogen Biogeochemistry 16 67 -83
Vahatalo A V and R G Wetzel (2004) Photochemical and microbial
decomposition of chromophonc dissolved organic matter during long
(months years) exposures Mar Chem 89 313 -326
Vahatalo A V and R G Zepp (2005) Photochemical mineralization of
dissolved organic nitrogen to ammonium in the Baltic Sea Environ Sci
Technol 39 6985 -6992
Vahatalo A V M Salkingla Salonen P Taalas and K Salonen (2000)
Spectrum of the quantum yield for photochemical mineralization of dis
solved organic carbon in a humic lake Limnol Oceanogr 45 664 -676
Vahatalo A V R G Wetzel and H W Paerl (2005) Light absorption by
phytoplankton and chromophonc dissolved organic matter in the drai
nage basin and estuary of the Neuse River North Carolina (U S A)
Freshwater Biol 50 477 -493
Vannote R L G W Mmshall K W Cummins J R Sedell and C E
Cushing (1980) The river continuum concept Can J Fish Aquat Set
37 130 -137
Velde B (1992) introduction to Clay Minerals Chemistry Origins Uses
and Environmental Significance 1st ed 198 pp CRC Press London
Webster J R and B C Patten (1979) Effects of watershed perturbation
on stream potassium and calcium dynamics Ecol Monogr 49 51 -72
Wetzel R G (2001) Limnology Lake and River Ecosystems 3rd ed
1006 pp Elsevier San Diego Calif
White E M P P Vaughn and R G Zepp (2003) Role of photo Fenton
reaction in the production hydroxyl radicals and photobleaching of co
lored dissolved organic matter in a coastal river of southeastern United
States Aquat Sci 65 402 -414
11 of 12

G03019 RIGGSBEE ET AL PHOTOCHEMICAL SOURCE OF CARBON IN RIVERS G03019
Wolman M G (1967) A cycle of sedimentation and erosion in urban river
channels Geogr Ann 49A 385 -395
Zuo Y and J Hoigne (1993) Evidence of photochemical formation of
H202 and oxidation SO in authentic fog water Science 260 71 -73
M W Doyle Department of Geography University of North Carolina
Campus Box 3220 Chapel Hill NC 27599 3220 USA
D M Leech Institute of Marine Science University of North Carolina
3431 Arendell Street Morehead City NC 28557 USA
C H Orr National Center for Earth Surface Dynamics Saint Anthony
Falls Laboratory University of Minnesota 23rd Ave SE Minneapolis MN
55414 USA
J A Riggsbee and R G Wetzel Department of Environmental Sciences
and Engineering University of North Carolina Chapel Hill NC 27599
7431 USA (anggsbee@restorationsystems com)
12 of 12

WATER RESOURCES RESEARCH VOL 44 W10411 doi 10 1029/2007WR006457 2008
Hem
IVFul
Article
Optical water quality in rivers
J P Julian 12 M W Doyle 1 S M Powers 3 E H Stanley 3 and J A Riggsbee4
Received 20 August 2007 revised l I June 2008 accepted 23 July 2008 published 18 October 2008
[1] Optical water quality (OWQ) governs the quantity and quality of light in aquatic
ecosystems, and thus spatiotemporal changes in OWQ affect many biotic and
abiotic processes Despite the fundamental role of light in rivers studies on nvenne OWQ
have been limited and mostly descriptive Here we provide a comprehensive,
quantitative analysis of the controls and spatiotemporal dynamics of nverme OWQ
focusing on the inherent optical properties (IOPs), which are those that are only affected
by water constituents and not by changes in the solar radiation field First we briefly
review the constituents attenuating light in rivers Second we develop a new method for
partitioning (light) beam attenuation into its constituent fractions This method
distinguishes between absorption and scattering by dissolved and particulate constituents,
and further isolates particulates into mineral and organic components Third we
compare base flow IOPs between four rivers with vastly different physical
characteristics to illustrate intersite variability Fourth, we analyze the spatial and temporal
patterns of IOPs for the four rivers Fifth, we quantify a longitudinal water clarity
budget for one of the rivers Finally, available data are synthesized to identify general
spatial trends robust across broad geographic areas Temporal trends in IOPs were largely
dictated by storm frequency while spatial trends were largely dictated by channel
network configuration Generally, water clarity decreased with increasing discharge
primarily owing to greater scattering by particulates and secondarily to greater absorption
by chromophoric dissolved organic matter Water clarity also generally decreased
longitudinally along the river owing to increased particulate inputs from tributaries,
however, for pear shaped dendritic basins, water clarity reached a mmiinum at —70% of
the channel length and then increased By illustrating the controls and spatiotemporal
variability of nvenne OWQ these findings will be of interest to water resource
managers and fluvial ecologists and specifically for remote sensing of fluvial
environments and river plumes in receiving waters
Citation Julian J P M W Doyle S M Powers E H Stanley and J A Riggsbee (2008) Optical water quality in nvers Water
Resour Res 44 W10411 doi 10 1029/2007WR006457
1 Introduchon
[-)] Optical water quality (OWQ) has been defined as
the extent to which the suitability of water for its func
tional role in the biosphere or the human environment is
determined by its optical properties [Kirk 1988] Accord
mgly OWQ governs the behavior of photons in aquatic
ecosystems and determines underwater light quantity (num
her of photons) and quality (spectral distribution) It there
fore influences primary productivity water temperature
faunal movements photodegradation of organic matter
and numerous other photoassisted biogeochemical reactions
[Wetzel 2001] Changes in OWQ can indicate important
'Department of Geography University of North Carolina Chapel Hill
North Carolina, USA
2Now at Department of Geography University of Oklahoma Norman
Oklahoma USA
3Center for Limnology University of Wisconsin Madison Wisconsin
USA
°Department of Environmental Sciences and Engineering University of
North Carolina Chapel Hill North Carolina USA
Copyright 2008 by the American Geophysical Union
0043 1397/08/2007WR006457$09 00
environmental trends such as eutroph►cat►on sedimentation
or general water quality degradation Additionally OWQ is
a key component of aesthetics recreation and management
of water resources [Davies Colley et al 2003] Thus OWQ
reflects prevailing environmental conditions and dictates
multiple aspects of ecosystem structure and function
[3] The significance of light has long been recognized in
oceans estuanes and lakes but has usually been ignored
and at most dealt with in a descriptive qualitative fashion in
rivers Of the body of work that exists on rivers most
address only individual influences such as light attenuation
by sediments Further the high vanability and difficulty of
characterization of nvenne OWQ [Davies Colley et al
2003] has delayed a comprehensive understanding The
lack of information persists despite the central role ascribed
to light availability in fluvial ecology models such as the
River Continuum Concept (RCC Vannote et al [1980])
Nonetheless the eclectic nature of OWQ and the ease of
field measurement has resulted in its adoption as a water
quality standard in some countries [Davies Colley et al
2003]
[4] The goal of this paper is to provide a comprehensive
overview of the spatial and temporal dynamics of nvenne
W10411 1 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
OWQ focusing on the inherent optical properties (IOPs)
those that are only affected by water constituents and not by
changes in the solar radiation field First the constituents
influencing IOPs are briefly reviewed Second a new
method is developed for partitioning light beam attenuation
into its constituent fractions Third we compare base flow
IOPs between four rivers with vastly different physical
characteristics to illustrate mtersite variability Fourth we
analyze the spatial and temporal distributions of IOPs for
the four rivers Fifth we quantify a water clanty budget for
one of the rivers including tributary inputs Finally avail
able data are synthesized to identify general spatial trends
robust across broad geographic areas Throughout the paper
we place our findings in the context of prevailing fluvial
ecosystem theory
Components of Optical Water Quality
[s] When light enters water it interacts with dissolved
and particulate substances by the processes of absorption
and scattering Although light is mostly lost through ab
sorption scattering can be a mayor influence on the quantity
of light because it increases the probability of absorption
[Kirk 1994] Absorption influences the quality of light by
selectively removing photons with specific wave bands The
absolute quantities of scattering and absorption are
expressed by an absorption coefficient (a) and a scattering
coefficient (b) which respectively are the fraction of radiant
flux (light per time) that is absorbed and scattered by an
infinitesimally thin layer of aquatic, medium per unit dis
tance along the light path Together a and b establish the
(light) beam attenuation coefficient (c) the fraction of
radiant flux that is lost over the infinitesimally thin layer
of aquatic medium in m -i
c= a + b (1)
This value of c is closely and inversely related to the visual
clanty of water [Davies Colley et al 20031 Accordingly c
is low for optically clear rivers and high for turbid rivers
[6] In rivers the amount of light at depth is ultimately
dictated by the diffuse attenuation coefficient (Kd) which
depends weakly on features of the light field such as solar
zenith angle, the ratio of diffuse to direct solar radiation, and
diffuse light within the water column While Kd is more
relevant to primary productivity (via light penetration) c is
more relevant to human aesthetics and faunal movements
(via visibility) Both variables are important to water
resources and aquatic ecology but we focus here on the
inherent optical properties of a b and c so that we can use
additive principles and compare spatiotemporal trends in
OWQ not associated with changes in the solar radiation
field That is we wanted to focus on the rivers composi
tional variability in order to parse out the different controls
on beam attenuation within the water column (e g scat
tenng by particulates versus absorption by dissolved con
stituents) Further Kd and c are strongly correlated
[Davies Colley and Nagels 2008] and thus trends in Kd
generally follow those of c
[7] Any component of the water column can absorb and
scatter light but there are only five that significantly
attenuate light in rivers water (w) chromophonc dissolved
organic matter (CDOM) inorganic suspended sediment
W10411
(SS) nonalgal particulate organic matter (POM) and phy
toplankton ( PHYTO) [Davies Colley et al 2003] Because
beam attenuation is an additive process [Kirk 1994] the
sum of beam attenuation by each one of these components
sets the clarity of a river such that
C = C + CCDOM + CSS + CPOM + CPHYTO (2)
c= C, +cd +cP (3)
where cd is the beam attenuation coefficient of the dissolved
constituents (CDDom) and Cp is the beam attenuation coefficient
of the particulate constituents (css + cpom + cptrt^rO)
[8] While every river possesses a unique OWQ regime
the spatial and temporal trends of the above five com
ponents [Davies Colley et al 2003 Golladay 1997
Reynolds 2000 Sedell and Dahm 1990 Syvitski et al
2000 Walling and Webb 1992 Wetzel 20011 allow for a
few generalizations Temporally rivers have the greatest
clanty during base flow (low Q) and the lowest clanty
during and immediately following floods (high Q) Spa
tially many headwater streams have high clanty owing to
very low CDOM SS POM and PHYTO concentrations
As a river increases in size downstream and source areas
of SS and POM are accessed the river becomes more
turbid and clanty decreases In the lowest reaches of a
river the main stem channel becomes more hydrologically
connected to its floodplam thereby increasing supply of
CDOM to the river The longer residence time of the lower
reaches also allows for a greater abundance of PHYTO
This trend of decreasing clanty along the river continuum
(headwaters to mouth) is an underlying tenet of the RCC
[Vannote et al 1980] but to our knowledge has not been
empirically verified
[9] This brief review provides a basic understanding of
nvenne OWQ but also highlights the fact that comprehen
sive quantitative studies of OWQ are rare and that much of
our current understanding of light driven processes in rivers
is based on assumed knowledge about spatial and temporal
patterns of water constituents To test some of these pre
vailing assumptions we analyze IOPs along the river
continuum in two Midwestern rivers (Baraboo River and
Wisconsin River Wisconsin USA) and compare published
synoptic data sets We also analyze temporal IOPs in a small
Midwestern stream (Big Spring Creek Wisconsin USA)
and a large Southeastern river (Deep River North Carolina
USA)
3 Study Saes
[io] Four nontidal freshwater U S rivers were selected
for our study (Figure 1) We assessed temporal trends in
IOPs on two of the rivers Deep River (DR) and Big Spring
Creek (BSC) The dissimilarities between these two rivers
allowed us to investigate water clanty over a large range of
physical characteristics from a small relatively clear stream
whose hydrology is driven by groundwater (BSC) to a large
relatively turbid river whose hydrology is predominantly
driven by surface runoff (DR) We assessed spatial trends in
IOPs on the Wisconsin River (WR) and Baraboo River
(BR) The dissimilarity in flow regulation between these
two rivers allowed us to investigate water clanty along the
2 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
4648 N
42 57' N
3614 N
35 25 N
8022 W
7863 W
W10411
93 30 W 88 93 W
Figure 1 Optical water quality study sites Major tnbutanes are depicted for all four basins Gage
identifies where Q was measured and water samples were collected Big Spring Creek Basin is adjacent
to but not located in the Wisconsin River Basin
river continuum for a heavily regulated river (WR) and an
unregulated river (BR)
[ii] Deep River is a sixth order stream located in the
Central Piedmont of North Carolina (Figure 1) The
2770 km2 watershed of the DR gage site is predominantly
forest (72 %) followed by agriculture (25 %) and urban
(3 %) Its npanan comdor is composed mostly of mixed
hardwood forest The basin receives 110 cm/a of precipita
tion with no distinct seasonality (NOAA National climate
data 2007 available at http / /www noaa gov /climate html)
Most of the urbanization in the basin is located in the
headwaters which together with its heavily entrenched
channels leads to high flashy flood flows dunng stones
[12] Big Spnng Creek is a second order stream located in
the Central Plain of Wisconsin (Figure 1) Its 21 1 km2
drainage basin is mostly agriculture (46 %) followed by
forest (31%) grassland (21%) and wetland (2 %) The
npanan corridor of BSC is composed of a mixture of reed
canary grass (Phalaris arundinacea) and mixed hardwood
forest Its basin receives 84 cm/a of precipitation with a
seasonal peak in monthly precipitation during the summer
(see http / /www noaa gov /climate html) BSC is a spring
fed stream with relatively constant discharge
[13] Baraboo River is a sixth order stream that begins in
the Western Uplands of Wisconsin and meanders through
the nonglaciated driftless area of central WI before empty
mg into the Wisconsin River (Figure 1) BR drops in
elevation from 420 to 235 m AMSL over a length of
187 km The 1690 km2 Baraboo River Basin is mostly
agriculture (47 %) followed by forest (31 %) grassland
(15 %) wetland (5 %) urban (1%) and barren (1%) Its
npanan corridor is composed mostly of mixed hardwood
forest and various grasses BR histoncally had nine dams on
its main stem (see Wisconsin Department of Natural
3of19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS W10411
Table 1 Temporal Sampling of OWQ at Big Spring Creek and Deep River
Automated samples were collected with Teledyne ISCO 6712 autosamplers Manual samples were collected following the protocol in section 4 1 All
samples were kept dark and refrigerated at —4 C and analyzed within 72 hours of sample collection with only two exceptions (two flood samples for
BSC) Abbreviations are as follows DR Deep River BSC Big Spring Creek
bThe two flood samples for BSC were collected at a station —2 km downstream of the study site A paired t test (n = 44) revealed that c was not
statistically different between these two sites (t = —1 36 p = 0 18)
Resources Dam Database available at http //dnr wi gov/
org /water /wm /dsfm/dams /datacentral html hereinafter re
ferred to as WDNR 2006) All nine dams have been
removed the last one in 2001 and now its entire 187 km
main stem is free flowing
[14] Wisconsin River is a seventh order stream that
begins at Lac Vieux Desert in the Northern Highlands of
Wisconsin and empties into the Mississippi River (Figure 1)
It drops in elevation from 515 to 185 m AMSL over a
length of 684 km The 31 400 km2 Wisconsin River Basin
is mostly forest (41%) followed by agriculture (27 %)
wetland (15 %) grassland (11 %) open water (3 %) urban
(1%) shrubland (1 %) and barren (1 %) Its npanan comdor
is composed of a mosaic of wetlands praine oak savanna
and floodplam forest There are currently 26 main stem
dams on the Wisconsin River (WDNR 2006)
4 Methods
4 1 Sample Collection
[15] We compared short term (3 — 10 d) and long term
(April September) changes in OWQ during base flow and
flood conditions at DR and BSC to assess temporal trends
(Table 1) We assessed longitudinal trends in OWQ by
performing synoptic surveys along the continuum of BR
and WR Water samples were collected during base flow
from 23 main stem locations and 7 tributaries along BR on
13 August 2006 and from 20 main stem locations along WR
on 16 September 2006 All samples were collected in acid
washed amber polyethylene bottles except DOC samples
which were collected in precombusted glass vials treated
with 600 fiL of 2M HCl All filtered water samples
including DOC were obtained using Whatman GF /F
(0 7 µm) glass fiber filters [Wetzel and Likens 2000]
Water chemistry and OWQ analyses were performed within
72 h of sample collection All water samples were kept
dark and refrigerated at —4 °C until analysis which mini
mized any changes in water chemistry during this time
4 2 Hydrology
[16] We obtained 15 min discharge records from the
USGS gages #05405000 and #05407000 for BR and WR
respectively Discharge records for BSC and DR were
obtained from stage Q rating curves we developed using
15 min water level readings from stage recorders (Intech
WT HR 2000 for BSC and HOBO 9 m for DR) and in situ Q
measurements taken with a Marsh McBsrney current meter
4 3 Water Chemistry
[v] We measured total suspended solids (TSS) and its
fractionated components of SS and POM on all water
samples according to APHA Standard Methods procedure
2540D/E [Clesceri et al 1998] using 15 µm glass fiber
filters (Pro Weigh Environmental Express) We measured
DOC as nonpurgeable organic carbon (NPOC) with a
Shimadzu TOC Vcsh Analyzer according to APHA Stan
dard Methods procedure 5310B [Clesceri et al 1998] We
used chi a concentration as a proxy for PHYTO concentra
tion For DR BSC and BR we measured chi a with a
Turner Designs TD 700 fluorometer according to APHA
Standard Methods procedure 10200H [Clesceri et al 1998]
using Whatman GF/F glass fiber filters For WR we
measured chi a with a Beckman DU Series 600 UVNIS
spectrophotometer according to Hauer and Lamberts
[ 1996]
4 4 Optical Measurements
4 4 1 Turbidity (T„)
[18] We measured T„ with a Hach 21OOP turbidimeter in
nephelometnc turbidity units (NTU) which is a relative
measure of b [Kirk 1994] We used the average value of
three T, measurements for each sample thoroughly mixing
the sample prior to each measurement
442 Inherent Optical Properties
[19] We used a Beckman DU Series 600 UVNIS spec
trophotometer to determine the 1OPs of water samples The
spectrophotometer measured the amount of incident radiant
flux (4io) that was received by a light detector ((D) after
being transmitted through a water sample path length (r)
All water samples were contained in the same quartz cuvette
(r = 0 01 m) and analyzed at room temperature (2 IT)
Adopting the method of Brscaud et al [1983] we derived
the beam attenuation coefficient (c) with a Beckman turbid
ity cell holder (TCH) which prevented scattered light from
reaching the light detector by reducing the collection angle
to 0 94° (collimated light beam) and moving the water
sample to 52 mm from the light detector With this config
uration the light detector only captured the incident light
that was left after absorption and scattenng by the water
sample Using the TCH c was calculated as follows
c = — In(`D/,Do)Tc,/r (4)
We derived the absorption coefficient (a) by using a
Beckman standard cell holder (SCH) which placed the
4 of 19
21 -30
May 2006
14-16
June 2006
11-17
July 2006
29 Aug to
11 Sep 2006
24 -26
April 2006
15 -24
June 2006
24 June 2005 to
18 Sep 2006
Location
DR
DR
DR
DR
BSC
BSC
BSC
Method
automated
manual
automated
automated
automated
automated
manual
Flow
base flow
flood
base flow
flood
base flow
base flow
base flow /floodb
Sample interval (hours)
12
—24
6
6
4
6
discrete
Sample number
20
3
25
50
12
36
22/2
Automated samples were collected with Teledyne ISCO 6712 autosamplers Manual samples were collected following the protocol in section 4 1 All
samples were kept dark and refrigerated at —4 C and analyzed within 72 hours of sample collection with only two exceptions (two flood samples for
BSC) Abbreviations are as follows DR Deep River BSC Big Spring Creek
bThe two flood samples for BSC were collected at a station —2 km downstream of the study site A paired t test (n = 44) revealed that c was not
statistically different between these two sites (t = —1 36 p = 0 18)
Resources Dam Database available at http //dnr wi gov/
org /water /wm /dsfm/dams /datacentral html hereinafter re
ferred to as WDNR 2006) All nine dams have been
removed the last one in 2001 and now its entire 187 km
main stem is free flowing
[14] Wisconsin River is a seventh order stream that
begins at Lac Vieux Desert in the Northern Highlands of
Wisconsin and empties into the Mississippi River (Figure 1)
It drops in elevation from 515 to 185 m AMSL over a
length of 684 km The 31 400 km2 Wisconsin River Basin
is mostly forest (41%) followed by agriculture (27 %)
wetland (15 %) grassland (11 %) open water (3 %) urban
(1%) shrubland (1 %) and barren (1 %) Its npanan comdor
is composed of a mosaic of wetlands praine oak savanna
and floodplam forest There are currently 26 main stem
dams on the Wisconsin River (WDNR 2006)
4 Methods
4 1 Sample Collection
[15] We compared short term (3 — 10 d) and long term
(April September) changes in OWQ during base flow and
flood conditions at DR and BSC to assess temporal trends
(Table 1) We assessed longitudinal trends in OWQ by
performing synoptic surveys along the continuum of BR
and WR Water samples were collected during base flow
from 23 main stem locations and 7 tributaries along BR on
13 August 2006 and from 20 main stem locations along WR
on 16 September 2006 All samples were collected in acid
washed amber polyethylene bottles except DOC samples
which were collected in precombusted glass vials treated
with 600 fiL of 2M HCl All filtered water samples
including DOC were obtained using Whatman GF /F
(0 7 µm) glass fiber filters [Wetzel and Likens 2000]
Water chemistry and OWQ analyses were performed within
72 h of sample collection All water samples were kept
dark and refrigerated at —4 °C until analysis which mini
mized any changes in water chemistry during this time
4 2 Hydrology
[16] We obtained 15 min discharge records from the
USGS gages #05405000 and #05407000 for BR and WR
respectively Discharge records for BSC and DR were
obtained from stage Q rating curves we developed using
15 min water level readings from stage recorders (Intech
WT HR 2000 for BSC and HOBO 9 m for DR) and in situ Q
measurements taken with a Marsh McBsrney current meter
4 3 Water Chemistry
[v] We measured total suspended solids (TSS) and its
fractionated components of SS and POM on all water
samples according to APHA Standard Methods procedure
2540D/E [Clesceri et al 1998] using 15 µm glass fiber
filters (Pro Weigh Environmental Express) We measured
DOC as nonpurgeable organic carbon (NPOC) with a
Shimadzu TOC Vcsh Analyzer according to APHA Stan
dard Methods procedure 5310B [Clesceri et al 1998] We
used chi a concentration as a proxy for PHYTO concentra
tion For DR BSC and BR we measured chi a with a
Turner Designs TD 700 fluorometer according to APHA
Standard Methods procedure 10200H [Clesceri et al 1998]
using Whatman GF/F glass fiber filters For WR we
measured chi a with a Beckman DU Series 600 UVNIS
spectrophotometer according to Hauer and Lamberts
[ 1996]
4 4 Optical Measurements
4 4 1 Turbidity (T„)
[18] We measured T„ with a Hach 21OOP turbidimeter in
nephelometnc turbidity units (NTU) which is a relative
measure of b [Kirk 1994] We used the average value of
three T, measurements for each sample thoroughly mixing
the sample prior to each measurement
442 Inherent Optical Properties
[19] We used a Beckman DU Series 600 UVNIS spec
trophotometer to determine the 1OPs of water samples The
spectrophotometer measured the amount of incident radiant
flux (4io) that was received by a light detector ((D) after
being transmitted through a water sample path length (r)
All water samples were contained in the same quartz cuvette
(r = 0 01 m) and analyzed at room temperature (2 IT)
Adopting the method of Brscaud et al [1983] we derived
the beam attenuation coefficient (c) with a Beckman turbid
ity cell holder (TCH) which prevented scattered light from
reaching the light detector by reducing the collection angle
to 0 94° (collimated light beam) and moving the water
sample to 52 mm from the light detector With this config
uration the light detector only captured the incident light
that was left after absorption and scattenng by the water
sample Using the TCH c was calculated as follows
c = — In(`D/,Do)Tc,/r (4)
We derived the absorption coefficient (a) by using a
Beckman standard cell holder (SCH) which placed the
4 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
025-
T
0.20
Q �
015
0
a01oS,
005-
0 00
300
OF 440 TCH OF cd + c,,
SCH -UF ad + aP
TCH F cd
SCH F ad
C1►1-a 1
absorption
peak f(b/a)
1 f(b)
(C ;m index)
350 400 450 600 650 800 650 700 750 SW
Wavelength, A (nm)
W10411
Figure 2 Heuristic diagram of the four configuration spectrophotometer scan The magnitude of
absorbance at 740 nm illustrates the degree of scattenng The magnitude of absorbance at 440 nm of
filtered water samples illustrates the absorption by CDOM The proportional spacing between the top two
absorbance curves illustrates the scattenng to absorption ratio (b /a) The relative height of the peak in the
SCH OF absorbance curve at 675 nm indicates the magnitude of light attenuation by phytoplankton
water sample 10 mm from the light detector and increased
the collection angle to 14° At this close proximity and
comparatively large collection angle we expect on the basis
of published volume scattenng functions [Petzold 19721
most ( -75 %) of the scattered light to be detected thus
absorption can be quantified once a correction for
undetected scattered light is applied Residual scattenng
not captured by the light detector was corrected for by
subtracting out the apparent absorption coefficient at
740 nm (X740) because essentially all measured absorption
at 740 nm is due to scattenng [Davies Colley et al 2003]
Using the SCH a was calculated as follows
X = — ln( 'P/'bo)sCH/r (5)
a =X —X740 (6)
where X is the apparent absorption coefficient for the
measured wavelength Equation (6) assumes that angular
distribution of scattenng is spectrally independent and that
the spectral trend of scattenng is negligible Using equation
(1) we calculated the scattering coefficient (b) by
subtracting a from c as recommended by Davies Colley
et al [2003]
4 4 3 Spectrophotometer Scans
[20] We scanned each water sample in 1 nm intervals
between 340 and 740 nm at 1200 nm/min Each scan took
approximately 20 s thus we assumed that particulate
settling was minimal Each sample was thoroughly mixed
prior to each scan In order to denve the variables in
equations (4) —(6) and partition c (described below) we
performed four configurations of scans on each water
sample (Figure 2) TCH OF (turbidity cell holder unfiltered
sample) TCH F (turbidity cell holder filtered sample)
SCH OF (standard cell holder unfiltered sample) SCH F
(standard cell holder filtered sample) We performed 3
scans for each configuration and used mean values for
subsequent analyses From the spectrophotometer scans
we used readings at 440 rim (index of CDOM) 740 nm
(residual scattenng) and the average of 400 -700 nm
(PAR) Unless denoted by a subscript identifier (e g
a440) all reported attenuation coefficients are average values
for PAR
[21] We also used the spectrophotometer scans to com
pare IOPs between the four study sites and to previous
studies The spectrophotometer scans (Figure 2) illustrate
the change in absorbance (D) with wavelength (A) where
D = logio (41o/`F) (7)
The magnitude of absorbance at 740 nm illustrates the
degree of scattenng in the water column which indicates
the concentration of particulates since scattering by
dissolved constituents is negligible The magnitude of
absorbance at 440 rim of filtered water samples illustrates
the absorption by CDOM The proportional spacing
between the top two absorbance curves (TCH OF and
SCH UF) illustrates the scattenng to absorption ratio (b /a)
which indicates the dominant process of beam attenuation
The magnitude of beam attenuation by PHYTO is indicated
by the relative height of the peak in the SCH OF absorbance
curve at 675 nm [Gallegos and Neale 2002] which is an
absorption peak of chl a
4 5 Partitioning Light Beam Attenuation
[22] We partitioned the beam attenuation coefficient into
its constituent fractions (equation (3)) by using combina
tions of TCH versus SCH and OF versus F (Figure 3)
Using the TCH on an unfiltered sample quantifies the
collective beam attenuation coefficient by the dissolved
(cd) and particulate (cp) constituents Because the spectro
photometer was blanked with Milli Q water pnor to meas
urements we added the beam attenuation coefficient of pure
water (cx,) to the TCH OF reading to obtain the total beam
attenuation coefficient (c) Values for c„, a,,, and b,, were
5of19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
UnBltei ed
Sample
Filtei ed
Sample
M
(UF — IF)
Pme
Ram
Turbidity Standard
Cell Holdei Cell Holder
Figure 3 Partitioning of the beam attenuation coefficient using the four configuration spectro
photometer scan
obtained from the data of Buiteveld et al [1994] Using the
SCH on an unfiltered sample quantifies the collective light
absorption coefficient by the dissolved (ad) and particulate
(ap) constituents We added the absorption coefficient of
pure water (aK,) to the SCH OF reading to obtain the total
light absorption coefficient (a) Using the TCH on a filtered
sample quantifies cd Using the SCH on a filtered sample
quantifies ad We derived particulate attenuation coefficients
by subtracting the dissolved and water attenuation coeffi
cients from the total attenuation coefficients (equation (3))
For example Cp = C — Cd — c. (TCH OF — TCH F
Figure 3) We derived scattering coefficients by subtracting
the absorption coefficients from the beam attenuation coef
ficients (equation (1))
[23] We partitioned cp into css and cpom by using the ap
and by of water samples where TSS was 100% POM When
the particulates in a water sample are composed entirely of
POM by can be attributed entirely to POM (bp = bpom)
Because absorption by SS is usually negligible (Davies
Colley et al [2003] exceptions include iron oxides Babin
and Stramsla [2004]) ap for any water sample can be
attributed entirely to POM (ap = apom) If we define K as
bpomlapoM then under this scenario bpou = Kap which
allows us to approximate the beam attenuation coefficient
of POM
cpom = apoM + bpoM = ap + Kap (8)
This equation assumes that K is a constant for all POM in
the water column It also assumes that cpHyTo is negligible
or either incorporated into cp0m
4 6 Water Clarity Budget
[24] We quantified the effect of tributaries on spatial
OWQ trends by creating a water clarity budget for the
W10411
Baraboo River using the mass balance approach of Davies
Colley et al [2003]
Cd,Qds = C.Qus + Cmb Qmb (9)
where c is the beam attenuation coefficient in m -1 Q is
discharge in m3 /s and the subscripts ds us and tnb denote
downstream upstream and tributary respectively This
method assumes that c is volume conservative where
constituents do not experience physical or chemical changes
(e g sedimentation of SS) between the upstream and
downstream sites To obtain c we used equation (4) on
water samples collected from seven confluences At each
confluence we sampled immediately upstream of the
confluence (c„s) at the tributary outlet before it entered
the main stem channel (C, b) and below the confluence at a
sufficient distance downstream to allow full mixing but
upstream of the next downstream confluences (cd,.) Q was
derived with the weighted area method [Gordon et al
2004] using the downstream USGS gage at river kilometer
(RK) 160 Watershed areas were calculated with the
ArcHydro extension in ArcGIS 9 1 (ESRI) We used high
resolution (1 24 000) national hydrography data (USGS) to
characterize stream link magnitudes (i a stream order via
the Strahler method) [Gordon et al 2004] Mayor
tributanes (in the sense of Benda et al [2004]) were
identified on the basis of a stream order greater than or
equal to n — 1 where n is the stream order of the main stem
channel before the confluence
4 7 Synoptic Optical Water Quality Data Sets
[25] We assessed global longitudinal trends in OWQ by
comparing our two longitudinal c profiles from BR and WR
with published synoptic data sets that met two conditions
(1) OWQ was measured in at least five locations from near
6 of 19

W10411 JULIAN ET AL
OPTICAL WATER QUALITY IN RIVERS
Table 2 Discharge and Water Chemistry of Gage Sites
Deep River at
Big Spring Creek at
Baraboo River at
Wisconsin River at
Glendon
Big Spnngb
La Valleb
Muscodad
Q(m3 /s) 97±144
0291002
96 ±60
1931±777
DOC (mg/L) 6 8 1 14 (65)
12 t 0 3 (94)
2 7± 0 2 (43)
6 9 1 0 6 (8)
SS (mg/L) 20 1 t 55 9 (124)
4 3± 3 0 (64)
60 7 t 24 3 (50)
28 1 t 39 2 (32)
POM (mg/L) 5 2 1 8 4 (124)
2 5± 1 0 (64)
8 9± 2 3 (50)
16 6 1 5 1 (32)
chl a (µg/L) l 5 t 1 1 (21)
6 3± 10 (10)
28 0 t 11 3 (10)
45 4 t 23 3 (7)
Values are mean t standard deviation (number of observations) Q data for all four sites is from water year 2006
bFrom this study
Q from USGS gage 05405000
dFrom Popp [2005] and USGS gage 05407000
the headwaters to the river s mouth and (2) the main stem
channel was greater than 100 km Three data sets
fulfilled these criteria all from New Zealand Motueka
River (110 km Davies Colley [1990]) Pomahaka River
(147 km Harding et al [1999]) and Waikato River
(330 km Davies Colley [1987]) The Waikato River study
reported Secchi disk depth (zsD) which we converted to c
using the method of Gordon and Wouters [1978] (c = 6 1zso)
The Pomahaka and Motueka River studies reported black
disk visibility (YBD) which we converted to c using the
method of Davies Colley [1988] (c = 4 81yBD) We used
these five synoptic OWQ surveys to test the prediction of the
RCC [Yannote et al 1980] that water clarity decreases along
the river continuum from headwaters to mouth
5 Results
5 1 Optical Water Quality Comparisons
[26] Big Spring Creek (BSC) had the highest clarity
because of its low SS POM DOC and PHYTO (Table 2)
These characteristics caused the water of BSC to be
relatively colorless owing to the lack of scattering or
absorption of light BSC had the lowest average base flow
c at 2 73 t 0 89 m -t (meantstd dev) and the lowest
average base flow T, at 4 0 f 13 NTU of the four study
sites (Table 3) Deep River (DR) had a yellowish hue owing
to preferential blue light absorption by its high DOC
concentration The average base flow c and T„ for DR
was 5 78 f l 57 m -i and 5 0± 19 NTU respectively
Wisconsin River (WR) at Muscoda also had a yellowish hue
due its high DOC concentration The c and T for WR at
Muscoda were 15 71 m -t and 13 6 NTU Baraboo River
(BR) at La Valle had the lowest clarity predominantly
because of high SS and POM which imparted a dark
brownish hue on the water This site had the highest c
Table 3 Base Flow Turbidity and IOPs of Study Sites'
Big Spring Wisconsin
Deep River at Creek at Baraboo River at River at
Glendon Big Spring La Valle Muscoda
T (NTU) 5 0± 19 4 0± 13 274 136
C (m ') 5 78 t 1 57 2 73 t 0 89 2926 1571
b/a 125 t 0 29 2 63 t 0 87 660 508
a440 (m -1) 4 10 f 109 0 61 t 0 20 160 236
Study sites are Deep River (n = 74) Big Spring Creek (n = 49)
Baraboo River (n = 1) and Wisconsin River (n = 1) Values are mean t
standard deviation
W10411
and T„ of the four study sites at 29 26 m- i and 27 4 NTU
[27] Spectrophotometer scans of base flow samples illus
trated the relative differences in IOPs among the four study
sites (Figure 4) BR had the highest TCH OF absorbance
curve at 740 nm and thus had the highest total scattering
coefficient (b) at 25 41 m -i followed by WR at 13 13 DR
at 4 39 and BSC at 2 53 We found a strong correlation
between TSS (SS + POM Table 2) and b (r2 = 0 98 p =
0 027) which supports the expected increased scattering
with increased concentration of particulates DR had the
highest SCH F absorbance curve at 340 nm and thus had the
highest CDOM absorption coefficient (a440) at 4 44 m -i
followed by WR at 2 36 BR at 1 60 and BSC at 0 41 DOC
explained 82% of the variation in a440 although the
regression was not statistically significant (p = 0 135) likely
owing to the small sample size (n = 4)
[28] At all four sites scattering was the dominant process
of beam attenuation (b /a > 1) with BR having the highest
b/a at 6 60 followed by WR at 5 08 BSC at 4 10 and DR
at 1 64 The magnitude of beam attenuation by PHYTO
was negligible at DR and BSC because of the lack of a
peak at 675 nm in the SCH OF absorbance curves
(Figure 4) Their low chl a concentrations (Table 2) support
this result BR and WR had small peaks at 675 nm owing
to higher chl a which still contributed minimally to overall
beam attenuation
[29] Turbidity was a highly significant (p < 0 001)
predictor of c at all four sites (Figure 5) The regression
slopes of —I for BR and WR which represent longitudinal
changes in c versus T„ throughout the basin agree well with
other studies that measured both variables across space
[Davies Colley 1987 Davies Colley and Smith 1992]
The much lower regression slopes for DR and BSC which
represent changes in c versus T„ with varying discharge at a
station was likely due to the ratio of 900 scattering to total
scattering (T„ /b) being lower for the larger particles sus
pended by more energetic flows [Davies Colley and Close
1990]
5 2 Temporal Trends Deep River and Big Spring
Creek
5 2 1 Turbidity and Discharge
[30] Turbidity generally increased with increasing Q for
DR and BSC (Figure 6) Q explained 77% of the variation
in T„ at DR (Figure 6b p < 0 001) We attribute this
unexplained variation to hysteresis interstorm and seasonal
effects For example T. values for the storm on 14 June
2006 were lower despite being a larger flood than the
7of19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS W10411
a5
0.20
Q
Q 015
0.10
.c
Q
Dos
A Deep River (DR)
—TQWF
—Sa+uF
TC"
sc14-F
025
c =707mi
020
b!a = 1 64
a44o = 4 44 m i 015
0.00
300 360 400 460 600 360 600 960 700 1'80 800
° C Baraboo River (BR)
c= 2926m1
ba
o.zo
p / = 6 60
C015- a440 = 160 m
010
005
h�
°00
300 350 400 460 600 sm 800 850 700 750 800
Wavelength, A (nm)
010
005
B Big Spring Creek (BSC)
c =3 15m i
b/a =410
a440 =041 m i
000 i iI =r—e i —ri
300 380 400 460 500 s60 800 860 700 760 800
°.n D Wisconsin River (WR)
020 c= 1571m�
015 b/a =508
010
a440 = 2 36 m
000
900 350 400 160 500 560 $00 560 700 750 800
Wavelength, A (nm)
Figure 4 Representative spectrophotometer scans of the four study sites during base flow
storm on 30 August 2006 (Figure 6a) The two likely causes
for this scenano are (1) There was a separate flood on 13
June 2006 that depleted the accumulated source of fine
sediment and POM for the 14 June flood and/or (2) More
sediment and POM were available for the 30 August storm
owing to crop harvesting during this time The relationship
between c and Q at DR (r2 = 0 71 c = 1 43Qi 04) was
similar to the relationship between T„ and Q
[31] Discharge explained only 27% of the vanation in T„
at BSC (Figure 6d p < 0 001) We attribute most of this
unexplained variation to seasonal effects The reduced
vegetative ground coverage of BSC basin during the winter
allowed greater surface sediment runoff especially during
the numerous snowmelt runoff events that occurred in
central Wisconsin during the 2005 -2006 winter This
scenano is the likely cause of the two high T„ measurements
in March 2006 (Figure 6c) The other considerable seasonal
effect on T„ in BSC was the die off of in channel vegetation
dunng late summer BSC had a dense benthic coverage of
aquatic macrophytes, which began to senesce in late July
[Zahn, 2007] This senescence not only added plant frag
ments to the water column but also fine sediment that was
previously trapped by the vegetation This scenano is the
likely cause of the increasing T„ values starting in August of
both years Another contributing factor to increased T„ at
BSC was observed bioturbation with the greatest turbidity
pulses being caused by cows and geese The extremely high
T„ in February 2006 (64 NTU) was most likely caused by
one of these two animals The relationship between c and Q
at BSC (r2 = 0 43 c = 1370 9Q4 85) was similar to the
relationship between T. and Q
5 2 2 Base Flow Beam Attenuation of Big Spring
Creek
[32] The clanty of Big Spring Creek vaned relatively
little during the 10 d base flow period from 15 to 24 June
2006 (Figure 7a) Particulates (c, 81%) accounted for most
of the beam attenuation followed by CDOM (cd 13 %) and
water (cµ, 6 %) The particulates consisted of 47% POM
(2 2 mg/L) and 53% mineral sediment (2 7 mg/L) The
concentration of chi a was relatively low and constant over
the 10 d (6 3 t 10 pg/L) The base flow period of BSC was
characterized by small and brief pulses of SS POM and
CDOM Overall CDOM (a440) remained fairly constant at
0 67 m -i and TSS decreased from 5 4 to 3 4 mg/L The
decrease in TSS was therefore the cause for the decrease in
c over the 10 d period from 3 3 to 2 0 m- i
5 2 3 Base Flow Beam Attenuation of Deep River
[33] The clanty of Deep River increased slightly during
the 10 d base flow period from 21 -30 May 2006
(Figure 7b) During this base flow period cp accounted
for most (64 %) of the beam attenuation, followed by cd
(33 %) and cw (3 %) The particulates consisted of 34% POM
(2 1 mg/L) and 66% mineral sediment (4 1 mg/L) The
concentration of chi a was minimal and relatively constant
over the 10 d (1 2 t 0 1 tig/L) The base flow period of DR
was characterized by decreases in SS (5 6 to 2 9 mg/L) and
CDOM (4 4 to 2 2 m-') resulting in a decrease of c from
6 7 to 3 6 in-' During this time POM % increased at an
average rate of 3 0% per day (20 to 50 %) TSS however
8 of 19

W10411
200
E 175
v
e
150
a�
E 125
V 100
C
2
3 75
m 50
E
e 25
0
0
50
E
40
v
1>: 30
i3
v
c
0
20
�3
>r
10
E
07
JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
5n
A Deep River (DR)
00
0
O O
O
0
40
30
20
ox y =054x +418
0
r2 =097
50 100 150 200 250 300 350
C Baraboo River (BR)
G
® G
0
O
C
10
B Big Spring Creek (BSC)
y =038x +123
02 =099
W10411
0; , i. 1, i I i i
0 20 40 60 BO 100
30
D Wisconsin River (WR)
2s
20
is
10
G y=1 03x-072
5
=096
(+136
-J90
0 Or
0 10 20 30 40 50 0 5 10 15 20 25 30
Turbidity T. (NTU) Turbidity, T. (NM
Figure 5 Beam attenuation coefficient (c) versus turbidity (T „) for the four study sites DR and BSC
illustrate changes in c versus T„ in response to changes in Q at a station BR and WR illustrate
longitudinal changes in c versus T. throughout the basin Note the different x and y axes between the
plots
remained fairly constant at 6 3 mg /L suggesting that
sediment was settling out while additional sources of
POM were being added to the water column During the
other base flow sampling period (11 -17 July 2006 data not
illustrated) POM % increased at an average rate of 4 5%
per day (20 to 47 %) while TSS remained fairly constant at
7 6 mg/L
5 2 4 Flood Beam Attenuation of Deep River
[34] In contrast to the limited change in clarity during
base flow the magnitude and composition of c varied
greatly through a flood at DR on 30 August 2006
(Figure 7c Q,,.k = 60 m3 /s recurrence interval of
—2 months) This flood occurred following a prolonged
(—I month) low flow period (Figure 6a) and thus preflood
water column concentrations of TSS (3 0 mg/L) and CDOM
(2 7 m -1) were relatively low Before the flood c was
3 5 m- i with c, accounting for most beam attenuation
(60 %) followed by cd (35 %) and cW (5 %) Preflood POM
averaged 87% of TSS The value of c increased rapidly
during the rising limb of the flood due mostly to a pulse of
TSS and c reached a maximum of 137 3 m -i at 12 h after
Qpeak This tag was most likely caused by either a delayed
sediment peak resulting from the flood wave s celerity
being faster than the water velocity or an additional TSS
pulse from a tributary with a slower travel time As
particulates settled out of the water column following Qpeak
c decreased exponentially until it reached its average base
flow value of 5 8 m -i at 8 d following Q,,ok CDOM also
increased in response to the flood and maintained elevated
concentrations during the entire sampling period which is
characteristic of terrestrial subsurface flow following a dry
period [Walling and Webb 1992] Consequently the relative
proportion of beam attenuation by CDOM increased fol
lowing the flood reaching a maximum of 53%
5 2 5 Components of Beam Attenuation
[35] Partitioning the total beam attenuation coefficient (c)
by means of equation (3) and Figure 3 revealed that
scattering by particulates (bp) was the dominant process of
midsummer base flow beam attenuation at DR and BSC
(Table 4) Absorption by CDOM (ad) and particulates (a,)
were the two other main contributors to base flow beam
attenuation at both sites For all combined base flow
9 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
Deep River (DR)
180
,.150
Ol
t
y 60
0
30
O 4'M
lama
A
12M lalm 4/x08 via 8208
Big Spring Creek BSC)
to
ti Od
e�
M
04
_A
oa
00
Of105 W31M ila= 3MM saves 8f300O
380
° 900 300
P 240 240
o a.
P
° 180 ISD
o ay
120 12 120
C
80 OD
0 0
tam 0
100 100 7
W10411
80 8D
F
40 40
i�
7
m F M
p 0 4-
01
30 OD 80 120 150 100
Discharge, Q (m3 /s)
D °
Outlier y
P
y = 967 22X417
r -027
P
0
0 0 0
0 8
0
015 03 035 OA 045
Discharge Q (m3 /s)
Figure 6 Discharge (Q time series) and turbidity (T„ point samples) for (a and b) Deep River and (c
and d) Big Spring Creek The four tightly grouped sample intervals in Figure 5a are the four sampling
periods for DR The two tightly grouped sample intervals in Figure 5c are the two sampling periods for
BSC Circled values in Figure 5c are the two flood samples of BSC The outlier (from bioturbation) was
not factored into the regression analysis in Figure 5d Note the different x and y axes between DR and
BSC
sampling at BSC c averaged 2 73 f 0 89 m-1 of which
82% was from TSS (cp) 12% from CDOM (cd) and 6%
from water (c„,) For all combined base flow sampling at
DR c averaged 5 78 f 1 57 m-1 of which 60% was from
TSS (cp) 37% from CDOM (cd) and 3% from water (c„,)
[36] Using water samples where TSS was virtually 100%
POM we found that bpomlapom (or K see Section 4 5 ) for
DR was —3 (3 06 t 0 65 n = 5) There were no water
samples from BSC where TSS was 100% POM and
therefore we used K from DR for BSC This assumes that
POM composition is similar between DR and BSC which
is reasonable on the basis of their riparian zones both being
composed of mixed hardwood forest Assuming that K
equals 3 the beam attenuation coefficient of POM (ep0m)
is approximately 4ap (equation (8)) Using equation (8) and
Table 4 we calculated the amount of base flow beam
attenuation by water CDOM SS and POM at each site
(Figure 8) Beam attenuation by PHYTO was included in
POM but given its low concentrations at both sites
(Table 2) its contribution to beam attenuation was probably
minimal Vahatalo et al [2005] found that aPtiyro for the
Neuse River basin which is adjacent to the Deep River
basin and had slightly higher chl a concentrations than DR
contributed 2 3 f 2 9 % to a During base flow at DR POM
(43 %) was the greatest contributor to beam attenuation,
followed by CDOM (37 %) SS (17 %) and water (3 %)
During base flow at BSC POM and SS both contributed
41% to total beam attenuation followed by CDOM (12 %)
and water (6 %)
5 3 Spatial Trends Baraboo River and Wisconsin
River
5 3 1 Wisconsin River Continuum
[37] Particulate and dissolved concentrations in the water
column fluctuated greatly along the 684 km WR for the first
550 km with sporadic increases and decreases in all four
components (Figure 9a) The large fluctuations in water
chemistry were likely associated with major tributary inputs
and impoundments along this section of river (Figure 9b)
Downstream of the last main stem dam (RK 538) SS POM
and PHYTO steadily increased while CDOM remained
fairly constant SS POM and PHYTO all reached their
maximum values at the last sampling site (RK 674) The
scattering to absorption ratio (b /a) along WR was highly
10 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
E
1
A Big Spring Creek baseflow
•f�tlf
_� el
Cl �
B Deep River baseflow
m
:an.um"
no.w
i —O
e
o m r n w m w= m it
C Deep River flood
n I—
e '
i
e
w = w s m ile
Tb- (h-n)
W10411
is
r
b
n 9
Y
m
0
IN r n r rm w am NO an M NO
I "% A
1
e 71 r n 0 M W W W iM M IN s
Time (bam)
Figure 7 Partitioned beam attenuation at (a) Big Spring Creek during base flow (15 -24 June 2006)
(b) Deep River during base flow (21 -30 May 2006) and (c) Deep River during a flood (29 August to 11
September 2006) Note the different y axes between base flow and flood
irregular ranging from 1 8 (RK 205) to 5 3 (RK 674)
indicating large changes in SS and POM relative to CDOM
[38] The beam attenuation coefficient (c) along WR
followed a similar trend as SS and POM by fluctuating
between 4 1 and 13 8 m-' for the first 548 km and then
steadily increasing after the last main stem dam reaching a
maximum of 22 8 m -' (Figure 9b) There were two local
peaks in c along WR both of which occurred immediately
downstream of confluences with turbid mayor tributaries
Between RK 250 and 292 (Big Rib River confluence at RK
256) c increased from 8 9 to 13 8 m -' Between RK 488
and 524 (Baraboo River confluence at RK 506) c also
increased from 8 9 to 13 8 m- i The c of BR before it
entered WR was 25 2 m -' (Figure 10b)
5 3 2 Baraboo River Continuum
[39] Water chemistry along the 187 km BR (Figure 10a)
fluctuated less than along WR CDOM remained fairly
constant along the entire length of BR SS and POM
increased slightly over the first 28 km and then rapidly
over the next 46 km After RK 74 SS decreased gradually
and POM decreased rapidly The increase in SS and POM at
RK 28 was immediately downstream of the confluence of a
turbid mayor tributary (Cleaver Creek RK 25) PHYTO
along BR was not measured directly and therefore we relied
on the relative peak height at 675 nm in the SCH OF
absorbance curve (Figure 2) to make inferences on its
longitudinal distribution PHYTO was minimal in the head
waters (i a no peak) increased gradually to RK 40 and
then decreased gradually toward the mouth of BR This
decrease in PHYTO at RK 40 coincided with a sharp
increase in c (Figure 10b) indicating less favorable con
ditions for their growth The scattering to absorption ratio
(b /a) increased along BR from 0 8 (RK 3) to 7 8 (RK 142)
before decreasing to 5 9 at the mouth (RK 18 1) The
increase in b/a was associated with increased concentrations
X
of SS and POM while CDOM remained relatively constant
The decrease in b/a over the last 39 km of BR was
associated with decreased concentrations of SS and POM
(Figure 10a) and lower channel gradient (Figure 10b)
which indicates that the particulates were likely settling
out of the water column over this reach
[40] The trend of c along BR was similar to that of SS and
POM (1) increasing gradually over the first 38 km (2)
increasing rapidly over the next 34 km (3) increasing
gradually over the next 70 km and (4) decreasing rapidly
over the last 39 km (Figure l Ob) These trends in c matched
the pattern of mayor confluences along BR where c in
creased rapidly after three mayor confluences and began to
decrease 40 km downstream of the last mayor confluence
5 4 Water Clarity Budget of Baraboo River
[41] We used the synoptic c and Q data through the BR
watershed to develop a water clarity budget (Figure 11) in
which we quantified the relative influence of tnbutary
clarity (CM,b) on main stem clarity (cus equation (9)) All
but two of the tributaries sampled were mayor tributaries
(Kratche Creek and Narrows Creek) and two of the major
tributaries from Figure 10b were not sampled (Cleaver
Creek at RK 25 and Seymour Creek at RK 34) Generally
Crnb and Qo-lb increased in the downstream direction The
value of c,,, increased in the downstream direction for the
first 73 km but then leveled off or decreased The rate of
increase in c,,,. (0 38 m-1/km) over the first 73 km was more
than two times the rate of increase in crab (0 17 m -'/km)
which resulted in a clarity inversion in which Cr,,b was
greater than c,,, in the upper basin but lower than the c..
in lower basin Accordingly the largest increase in c
( +3 22 m -') occurred in the upper basin at the West Branch
Baraboo River confluence while the largest decrease in c
( -4 82 m ') occurred in the lower basin at the Narrows Cr
confluence
11 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS W10411
O� N
h r
a � �
� O O
-H -H
N
�D O_
O V
N —
P O O
-H -H
M S
O O
O O
3
O O
V1 N
h r
M l�
o r
-H -H
o�
M ^'
,= o
N--
C13 C
-H -H
b N
00
O� M
a --
� o 0
-H -H
O M
00 M
—o
00
� o 0
-H -H
N O
r;
M 00
-fi-H
V N
f� N
A U C) O
w
-H -H
ON O
o v �
N O O
C
O r 00
U � 00
* O
00 M
r- r-
e �N
d
F Qm
U
on
a
W
U
0]
a
x
0
3
0
w
c
0
ft""M V9 sprmqCroft
Figure 8 Component contributions to the total beam
attenuation coefficient at Deep River and Big Spring Creek
during base flow
[42] The predicted product of cd,Qd,.* (via equation (9))
and the actual product of cdsQds (via Figure 11) agreed fairly
well (Table 5) All predicted products were within 20% of
the actual product except the two uppermost confluences
These two exceptions may have been caused by the greater
variability in mixing/sedimentation processes in headwater
streams and/or the greater uncertainty of Q for small
watersheds The other five confluences suggest that clarity
in BR is generally volume conservative
6 Discussion
6 1 Water Clarity in Rivers
6 11 Five Components
[43] Water clarity in rivers is dictated by the trends of five
components pure water suspended sediment particulate
organic matter chromophor►c dissolved organic matter and
phytoplankton The optical properties of pure water remain
constant and therefore its contribution to beam attenuation
decreases with increases in any of the other four compo
nents Using a wide variety of rivers we found that water
clarity is primarily dictated by the particulates in the water
column rather than by dissolved constituents Our results are
similar to Davies Colley and Close [1990] who analyzed
96 New Zealand rivers during base flow and found that
87% of the total light beam attenuation was attributed to
particulates
[44] Our study also showed that during floods the
dominance of c, increases (Figure 7c) as SS and POM
increase The relative dominance of SS versus POM is
likely to vary between (Figure 8) and within rivers
(Figure 9a) owing to source limitations For example the
water clarity of rivers in the Midwest USA such as BR that
drain areas with organic rich soils and abundant vegetation
is likely to be dominated by POM whereas the water
clarity of rivers in the Southwest USA such as the Colorado
River that drain areas of organic poor soils and sparse
vegetation is likely to be dominated by SS
[45] The contribution of CDOM to beam attenuation is
also likely to vary between rivers (Figure 8) owing to source
limitations However along the river continuum CDOM
typically remains fairly constant (Figure 10a) [Smith et al
1997] except in rivers with large water contributions from
12 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
0
1= 30
0 25
0
U
�v
r. 20
15
O 10
a
vi
`n 5
0
0
900
800
700
C
..] 600
c/1
d 500
c
0
w 400
a�
300
KIN
A Wisconsin River (WR)
� ss
0 POM
—A- CDOM
—e— PHYTO
0
Er Cr
Ja °
rr IT a
❑ A-
100 200 300 400
500 800
EMI
VIA
250
�o
200 -�
O
150
r
v
100
50
0
700
40
O �
A �
20
n �
t0 o
0
100 200 300 400 500 600 700
Distance from headwaters (km)
W10411
Figure 9 (a) Water chemistry and (b) longitudinal profile and beam attenuation along Wisconsin River
on 16 September 2006 A major confluence is where a tnbutary with a stream order > n -1 enters WR
(n = stream order of WR before confluence)
wetlands [Gallegos 2005] heavily regulated rivers such as
WR (Figure 9a) and heavily disturbed rivers [Davies Colley
1987] The temporal trends in CDOM are mostly influenced
by the hydrologic regime of the river (Figure 7) Because
most of the CDOM present in rivers is derived from displaced
soil moisture [Webster et al 1995 Wetzel 2001] the
contribution of CDOM to beam attenuation is usually greater
following storms and increases as particulates settle out of the
water column (Figure 7c)
[46] We did not quantify cPHYTo but other nvenne OWQ
studies [Davies Colley and Close 1990 Duarte et al
2000 Yahatalo et al 20051 found that the contribution of
PHYTO to light attenuation was either minimal or negligi
ble over a wide range of rivers owing to unfavorable
conditions for phytoplankton growth While particulates
dominate beam attenuation for most rivers there are excep
tions most notably in tidal and blackwater rivers [e g
Gallegos 2005] In these rivers PHYTO and CDOM have
a much greater influence on beam attenuation Future OWQ
research opportunities can be directed toward determining if
trends observed here hold for diverse types of rivers
worldwide
6 12 Optical Water Quality Measurements
[47] Rivenne optical water quality has been measured
using a variety of instruments including a beam transmis
someter [Davies Colley and Smith 1992], Secchi disk
13 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
60
E 50
2
O
O 40
U
c
E
20
CA
Cn 10
0
900
800
700
600
Q
0 5m
i#
W
400
0101
200
A 4 Baraboo River BR)
0
--*--SS
-0 POM Q
—� CDOM
a o �
0
0
0
ao
- -�► - -- ---- — At —Ar- — —♦
0 25 50 75 100 125 150 175
200
40
25 50 75 100 125 150
Distance from headwaters (km)
30
CD
20
� m
n m
10
0
175 200
Figure 10 (a) Water chemistry and (b) longitudinal profile and beam attenuation along Baraboo River
on 13 August 2006 A mayor confluence is where a tributary with a stream order > n -1 enters BR
(n = stream order of BR before confluence) There were no dams along BR
[Davies Colley 1987] black disk [Davies Colley 1990]
and spectrophotometer [Vahatalo et al 2005] While each
method has its advantages and disadvantages [see Davies
Colley et al 2003] we used a spectrophotometer because of
its versatility By using the four configuration spectropho
tometer scan we were able to distinguish between absorption
and scattering of both particulate and dissolved constituents
We were also able to distinguish beam attenuation between
SS and POM using an empirical coefficient (bpomlaPom)
from one of our study sites This scattering to absorption ratio
for POM will need to be further investigated over a wider
range of rivers before this new method becomes practical In
addition to our compartmental method there are other
sophisticated techniques that can be used to distinguish
different components of light attenuation and provide highly
W10411
accurate nverme OWQ measurements [Davies Colley and
Smith 1992 Roesler 1998]
[48] Despite the utility and accuracy of these sophisticated
methods the time detail and cost involved in such analyses
may not make them practical tools for water resource
managers to assess nverme OWQ We therefore recommend
the use of turbidity (T „) as a relative measure of OWQ
Comparisons of T„ and c showed that T„ is a strong
predictor of water clarity (Figure 5) and data from studies
of New Zealand rivers produced similar relationships
[Davies Colley, 1987 Davies Colley and Nagels 2008
Davies Colley and Smith 1992 2001 Smith et al 1997]
The use of T„ as a measure of OWQ over light attenuation
coefficients is advantageous because (1) there is a longer
and more extensive record of T„ in rivers (e g USGS water
14 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
c Q
1 1
1 59 (002)
167 (0 01) Missouri Cr
C I 1 17 (003) (RK 4)
i 2 49 (0 04
Lydon Valley Cr 0 81 (0 04)
(RK 8) 2-91(008
2 91 (0 08)
382 (062)
Kratche Cr 10 78 (0 01)
(RK 28) 362 (063)
W Branch Baraboo R. 14 36 (0 62) ®
(RK 40) 748(l
4 90 (0 07). Spring Valley Cr
328 (0 15) (RK 9)
Im
f�
Cr
O
O
M
1
Legend
C(W)
0 -099
1 -499
®5 -1499
—
15— 29 99
— >30
UPStre=
tributary o1 tnbutary
downstream
01 tntwtary
W10411
Figure 11 Water quality budget for Baraboo River All but two of the tributanes are mayor tnbutanes
(Kratche Creek and Narrows Creek) The distance of the confluence from the headwaters in km (RK) is
given below its name Schematic is not drawn to scale
quality monitoring network) (2) T. is easier and less
expensive to measure and (3) T„ is increasingly becoming
a popular metnc in fluvial ecology studies Disadvantages
of using T„ are (1) it does not use fundamental scientific
units reducible to mass or length (2) its value is instrument
specific owing to different optical designs and (3) it is not
sensitive to changes in absorption [Davies Colley and
Smith 2001] Methods to overcome these limitations are
to use the same instrument for all measurements (e g Hach
2100P) and perform an in situ calibration with a relevant
optical property (e g Figure 5) The use of T. as a relative
Table 5 Predicted Versus Actual Tnbutary Effects on Beam
Attenuation in Baraboo Rivera
RK
29 26 (2 19)
Little Baraboo R
5 76 (0 67) —
(RK 73)
28 59 (2 86)
4
3428 (3 86)
Narrows Cr
16 63 (0 64)-
(RK 115)
2946 (4 50)
4 90 (0 07). Spring Valley Cr
328 (0 15) (RK 9)
Im
f�
Cr
O
O
M
1
Legend
C(W)
0 -099
1 -499
®5 -1499
—
15— 29 99
— >30
UPStre=
tributary o1 tnbutary
downstream
01 tntwtary
W10411
Figure 11 Water quality budget for Baraboo River All but two of the tributanes are mayor tnbutanes
(Kratche Creek and Narrows Creek) The distance of the confluence from the headwaters in km (RK) is
given below its name Schematic is not drawn to scale
quality monitoring network) (2) T. is easier and less
expensive to measure and (3) T„ is increasingly becoming
a popular metnc in fluvial ecology studies Disadvantages
of using T„ are (1) it does not use fundamental scientific
units reducible to mass or length (2) its value is instrument
specific owing to different optical designs and (3) it is not
sensitive to changes in absorption [Davies Colley and
Smith 2001] Methods to overcome these limitations are
to use the same instrument for all measurements (e g Hach
2100P) and perform an in situ calibration with a relevant
optical property (e g Figure 5) The use of T. as a relative
Table 5 Predicted Versus Actual Tnbutary Effects on Beam
Attenuation in Baraboo Rivera
RK
CmbQlnb
c-Q-
Cd Qd
Cd Qd *
Cd Qd *lCd Qds
4
001
003
004
005
138
8
003
010
023
0 13
056
9
033
023
049
057
1 16
28
0 15
234
228
250
1 10
40
890
3 87
1143
12 77
1 12
73
386
64 18
8185
6803
083
115
1059
13244
13259
14303
108
Here cd,Qd * is the predicted product according to equation (9) and
cd Qd is the actual product according to Figure 10
measure of OWQ is probably only valid for nontidal
nonblackwater rivers where scattering is the dominant
process of light attenuation [Davies Colley and Smith
2001 ] In tidal and black-water rivers where absorption is
likely to be the dominant process of light attenuation other
measures such as CDOM or chl a will need to be used
6 2 OWQ Across the Hydrograph
[49] Every study that has compared OWQ to Q including
this study has found that water clarity decreases (c
increases) exponentially with increasing Q (c = aQ °) due
primarily to increased TSS [e g Davies Colley 1987 1990
Smith et al 1997] The rating coefficient (a) and exponent
(3) are river dependent but in general 0 is highest for
rivers with large sources of readily available sediment or
organic matter [Davies Colley 1990 Davies Colley et al
1992] The source of readily available sediment is mflu
enced by basin geology topography land use and storm
frequency [Syvitski et al 2000] Our results suggest that
storm frequency is the dominant control on Q For example
even though the DR basin has more readily available
sediment owing to greater relief and more intensive land
use 0 is higher for BSC (4 85) than DR (1 04) which we
attribute to BSC s much lower storm frequency (Figure 6)
The low storm frequency of its basin allows BSC to remain
clear for most of the year owing to infrequent surface
15 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
100
.-.
v
V
10
^y
V
V
J
rrp
C
0
3
tHu2boo F- Unregulated Mostly Agnculture
ca
E
- -W=oosm R 26 dams Forest and Agncutturc
y
�-w,*c R. 8 dams, Apadture and Urban
m
- tr -YomalWn R Unctguleted 'Mostly CirassLul
-43 - 11otoeln R Unregulated, Mostly Forst
01
0 20 40 60 80 100
Normalized river length (% of total length)
Figure 12 Water clarity along the river continuum Degree of flow regulation and dominant land use is
provided next to the name of each case study
runoff This infrequent surface runoff also allows more time
for sources of SS and POM to accumulate which together
results in a high stormflow c to base flow c ratio (crpcbf)
The higher stone frequency of the DR basin sustains
elevated turbidity at base flow and also prevents large
source accumulations of SS and POM which together
results in a lower cslcbf hence a lower a Temporal
variations in OWQ can also be influenced by seasonal
effects such as exposed soil surface in winter crop harvest
mg and vegetation senescence (basin wide and in channel)
[5o] The change of c with Q is further influenced by the
composition of river water Our results from base flow and
stone sampling at DR showed that POM remained in the
water column longer than SS and thus its relative role in
beam attenuation increased with time following floods We
attribute this temporal trend to POM having a lower settling
velocity than SS Our study also showed that the contnbu
tion of CDOM to c increases following floods as partic
ulates settle out of suspension and terrestrial subsurface
contributions increase (Figure 7) Rivers in which PHYTO
significantly influences water clarity will most likely expe
nence diurnal and seasonal changes in c with Q owing to
the response of PHYTO to sunlight and temperature
[Reynolds 2000] In all the combined effects of storm
frequency seasonality and water chemistry are likely to
produce great temporal variations in OWQ for any given
river
6 3 OWQ Along the River Continuum
[51] We now address the prediction proposed by Yannote
et al [1980] that water clarity decreases (c increases) along
the river continuum Of the five case studies (section 4 7)
Motueka River had the least developed basin with most of
its area being forest and conservation lands [Basher 2003]
Accordingly Motueka River had the lowest c along its
entire length (Figure 12) The Pomahaka basin was also
W10411
relatively undeveloped with most of its land being grass
lands [Harding et al 1999] Accordingly Pomahaka River
had the second lowest c along its entire length The Waikato
River began at the outlet of Lake Taupo and thus was very
clear at its headwaters Intensive agriculture along the
Waikato River caused it to become progressively more
turbid [Davies Colley 1987]
[52] The average c values along the river continuum for
the two U S rivers were an order of magnitude higher than
the NZ rivers (Figure 12) which we attribute to greater
availability of organic rich fine sediments more aggressive
agricultural practices and poorer water quality manage
ment Two of the five rivers had main stem dams Waikato
River (8) and Wisconsin River (26) Reservoirs tend to
reduce SS POM [Grant et al 2003] and CDOM [Larson et
al 2007] and increase PHYTO [Yahatalo et al 2005] and
thus are likely to disrupt spatial trends in c (Figure 9)
Therefore the three unregulated rivers Baraboo Pomahaka
and Motueka provided the best case studies to analyze
water clarity along the river continuum
[53] The water clarity of the three unregulated rivers
followed a similar trend where c increased over the first
70% of the river continuum and then began to decrease
(Figure 12) We suggest that this asymptotic trend as well
as the longitudinal variability of water clarity is dictated by
the channel network configuration (► e density and location
of tributaries Benda et al [2004]) Tributaries are point
sources for all five light attenuating components and
therefore confluences should be sites where changes in
OWQ are most likely to occur Baraboo River (Figures 10b
and 11) provided an excellent example of confluence
effects on water clarity Between RK 20 and 40 three
mayor tributaries entered Baraboo River which coincided
with the greatest increase in c whereas 40 km after the last
mayor tributary c began to decrease For the Pomahaka
River Harding et al [1999] attributed the increase in c to
16 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
turbid inflows from tnbutanes draining agriculturally dom
mated regions Like Baraboo River the decrease in c over
the last 30% of Pomahaka and Motueka Rivers coincided
with the absence of major tributaries This lack of major
tributaries near the outlet of large rivers is consistent with
the Network Dynamics Hypothesis [Benda et al 20041
which states that the distance between geomorphically
significant tributaries increases with distance downstream
owing to the continually reduced drainage area available in
dendritic pear shaped basins The decrease in c along the
last 30% of the three unregulated rivers was most likely the
result of a decreased supply of suspended particulates from
large tributaries and we expect a similar asymptotic trend
for other dendntic pear shaped basins
[54] Wisconsin River provided a counterexample to the
above pattern as five major tributaries enter the channel
over its last 38% and c increased (Figure 9b) These tributary
locations were a consequence of a rectangular shaped basin
providing a relatively constant available drainage area along
the river s continuum (Figure 1) Major tributaries are point
sources of SS and POM which likely caused the increase in
c over the last 30% of Wisconsin River (Figure 12) We
therefore expect the trend of c along the river continuum to
vary for different basin configurations
[55] Our hypothesis of channel network configuration
dictating OWQ shares some of the principles of the Link
Discontinuity Concept (LDC) [Rice et al 2001] The LDC
states that tnbutanes are not Just disruptions to the river
continuum that temporarily reset downstream changes in
physical conditions as proposed by the RCC but rather by
defining patterns of water and sediment flux they are
entirely responsible for moderate and large scale variations
in physical habitat along all river channels [Rice et al
2001 ] Spatial patterns in water clarity are consistent with the
LDC that rivers may be more appropriately viewed as a
series of links where two separate fluxes of water and
sediment meet to form a new channel (equation (9)) In
order to apply the LDC to optical water quality we need to
include CDOM POM and PHYTO fluxes as well Applying
this links concept to OWQ assumes some degree of volume
conservation which we found for Baraboo River (Figure 11
and Table 5) Volume conservation will not always apply due
mainly to sedimentation especially at headwater links [Gomi
et al 2002] but for larger rivers we expect this mass
balance approach to predict downstream water clarity within
20% (Table 5) There are also biochemical transformations
that could affect volume conservation [e g Moreira Turcq et
al 2003] but their effect is probably negligible owing to the
dominance of particulates on beam attenuation
[56] The major limitation of the links concept for OWQ
is that changes in water clarity occur in the absence of
tributaries as well Owing to the increasing contribution of
particulate free groundwater to total Q [Leopold and
Maddock 1953] and the increasing potential of sedimenta
tion and hyporheic exchange processes in the downstream
direction [Packman et al 2000] the absence of major
tributaries typically leads to downstream decreases in c So
even though water clarity is strongly influenced by tributary
inputs the entire basin configuration must be assessed in
order to develop accurate OWQ budgets for rivers
Conclusion
W10411
[57] While light is recognized as a primary limiting
variable in rivers it has received comparatively limited
empirical study Water resource managers should be aware
of spatial and temporal variability of OWQ as it is an
important indicator of water quality change and dictates
aesthetics of water resources of interest to the general
public Ecologically light availability is likely to become
an increasingly important regulator of primary production
and species composition in rivers subject to greater human
land use and nutrient enrichment [Hilton et al 2006]
Anthropogenic land use has also resulted in rivers with
higher turbidity [Walling and Fang 2003] reducing visual
habitat and predation for sighted animals like fish By
knowing the controls and spatiotemporal trends of riverme
OWQ fluvial ecologists will be more able to quantify
aquatic light regimes and understand consequences of light
variability on multiple ecological processes Additionally
remote sensing applications will benefit from OWQ studies
as the optical characteristics of the water column must be
known to derive its depth and composition from radiance
measurements [Jensen 2007]
[58] Most of the referenced literature in this treatise is
derived from studies in New Zealand The reason most
riverme OWQ studies have been performed in New Zealand
is that they have and regulate OWQ standards [see Davies
Colley et al 2003] We advocate broad adoption of similar
OWQ standards to foster ecosystem health and protect the
recreational (and aesthetic) quality of waters Designating
and regulating these standards will require considerable
monitoring From a management perspective our study
suggests that tributaries should be monitored with greater
frequency and extent since they are the point sources for the
components that set OWQ Further we suggest that the
abovementioned trends and concepts particularly the role of
channel network configuration can also be used to under
stand the spatiotemporal trends of other water quality
variables The OWQ of rivers has greater significance
because it affects receiving waters such as lakes estuaries
and coastal environments whose biota greatly depend on
aquatic light availability including coral [Fabncius 2005]
birds [Henkel 2006] and submersed aquatic vegetation
[Dennison et al 19931 This study has highlighted the high
spatiotemporal variability of riverme OWQ and in doing so
has opened up a number of promising research avenues
including the need to understand the effects of land use and
climate change on OWQ as critical steps toward a broader
awareness of the fundamental role of light as a driver of
multiple processes in fluvial ecosystems
Notation
17 of 19
a total absorption coefficient (in-')
ad absorption coefficient of dissolved
constituents (in-')
a, absorption coefficient of particulates (m -i)
a„ absorption coefficient of pure water (in-])
a44o absorption coefficient at 440 nm (in-')
b total scattering coefficient (m -i)
bd scattering coefficient of dissolved
constituents (M-1)
b, scattering coefficient of particulates (in-')

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
bµ,
scattering coefficient of pure water (m -1)
bpoM
scattering coefficient of particulate organic
matter (m- 1)
c
total (light) beam attenuation coefficient (m -1)
cd
beam attenuation coefficient of dissolved
constituents (m -1)
cp
beam attenuation coefficient of
particulates (m -1)
c„,
beam attenuation coefficient of
pure water (m -1)
chl a
chlorophyll a (µg/L)
YBO
black disk visibility distance (m)
zso
Secchi disk visibility depth (m)
AMSL
above mean sea level
BSC
Big Spring Creek
BR
Baraboo River
CDOM
chromophonc dissolved organic matter
(a440 index)
D
absorbance (loglo %/ (P)
DOC
dissolved organic carbon (mg/L)
DR
Deep River
F
filtered water sample
IOP
inherent optical property
K
bpomlapom
OWQ
optical water quality
4D0
incident radiant flux (molls)
(D
transmitted radiant flux ( molls)
PHYTO
phytoplankton (µg/L)
POM
particulate organic matter (mg/L)
Q
discharge or volume of water per time (m3 /s)
Qpeak
maximum discharge during a flood (m3 /s)
Qd,
discharge of river Just downstream of a
tributary (m3 /s)
Qtnb
discharge of tnbutary�ust before confluence
with main stem (m /s)
Qis
discharge of river dust downstream of a
tributary (m3 /s)
SCH
standard cell holder used to obtain a
SS
suspended sediment (mg/L)
T„
turbidity (NTU)
TCH
turbidity cell holder used to obtain c
TSS
total suspended solids (mg/L)
OF
unfiltered water sample
X
apparent absorption coefficient (m -1)
WR
Wisconsin River
[59] Acknowledgments This project was supported by the National
Research Initiative of the USDA Cooperative State Research Education
and Extension Service (CSREES grant 2004 35102 14793) and NSF grant
DEB 0321559 Additional support was provided by U S Fish and Wildlife
Service and Restoration Systems LLC A portion of laboratory resources
were provided by Stephen C Whalen and Robert G Wetzel Special thanks
to Bill Gisler and the town of Big Spring Wisconsin for site access
Rmchelle Baroudm Zack Feiner Victoria Julian Ryan Kromss Mark Lochner
Steve Neary Hollis Rhinelander Matt Smith Leah Vandenbusch Sally
Whmsler and Sarah Zahn assisted with field and laboratory work This
manuscript benefited greatly from reviews by Rob Davies Colley and two
anonymous reviewers
References
Babm M and D Stramskm (2004) Variations in the mass specific
absorption coefficient of mineral particles suspended in water Limnol
Oceanogr 49 756 -767
Basher L R (2003) The Momeka and Rmwaka catchments A technical
report summansig the present state of knowledge of the catchments
management issues and research needs for integrated catchment manage
W10411
ment 122 pp Manaakm Whenua Landcare Res Lincoln Neb
Benda L et al (2004) The network dynamics hypothesis How channel
networks structure reverie habitats BioScience 54 413 -427
dm 10 1641/0006 3568(2004)054[0413 TNDHHC12 0 CO 2
Bncaud A et al (1983) Optical efficiency factors of some phytoplankters
Limnol Oceanogr 18 816 -832
Buiteveld H et al (1994) The optical properties of pure water paper
presented at SPIE Ocean Optics XII Bergen Norway 13 -15 June
Clescen L S A E Greenberg and A D Eaton (1998) Standard Methods
for the Examination of Water and Wastewater 20th ed Am Public
Health Assoc Washington D C
Davies Colley R J (1987) Optical properties of the Waikato River
New Zealand Mitt Geol Palaontol Inst Univ Hamburg SCOPE/UNEP
Sonderband 64 443 -460
Davies Colley R J (1988) Measuring water clarity with a black disk
Limnol Oceanogr 33 616 -623
Davies Colley R J (1990) Frequency distributions of visual water clarity
in 12 New Zealand rivers N Z J Mar Freshwater Res 24 453 -460
Davies Colley R J and M E Close (1990) Water colour and clarity of
New Zealand rivers under baseflow conditions N Z J Mar Freshwater
Res 14 357 -365
Davies Colley R J and J W Nagels (2008) Predicting light penetration
into river waters J Geophys Res 113 G03028 dot 10 1029/
2008JG000722
Davies Colley R J and D G Smith (1992) Offsite measurement of the
visual clarity of waters Water Resour Bull 18 951 -957
Davies Colley R J and D G Smith (2001) Turbidity suspended sedi
ment and water clarity A review J Am Water Resour Assoc 37
1085 -1101 dm 10 1111 /J 1752 1688 2001 tb03624 x
Davies Colley R J et al (1992) Effects of clay discharges on streams 1
Optical properties and epilmthon Hydrobiologia 148 215 -234
dol 10 1007BF00006149
Davies Colley R J et al (2003) Colour and Clarity of Natural
Waters 310 pp Ellis Horwood New York
Dennison W C et al (1993) Assessing water quality with submersed
aquatic vegetation BioScience 43 86 -94 dom 10 2307/1311969
Duarte C M et al (2000) Particulate light absorption and the prediction
of phytoplankton biomass and planktonic metabolism in northeastern
Spanish aquatic ecosystems Can J Fish Aquat Set 57 25 -33
dom 10 1139 /cjfas 57 1 25
Fabncius K E (2005) Effects of terrestrial runoff on the ecology of corals
and coral reefs Review and synthesis Mar Pollut Bull 50 125 -146
dot 10 1016 /j marpolbul 2004 11 028
Gallegos C L (2005) Optical water quality of a blackwater river estuary
The Lower St Johns River Florida USA Estuarine Coastal Shelf Sci
63 57 -72 doi 10 1016 /j ecss 2004 10 010
Gallegos C L and P J Neale (2002) Partitioning spectral absorption in
case 2 waters Discrimination of dissolved and particulate components
Appl Opt 41 4220 -4233 dot 10 1364/AO 41 004220
Golladay S W (1997) Suspended particulate organic matter concentration
and export in streams J N Am Benthol Soc 16 122 -131 dor 10 2307/
1468245
Gomm T et al (2002) Understanding processes and downstream linkages
of headwater systems BuiSctence 51 905 -916 dom 10 1641/0006
3568(2002)052[0905 UPADLO12 0 CO 2
Gordon H R and A W Wouters (1978) Some relationships between
Secchi depth and inherent optical properties of natural waters Appl
Opt 17 3341 -3343
Gordon N D et al (2004) Stream Hydrology An Introduction for
Ecologists 429 pp John Wiley Hoboken N 1
Grant G E et al (2003) A geological framework for interpreting down
stream effects of dams on rivers in A Peculiar Raver Geology Geomor
phology and Hydrology of the Deschutes River Oregon edited by J E
O Connor and G E Grant pp 209 -225 AGU Washington D C
Harding J S et al (1999) Changes in agricultural intensity and river
health along a river continuum Freshwater Biol 42 345 -357
dom 10 1046/! 1365 2427 1999 444470 x
Hauer F R and G A Lamberh (1996) Methods in Stream Ecology
674 pp Elsevier New York
Henkel L A (2006) Effect of water clarity on the distribution of marine
birds in nearshore waters of Monterey Bay California J Field Ornithol
77 151 -156 dom 10 1111 /J 1557 9263 2006 00035 x
Hilton J et al (2006) How green is my river? A new paradigm of
eutrophmcation in rivers Set Total Environ 365 66 -83 dom 10 1016/
J scmtotenv 2006 02 055
Jensen J R (2007) Remote Sensing of the Environment An Earth
Resource Perspective 592 pp Prentice Hall Upper Saddle River N J
18 of 19

W10411 JULIAN ET AL OPTICAL WATER QUALITY IN RIVERS
Kirk J T O (1988) Optical water quality—What does it mean and how
should we measure it? J Water Pollut Control Fed 60 194 -197
Kirk J T O (1994) Light and Photosynthesis in Aquatic Ecosystems
509 pp Cambridge Unry Press New York
Larson 1 H et al (2007) Effects of upstream lakes on dissolved organic
matter in streams Limnol Oceanogr 52 60 -69
Leopold L B and T Maddock Jr (1953) The hydraulic geometry of
stream channels and some physiographic implications U S Geol Sury
Prof Pap 252 1 -57
Morena Turcq P F et al (2003) Chaaactenstics of organic matter in the
mixing zone of the Rio Negro and Rio Solimoes of the Amazon River
Hydrol Process 17 1393 -1404 dot 10 1002/hyp 1291
Packman A 1 et al (2000) A physicochemical model for colloid ex
change between a stream and a sand streambed with bed forms Water
Resour Res 36 2351 -2361 dot 10 1029/2000WR900059
Petzold T J (1972) Volume scattering functions for selected ocean waters
79 pp Visibility Lab Scripps Inst of Oceanogr San Diego Calif
Popp A S (2005) Longitudinal patterns of carbon chlorophyll and nu
tnents in the Wisconsin River M S thesis Unry of Wis Madison Wis
Reynolds C S (2000) Hydroecology of river plankton The role of vana
bility in channel flow Hydrol Process 14 3119 -3132 dot 10 1002/
1099 1085(200011/12)14 16/17 <3119 AID HYP137 >3 0 CO 2 6
Rice S P et al (2001) Tnbutanes sediment sources and the longitudinal
organisation of macromvertebrate fauna along nver systems Can J
Fish Aquat Sci 58 824 -840 dot 10 1139 /clfas 58-4 824
Roesler C S (1998) Theoretical and experimental approaches to improve
the accuracy of particulate absorption coefficients derived from the quan
titative filter technique Limnol Oceanogr 43 1649 -1660
Sedell J R and C N Dahm (1990) Spatial and temporal scales of dis
solved organic carbon in streams and rivers in Organic Acids in Aquatic
Ecosystems edited by E M Perdue and E T GJessing pp 261 -279
John Wiley Hoboken N J
Smith D G et al (1997) Optical characteristics of New Zealand rivers in
relation to flow J Am Water Resour Assoc 33 301 -312 dot 10 1111/
J 1752 1688 1997 tb03511 x
Syvitski J P et al (2000) Estimating fluvial sediment transport The
rating parameters Water Resour Res 36 2747 -2760 dot 10 1029/
2000WR900133
W 10411
Vahatalo A V et al (2005) Light absorption by phytoplankton and chro
mophonc dissolved organic matter in the drainage basin and estuary of
the Neuse River North Carolina (USA) Freshwater Biol 50 477 -493
dm 10 1111 /J 1365 2427 2004 01335 x
Vannote R L et al (1980) The river continuum concept Can J Fish
Aquat Set 37 130 -137
Walling D E and D Fang (2003) Recent trends in the suspended sedi
ment loads of the world s rivers Global Planet Change 39 111-126
dot 10 1016/S0921 8181(03)00020 1
Walling D E and B W Webb (1992) Water quality I Physical char
actenstics in The Rivers Handbook Hydrological and Ecological Prin
ciples edited by P Calow and G E Petits pp 48 -72 Blackwell Set
Malden Mass
Webster J R et al (1995) Organic processes in streams of the eastern
United States in River and Stream Ecosystems edited by C E Cushing
et al pp 117 -187 Elsevier New York
Wetzel R G (2001) Limnologv Lake and River Ecosystems 1006 pp
Elsevier New York
Wetzel R G and G E Likens (2000) Limnologcal Analyses 3 ed
429 pp Springer New York
Zahn S E (2007) Submerged macrophytes in Big Spring Creek WI
Distribution and influence on phosphorous dynamics 29 pp B S
thesis Univ of Wis Madison Wis
M W Doyle Department of Geography University of North Carolina
205 Saunders Hall Chapel Hill NC 27599 3229 USA (mwdoyle@
email unc edu)
J P Julian Department of Geography University of Oklahoma 100
East Boyd Street SEC Suite 684 Norman OK 73019 USA (]Julian@
on edu)
S M Powers and E H Stanley Center for Limnology University of
Wisconsin 680 North Park Street Madison WI 53706 USA (smpowers@
wise edu ehstanley@wisc edu)
J A Riggsbee Department of Environmental Sciences and Engineering
University of North Carolina, 166 Rosenau Hall Chapel Hill NC 27599
USA (anggsbee @restorationsystems com)
19 of 19

-e _1* 1 rwrs�
ENVIRONMENTAL SCIENCE
Aging Infrastructure and
Ecosystem Restoration
Martin W. Doyle,'* Emily H. Stanley,z David G. Havlick? Mark J. Kaiser,' George Steinbach,s
William L. Graf,6 Gerald E. Galloway,r J. Adam Riggsbee8
A us a result of recent infrastructure fail-
res, particularly the tragic failure of
he Interstate -35 bridge in Minnesota,
the U.S. Senate passed the National Infra-
structure Improvement Act (NIIA), which
would create the National Commission on the
Infrastructure of the U.S.A. The commission's
broad mandate would be to assess the nation's
infrastructure and its ability to meet current
and future demands. Such policy develop-
ment coincides with ongoing efforts to man-
age and restore degraded ecosystems. This
provocative intersection of aging infrastruc-
ture and environmental degradation provides
unprecedented and largely unappreciated
opportunities for ecosystem restoration.
Convergence of Phenomena
The United States is at an unusual juncture of
three growing phenomena. First is the wide-
spread decay of infrastructure. The 20th cen-
tury saw rapid growth in population, the econ-
omy, and infrastructure (see chart, right).
Many structures have been in place for 50
years or more, and an increasing portion of
national infrastructure is now approaching or
exceeding its originally intended design life
and will require over $1.6 trillion to reach
acceptable levels of safety and function (1).
The second phenomenon is the degrada-
tion of the environment and the loss of associ-
ated ecosystem services (2). Substantial eco-
logical degradation can be attributed, at least
in part, to infrastructure expansion. Roads
increase sediment erosion, fragment habitat,
and facilitate the spread of invasive species
(3). Dams and levees restrict fish migration
and have drastically altered river flow regimes
(4). Offshore platforms discharge waste,
release atmospheric pollution, and compete
with commercial fishing (5).
'University of North Carolina at Chapel Hill, NC 27599,
USA. 2University of Wisconsin, Madison, WI, USA.
3University of Colorado at Colorado Springs, CO, USA.
°Center for Energy Studies, Louisiana State University,
Baton Rouge, LA, USA. sCalifornia Artificial Reef
Enhancement Program, Santa Barbara, CA, USA. °University
of South Carolina, Columbia, SC, USA. 'University of
Maryland, College Park, MD, USA. gRestoration Systems,
LLC, Raleigh, NC, USA.
*Author for correspondence. E -mail: mwdoyle @email.
unc.edu
Third is the burgeoning of ecosystem
restoration as both a science and an industry
(6). To date, restoration has often been limited
in scale, and its effectiveness is frequently
unclear (7). Nevertheless, there is growing
demand, political will, and funding for restor-
ing degraded ecosystems (7, 8).
Restoration via Decommissioning
Decommissioning can take several forms.
including full removal, partial removal of key
components, or abandonment. Publicly ver-
sus privately owned infrastructure may differ
in decommissioning procedures and ability,
but all are subject to the National Environ-
mental Policy Act and equivalent state laws,
although these laws generally do not facilitate
ecological restoration.
On rivers, dam decommissioning is increas-
ingly common, whereas levee decommission-
ing is rare. Of the >79,000 dams in the United
States, 3500 have been rated as unsafe, collec-
tively in need of $30 billion for rehabilitation,
repair, or removal (1). Levee inventories are
less clear, but estimates exceed 25,000 km (9),
many with unknown structural integrity (10).
To date, >600 dams have been removed, prima-
rily for safety and economic reasons. Dam
removal is followed by rapid recovery of inver-
tebrate, fish, and riparian vegetation communi-
ties (11). Levee decommissioning, often aban-
doning breached levees, reduces economic
demands of levee reconstruction while improv-
ing floodplain habitat and water quality (12).
Of the >6 million km of roads in the United
States, 885,000 km are on public lands main-
tained by federal land agencies, a portion of
which are rarely used (13). The U.S. Forest
Service (USFS), with >250,000 km of roads
over 50 years old, estimated its maintenance
backlog at —$10 billion for 26% of system
roads (13). The USFS decommissioned 7900
Ian of roads between 2002 and 2005 and has
identified almost 300,000 km for possible
decommissioning over the next 40 years (14).
Decommissioning decreases economic liabui-
ties but is also an important tool for restoring
forest ecosystems (13).
In US. federal waters, there are >3900 off-
shore oil and gas platforms, primarily in the
Gulf of Mexico (GOM), about one -third of
Targeted decommissioning of deteriorated
and obsolete infrastructure can provide
opportunities for restoring degraded
ecosystems.
300
0
E 200
s
2 100
C
t
s
L
ao DAMS
60
40
20
0
600
400
200
900
E SEWER PIPE
Y
`0 600
0 300
0
6 SURF_ �A {E D ROADS
Y
0
4
0
2
0
4 OFFSHORE OIL/
3 GAS PLATFORMS
C
2
L
0 i
1900 1920 1940 1960 1980 2000
U.S. infrastructure inventory. For sources, see (21).
286 18 JANUARY 2008 VOL 319 SCIENCE www.sciencemag.org
Published byAAAS

r.
"e POLICYFORUM
which are idle (15) Fed
eral policies require that
platforms be removed with
in 1 year of lease expira
tion and >2700 platforms
have been removed Re
moved platforms have
been pnmanly shallow
water platforms and mostly
disposed of onshore (15)
Costs and environment im
pacts of removing deep
water platforms are sub
stantial
Full platform removal
has drawbacks including
environmental impacts
and loss of the potential
ecological value of the
structure as an artificial
reef (16) Research shows
that platforms facilitate
the expansion of coral populations in the GOM
(17) and act as refuges for juvenile fish
increasing fish production off the coast of
California (18) Rigs to Reefs Programs allow
reuse of decommissioned structures as artificial
reefs Through 2004 >190 retired platforms
were dedicated for fisheries enhancement
which reduced decommissioning costs and led
to >S20 million in industry donations to state
environmental management trust funds (I5)
Department of Defense (DOD) facilities
pose an unusual challenge and opportunity Of
the 257 million ha of federal lands in the United
States >10 million ha belong to the DOD (19)
Access restrictions have made military bases
some of the richest ecological reserves of any
of the nation s public lands (19) Through the
Base Realignment and Closure (BRAG) pro
gram 400 military sites were closed or reclas
sified between 1988 and 2002 To date man
agement of 21 bases on >445 000 ha has been
transferred from the DOD to the U S Fish and
Wildlife Service to become National Wildlife
Refuges (e g Jefferson Proving Ground be
came Big Oaks National Wildlife Refuge)
Perhaps the largest combination of in
frastructure management and ecosystem
restoration is the Comprehensive Everglades
Restoration Project (CERP) an effort to restore
the 2 3 million —ha watershed and its ecosystem
(8) Hydrology in the Everglades is mantpu
lated through hundreds of control gates thou
sands of kilometers of levees 2900 km of
canals and dozens of pump stations which
largely continue to function well The CERP is
based on infrastructure modification and par
tial removal to move the ecosystem to a more
natural and sustainable configuration through a
40 year $20 billion project
Infrastructure decommissioning and removal Removal of Carbonton
Dam resulted in rapid recovery of the federally listed endangered Cape Fear
shiner (Notropls mekistocholos) found in the former impoundment after <2
years of removal The dam was removed to generate environmental restora
tion credits which were then sold to offset stream impacts elsewhere
Policy Directions and Exit Strategies
Infrastructure policy should do more than fund
projects it should set national pnonties and
initiatives The National Commission on
Infrastructure would set such strategic pnon
ties and consider infrastructure financing reha
bilitation and maintenance Rehabilitation
under the NRA includes considering removal
of infrastructure that is deteriorated or no
longer useful When infrastructure has been
decommissioned, ecological restoration has
been a side benefit Prioritizing decommission
ing sites based on a combination of ecological
economic and safety concerns can benefit
multiple stakeholders possibly reducing over
all decommissioning costs (15)
Infrastructure decommissioning is likely
to occur during discrete windows of opportu
nity These may be policy related (e g expim
tion of a dam license) natural (e g flooding)
or through deliberate legislation (e g CERP)
However political will for such expenditures
is difficult to maintain particularly during
political transitions (20) A less broadly
applied funding mechanism is the use of
market like principles in which infrastructure
decommissioning is used to generate credits
to offset environmental impacts elsewhere
(see figure above)
The greatest lesson from current aging
infrastructure is the need for exit strategies
which vary greatly among infrastructure types
Policies for decommissioning dams are sur
pnsmgly rare and vague (11) whereas policies
for decommissioning offshore platforms are
unambiguous (I)-) Because the costs of
decommissioning and cleanup for infrastruc
ture can be substantial more explicit policies
should require provisions for decommissioning
as part of infrastructure license or lease terms
perhaps similar to that for offshore facilities
where bonding requirements are specified and
based on the estimated cost of full removal
Any infrastructure policy approach must
confront the national conundrum of pressing
infrastructure problems and continuing envi
ronmental degradation Specifically aNational
Commission on Infrastructure should squarely
face decommissioning as a viable option and
the environmental benefits gained through
such decommissioning should be assessed as
definable benefits and leveraged when possible
and practicable
References and Notes
1 American Society of Civil Engineers (ASCE) Report Cord
for America s Infrastructure (ASCE Newyork 2005)
2 Heinz Center State of the Nation s Ecosystems
(Cambridge Univ Press Cambridge 2002)
3 National Research Council (NRC) Assessing and
Managing the Ecological Impacts of Paved Roads
(National Academies Press Washington DC 2005)
4 C Nilsson et at Science 308 405 (2005)
5 S Patin Environmental Impact of the Offshore Oil and
Gas Industry (EcoMomtor Publishing East Northport NY
1999)
6 NRC Compensating forWeiland Losses Under the Clean
WoterAct (National Academy Press Washington DC
2001)
7 E S Bernhardt et at Science 308 636 (2005)
8 NRC Progress Toward Restoring the Everglades (National
Academies Press Washington DC 2006)
9 G Tobin WaterResour Bull 31 359 (1995)
10 Interagency Levee Policy Review Committee The
National Levee Challenge (Federal Emergency
Management Association Washington DC 2006)
11 M W Doyle et at Geomorphology 71 227 (2005)
12 D Galat et at BioScience 48 721 (1998)
13 D Havlick No Place Distant (Island Press Washington
DC 2002)
14 D Ihara et at Reinvesting in jobs Communities and
Forests (Center for Environmental Economic
Development Arcata CA 2003)
15 M Kaiser A Pulsipher Ocean Dev Int Law 36 125
(2005)
16 D M Schroeder M 5 Love Ocean Coast Monag 47 21
(2004)
17 P W Sammarco et at Mar Ecol Prog Ser 280 129 (2004)
18 M S Love et at Fish Bull 104 383 (2006)
19 M Leslie et at Conserving Biodiversity on Military Lands
(U S Department of Defense Washington DC 1996)
20 National Council on Public Works Improvement (NCPW0
Fragile Foundation Final Report to the President and
Congress (NCPWI Washington DC 1988)
21 Sources Population (U S Census Bureau 2007)
bridges [U S Federal Highway Administration (FHA)
2007] surfaced roads (FHA Annual Highway Statistics
surfaced roads include soil surfaced slag gravel or
stone asphalt or concrete) dams (ASCE National
Inventory of Dams New York 2007) sanitary sewer
pipe (U S Environmental Protection Agency The Clean
Water and Drinking Water Infrastructure Gap Analysis
2002) and offshore platforms (Minerals Management
Service U 5 Department of the Interior 2007) (instal
lation numbers are only for the U S outer continental
shelf)
22 Funding U S Department of Agnculture (2004 35102
14793) and NSF (DDIG 0521728) Views expressed here
do not represent the views of any supporting organiza
non We thank E 5 Bernhardt L Band M ] Small and
D Carr for comments
www sciencemag org SCIENCE VOL 319 18 JANUARY 2008
Published byAAAS
10 1126/science 1149852
287
00
0
0
N

JOURNAL OF GEOPHYSICAL RESEARCH VOL 113 G03022 doi 10 1029/2007JG000601 2008
Empirical modeling of light availability in rivers
J P Julian 12 M W Doyle 1 and E H Stan1ey3
Received 24 September 2007 revised 27 March 2008 accepted 3 April 2008 published 16 August 2008
[i] While the influence of hydrology and geomorphology on ecosystem limiting factors
in rivers has been well studied particularly habitat availability and nutrient cycling the
more fundamental limitation of light availability has received much less attention
Characterizing light regimes in rivers is optically complex and requires consideration of
five hydrogeomorphic controls topography, riparian vegetation channel geometry optical
water quality, and hydrologic regime To generalize and quantify these hydrogeomorphic
controls we developed an empirical model that predicts both spatial and temporal
variability of photosynthetically active radiation reaching the riverbed (benthic PAR) We
applied this benthic light availability model (BLAM) to two dissimilar systems a large
turbid river and a small optically clear river Comparisons between the two systems
revealed that the dominant control on temporal light availability for the large river was
discharge which accounted for 90% of the variation A dominant temporal control for the
small river did not emerge but instead was found to be a function of both above - canopy PAR
and discharge Spatially water depth accounted for 99% of the variation in benthic
PAR for the large river and riparian shading accounted for 93% of the variation for
the small river Channel orientation also had a mayor influence where an E W
configuration increased benthic PAR by as much as 108% relative to a N -S
configuration BLAM predictions agreed well with measured benthic PAR, within
39% on average over a 9 -d period BLAM is the first model to quantify benthic PAR
using all five hydrogeomorphic controls and thus provides a new tool for
investigating the role of light in fluvial ecosystem dynamics and for establishing light
availability targets in water resource management
Citation Julian J P M W Doyle and E H Stanley (2008) Empirical modeling of light availability in rivers J Geophys Res
113 G03022 doi 10 1029/2007JG000601
1 Introduction
[2] Many fundamental processes of aquatic ecosystems
are driven by light availability including photosynthesis
photochemical reactions thermal fluctuations and animal
behaviors [Wetzel 2001 ] While the influences of hydrology
and geomorphology on other ecosystem limiting factors are
increasingly studied (e g nutrient cycling habitat [Doyle
and Stanley 2006 Strayer et al 2006]) the more funda
mental limitation of light availability has received consid
erably less attention Light studies in rivers may have been
largely neglected because (1) of greater attention to
nutrients in controlling primary production (2) boundary
conditions (banks riparian vegetation) make ambient light
measurements challenging and (3) nvenne optical water
quality is highly variable and difficult to characterize
[Davies Colley et al 2003] The little information that is
'Department of Geography University of North Carolina Chapel Hill
North Carolina USA
2Now at Department of Geography University of Oklahoma Norman
Oklahoma USA
3Center for Limnology University of Wisconsin Madison Wisconsin
USA
Copynght 2008 by the American Geophysical Union
0148 0227/08/2007JG000601
available on nvenne light regimes is derived mostly from
New Zealand rivers under predominantly baseflow condi
tions limiting current understanding of the temporal and
spatial availability of light in rivers
[3] Most of our knowledge on aquatic optics is derived
from studies in oceans [Jerlov 1976 Mobley 1994] and
lakes [Kirk 1994 Wetzel 2001 ] These studies have shown
that once light enters water it is attenuated exponentially
with depth at a predictable rate depending on the type and
quantity of water constituents [Kirk 1994] Light availabil
ity in rivers is optically more complex [Davies Colley et al
2003 Westlake 1966] requiring consideration of channel
hydrology and geomorphology among other factors
[4] Characterizing the light environment in rivers requires
information on the surrounding topography riparian vege
tation channel geometry optical water quality and hydro
logic regime (Figure 1) These components hereafter
referred to as hydrogeomorphic controls are primarily
shaped by the river basin s climate and geology Topogra
phy including mountains canyon walls and riverbanks
affects light availability as an opaque bamer between solar
irradiance and the river Riparian vegetation also shades the
water surface but is not opaque The percentage of light that
riparian vegetation attenuates depends on the direction and
intensity of above canopy irradiance and the canopy struc
G03022 l of 16

G03022 JULIAN ET AL LIGHT IN RIVERS 603022
�A4
a
D04
Ecan
Kd = 1n(E01Eb,d) l y
Figure 1 Light availability in rivers E,Q„ is the total solar irradiance available to the river before any
shading from topography and riparian vegetation ES is the irradiance at the water surface after shading
from topography and riparian vegetation where s is the shading coefficient Eo is the irradiance that
enters the water column after reflection at the air water interface where r is the reflection coefficient Ebel
is the irradiance at the riverbed after attenuation from the water column which is dictated by the optical
water quality (via the diffuse attenuation coefficient Kd) and water depth (y)
ture including its type height density and spatial distnbution
[Song and Band 20041 Channel geometry refers to the three
spatial dimensions of planform, width and depth Planform
and width augment or mitigate terrestrial shading by influ
encing the size of the canopy opening relative to the sunpath
[s] Once light enters the water column the amount
reaching the riverbed (i e benthic light) is influenced by
water depth optical water quality and hydrologic regime
Optical water quality is the biogeochemical property that
dictates the rate of light attenuation with depth and is set by
the relative proportions of pure water chromophoric dis
solved organic matter suspended sediment particulate
organic matter and phytoplankton [Kirk 1994] Optical
water quality can vary widely spatially along a nver
[Davies Colley 1987 Julian et al 2008] and temporally
with discharge [Smith et al 1997 Julian et al 20081
Hydrologic regime (the frequency magnitude timing
duration and variability of streamflow [Poff et al 1997])
directly influences optical water quality and water depth
which in turn dictate the irradiance at depth in a river [Smith
et al 1997 Julian et al 2008]
[6] Most previous riverme light studies have only
assessed the control of optical water quality [Davies Colley
1987 Davies Colley and Close 1990 Davies Colley et al
1992 Koch et al 2004 Phlips et al 2000] The aquatic
controls of optical water quality and hydrologic regime have
been concomitantly addressed by only a few studies
[Davies Colley, 1990 Smith et al 1997] The terrestrial
controls of topography npanan vegetation and channel
geometry have been concomitantly addressed by only a
few studies as well [Davies Colley and Payne 1998
Davies Colley and Quinn 1998] The most comprehensive
nvenne light studies have assessed topography riparian
vegetation channel geometry and optical water quality
[DeNicola et al 1992 Taylor et al 2004] with hydrologic
regime omitted Further all of the above studies have been
site specific A comprehensive explicit and adaptable
framework for characterizing light regimes in rivers has
yet to be developed
[7] The goal of this study was to develop a comprehensive
and empirically based benthic light availability model
(BLAM) Specific objectives were to quantify the amount
of light attenuation by each hydrogeomorphic control derive
a comprehensive expression that incorporates both the
spatial and temporal variability of these controls and apply
this model to rivers with a wide range of physical character
istics First we outline the analytical framework of BLAM
for predicting photosynthetically active radiation (PAR
400 -700 nm) at the riverbed Second we apply BLAM to
two dissimilar systems a large turbid river and a small
optically clear river Third we compare model results of
these two rivers to assess the dominant controls on both
temporal and spatial light availability for nvers in general
Fourth we assess the accuracy of BLAM by comparing
modeled and measured PAR values at a transect in one of our
study reaches Finally we provide examples of applications
2of16

G03022 JULIAN ET AL.: LIGHT IN RIVERS G03022
A. BSC- closed canopy
B. DR- open canopy
Figure 2. Hemispherical canopy photos of transects at (A) Big Spring Creek (BSC) and (B) Deep River
(DR). The orientation of the BSC transect is West -East (azimuth = 90 °) and the orientation of the DR
transect is South -North (azimuth = 0 °). Both transects are forested. The DR transect has an open canopy
because of its greater width, 34 in compared to 6 in for the BSC transect. The dotted white line represents
the sunpath on 27 June for BSC and 27 August for DR.
for BLAM and how readily available or commonly collected
data can be used to assess light regimes at other sites.
2. Methods
2.1. Model Development
[8] To quantify benthic light availability, we combined
previously developed and verified optical and hydrological
methods. The first -order control on light availability is
above - canopy PAR (E a „) in mol m d -', where one
mol equals 6.02 x lo2f photons. E,Q„ is the total PAR (as
irradiance) available to the river before any shading from
topography or riparian vegetation (Figure 1). E,,Q„ can be
obtained directly from a local weather station, measured
directly with a PAR sensor, or modeled using solar simu-
lation software.
[9] Topography and riparian shading decrease the inten-
sity of PAR at the water surface, reducing E,Q„ to Es
(Figure 1). We refer to the ratio of Es:E,Q„ as the shading
coefficient (s), where s decreases with increased shading.
The shading coefficient can be derived from numerous
methods [see Davies - Colley and Payne, 1998; Davies - Colley
and Rutherford, 2005], but we prefer the "canopy photo
method,” where a hemispherical canopy photograph is
overlaid by the sunpath to calculate how much solar radiation
is transmitted through openings in the canopy (Figure 2).
After review of all the methods to quantify stream shade and
several pilot studies, we found that this method provided the
best combination of precision, simplicity, time - efficiency,
versatility, and affordability. Most other methods used to
quantify stream shade (e.g., clinometer, densiometer, and
solar pathfinder) assume an opaque canopy, which can
underestimate transmitted PAR by as much as 85%
[Chazdon and Pearcy, 1991]. The canopy photo method
was designed for forestry applications [Evans and Coombe,
1959], but has been successfully used to quantify stream
shade [Taylor et al., 2004].
[io] Reflection at the air -water interface decreases the
intensity of PAR that enters the water column, reducing Es
to Eo (Figure 1). We refer to the ratio of Eo:Es as the
reflection coefficient (r). The value of r can be found in situ
by measuring PAR immediately above (Es) and below (Eo)
the water surface. Alternatively, r can be estimated using
Fresnel's formula [Kirk, 1994; Mobley, 1994]. The product of
E,a,,, s, and r is the total PAR that enters the water column.
[ii] Once light enters the water column, it is attenuated
exponentially with depth due to scattering and absorption
by dissolved and particulate constituents. The PAR at depth
in the river is thus derived using a simple exponential model
[Kirk, 1994]:
Ed(y) =Eo•e ly (1)
where Ed(y) is downward PAR in µmol m-2 s -' at depth y
in m, and Kd is the diffuse attenuation coefficient for
downward PAR in m -'. Kd is predominantly set by the
optical water quality, and to a lesser degree by the solar
zenith angle and the ratio of diffuse to direct light. Kd can be
normalized to remove the effects of solar zenith angle and
ratio of diffuse to direct light [see Gordon, 1989], but for
most waters dependence of Kd on these two variables is
minimal [Baker and Smith, 1979; Zheng et al., 2002].
[12] Combining equation (1) with the quantifications of
shading and reflection allows calculation of PAR at the
streambed (Ebed) in mol m-2 d -' at one location in time:
Ebed = (ES". • s • r )e —KaY ( 2 )
Spatial variability of Ebed (i.e., longitudinally along the
river) can be derived by adjusting the shading and depth (s
and y). The other parameters of E, ,,,,, r, and Kd do not vary
appreciably along a river reach, defined here as a length of
river with no major confluences and longitudinally
consistent optical water quality.
3of16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
43 67 N
flow Big Spring Creek,
1 WI, USA
Sampling
station
50M
43 66 N
89 65 W
89 64 W
35 55 N
35 51 N
79 35 W 79 31 W
Figure 3 Big Spring Creek and Deep River study reaches Sampling stations are where discharge was
measured and temporal benthic light availability was assessed
[13] In addition to spatial distributions temporal vanabil
ity of Ebed (i a at a station over time) can be quantified We
do this by relating y and Kd to water discharge (Q)
Y = aQ" (3)
Kd = QQ� (4)
where a Q v and as are rating parameters for y and Kd We
used the power function to relate both variables to Q based
on previously developed empincal evidence from Leopold
and Maddock [1953] for water depth and Davies Colley
[1990] for optical water quality The combination of these
two relations modifies equation (2) into a temporally
variable form
Ebed = (Eton s Y)e —"'Q, (5)
Equation (5) therefore predicts the temporal variations in
benthic light availability as a function of discharge
variability while equation (2) predicts spatial vanations in
benthic light availability through a river reach We focus
here on light availability at the channel bed (Ebed) because it
provides a relatively fixed datum is the minimum value of
underwater irradiance and is the relevant quantity for
benthic plant growth This approach however can be used
to predict light availability at any depth in the water column
by simply adjusting y in equation (2)
22 Study Sites
[14] We applied BLAM to two river reaches Big Spring
Creek (BSC) a small relatively clear river whose hydrol
ogy is driven by groundwater and Deep River (DR) a
large relatively turbid river whose hydrology is predomi
nantly influenced by surface runoff The dissimilanties
between these two systems allowed us to (1) investigate
light regimes over a large range of physical characteristics
and (2) display quantitative outputs for a river influenced
more by terrestnal controls (BSC) versus one influenced
more by aquatic controls (DR)
[15] Big Spnng Creek is a 2nd order spnng fed river
located in the Central Plain of Wisconsin near Big Spring
WI (Figure 3) The BSC study reach was a 13 km section
downstream of Big Spring Dam a small run of river dam
Being a run of river dam it did not alter the hydrology of
BSC and compansons between upstream (of the dam) and
downstream stations revealed that downstream optical water
quality was not significantly affected by the dam [Julian
et al 2008] There were no major tributaries and optical
water quality was longitudinally consistent along the entire
study reach (J P Julian unpublished data 2006) Land
cover in its 21 1 km2 watershed was mostly agriculture
(46 %) followed by forest (31%) grassland (21%) and
wetland (2 %) The discontinuous riparian corridor of BSC
was composed of a mixture of reed canary grass (Phalaris
arundinacea) and mixed hardwood forest
[i6] Deep River is a 6th order river located in the Central
Piedmont of North Carolina near Carbonton NC (Figure 3)
The DR study reach was the 5 8 km section downstream of
the former Carbonton Dam which was removed in Decem
ber 2005 There were no major tnbutanes and optical water
quality was longitudinally consistent along the entire study
reach (J P Julian unpublished data 2006) The 2770 km2
watershed was dominated by forest (72 %) followed by
agriculture (25 %) and urban (3 %) land cover Most of the
urbanization in the basin was located in the headwaters
which together with its heavily entrenched channels lead to
high flashy flood flows during storms The nearly contra
uous riparian corridor of DR was composed of mixed
hardwood forest
2 3 Data Collection and Model Inputs
2 3 1 Above - Canopy PAR (Eeo„)
[n] We modeled Eea„ with Gap Light Analyzer (GLA)
software [Frazer et al 1999] using the parameters in
4of16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
Appendix A and the respective locations and elevations of
BSC and DR From GLA we derived an average daily E,u„
for BSC during 15 May to 15 September and an average
daily Ecan for DR during 1 May to 30 September We also
obtained actual daily E,u„ values from the UV B Monitoring
and Research Program [US Department of Agriculture
(USDA) 2007] which reported 3 min averages of 20 s
readings from a LI COR quantum sensor Sites NCO2
(60 km NE of DR) and WI02 (115 km N of BSC) were
used for DR and BSC respectively
2 3 2 Reflection Coefficient (r) and Diffuse
Attenuation Coefficient (Kd)
[ia] We measured r and Kd at various locations and
discharges along the study reaches using a LI COR LI
192 underwater quantum irradiance (PAR) sensor All
measurements were taken at unshaded locations during full
sun conditions between 15 May to 15 September 2006
between 0900 -1500 local standard time and using 15 s
averages We calculated r (EOIE,) by taking PAR measure
ments directly above the water surface (E,) and at 1 cm
below the water surface which we extrapolated to zero
depth (equation (1)) to obtain Eo A total of 27 and 25 r
measurements were taken at BSC and DR respectively In
addition to Eo we measured PAR at the riverbed (Ebed) and
at 10 cm intervals between these two depths We derived Kd
from the linear regression coefficient of In Ed(y) with
respect to y (equation (1)) A total of 34 and 21 Kd
measurements were taken at BSC and DR respectively
2 3 3 Shading Coefficient (s) and Water Depth (y)
[i9] We used synoptic sampling to quantify the within
reach variability of s and y We used a Nikon Coolpix 4500
camera with fisheye lens to collect digital hemispherical
canopy photos along the study reaches which we processed
and analyzed with GLA according to Frazer et al [ 1999] to
obtain ES and s We took 39 canopy photos along the
channel centerline of BSC on 27 June 2006 with an average
distance of 33 in between photos We took 22 canopy
photos along the channel centerline of DR on 27 August
2006 with an average distance of 264 in between photos
Photo locations were selected based on changes in channel
width canopy structure and channel orientation (azimuth)
[-)o] We quantified y along the channel centerline of both
study reaches using longitudinal profiles surveyed with a
total station (Trimble 3350DR) and graded prism rod at
baseflow conditions We measured 129 locations along BSC
on 15 June 2005 with an average interval of 10 in and 67
DR locations on 12 September 2005 with an average
interval of 86 in Survey locations were selected based on
changes in channel slope and water depth
2 3 4 Temporal Sampling
[ ,) i] At each river we established a fixed sampling station
where we quantified depth discharge and turbidity Tern
poral trends in these variables were assessed at a station
0 75 km downstream of the dam on BSC and 0 25 km
downstream of the former dam on DR (Figure 3) Water
depth was measured every 15 min by stage recorders (Intech
WT HR 2000 for BSC and HOBO 9 in for DR) We
calculated discharge (Q) using stage Q rating curves devel
oped with in situ Q measurements taken with a Marsh
McBirney current meter We estimated flood discharges at
DR with the weighted area method [Gordon et al 2004]
using a downstream USGS gage ( #02102000) for a refer
ence Q All reported Q and y are daily average values
[22] The rating parameters a and v were derived from the
regression of y versus Q (equation (3)) and 0 and w were
derived from the regression of Kd versus Q (equation (4))
We used turbidity (T „) as an intermediate regressor (i e Kd
was first regressed with respect to T„ then T„ was regressed
with respect to Q) due to the impracticality of measuring Kd
during high flows Because of the dominance of particulates
on light attenuation in rivers nverine optical water quality
can be characterized fairly accurately using T„ measured in
nephelometnc turbidity units (NTU) with a turbidimeter
[Kirk 1994 Julian et al 2008 Davies Colley and Nagels
2008] We measured T„ with a HACH 2100P turbidimeter
from water samples collected during various flow periods
(Appendix B)
2 4 Data Analysis
2 4 1 Statistical Methods
[23] To assess the dominant controls on benthic light
availability we compared correlations between Ebed and
the parameters of BLAM (equations (2) and (5)) One way
analysis of variance (ANOVA p < 0 05) was used to test for
differences in s among various riparian communities and
channel orientations We classified riparian community as
forest grass or mixed and channel orientation by the four
azimuthal axes 0 -180° 45 -225° 90 -270° and 135-
315° We used JMP IN 5 1 (SAS Institute Cary NC) to
perform all statistical tests
[14] We also used JMP IN 5 1 to perform Monte Carlo
simulations that quantified the frequency of daily Ebed for an
independent randomly selected E,Q„ and an independent
randomly selected Q which are the two temporally variable
parameters in equation (5) We used 10 000 iterations
(paired random samples) for each site selecting from
measured values of E,Q„ (via the weather station) and Q
(via the stage recorder) Distributions were tested for
normality using the Kolmogorov Smirnov test statistic
(D) where D < 0 05 indicated a normal distribution
2 4 2 Effect of Channel Orientation on the Shading
Coefficient
[15] While the effects of channel width and canopy
structure on s are intuitive (i a increased width increases
s increased canopy area and density decreases s) the effect
of channel orientation on s is more complicated and has
rarely been considered in light availability studies We
quantified the variation in s as a function of channel
orientation by keeping width and canopy structure constant
which we accomplished by rotating the canopy photos in
45° increments and reanalyzing in GLA (eight analyses for
each photo) For example by rotating the canopy photo 90°
in Figure 2B we changed its orientation from South North
to West East without altering its width or canopy structure
This technique was performed for four scenarios closed
canopy (forested banks narrow channel) open canopy
(forested banks wide channel) half canopy (one grassed
bank one forested bank) and no canopy (grassed banks)
2 4 3 Model Accuracy Assessment
[26] In order to assess the accuracy of BLAM we
compared modeled daily Ebed (equation (5)) to actual daily
Ebed (via PAR sensor) We measured E,„„ E, Eo and Ebed
5of16

G03022 JULIAN ET AL.: LIGHT IN RIVERS G03022
70
so
s0
�b 40
N
.10
20
10
0
511 011 7H OM 911
2006
70
so
so
40
30
v
20
10
1.4
1.2
1.0
0.8
00
M1
W
0.4
Of
to
1011
UO
120
too
d
so r
A
00
s
40
20
0 1 0
s/1 011 711 W 911 1011
2006
Figure 4. Temporal distribution of daily above - canopy PAR (E,,,), discharge (Q), and benthic PAR
(Ebed) at (A) Big Spring Creek and (B) Deep River. Note the different secondary y axes for Q between the
study sites.
continuously at BSC with four PAR sensors during 16 -25
June 2006. E,Q„ was monitored in 1 -min intervals with a
PAR sensor (HOBO, Onset) placed in a nearby open field.
The other three PAR sensors (LI -192, LI -COR) were set in
an array in BSC at a transect 175 m downstream of the
sampling station. We attached these three sensors to a metal
rod driven into the bed of the channel, with one sensor
located just above the water surface, one immediately below
the water surface, and one on the riverbed to measure E,r, E0,
and Ebed, respectively. The three sensors were connected
to a LI -COR LI -1400 data logger, which recorded PAR in
15 -min intervals. These measurements were integrated
and summed to obtain daily PAR. We leveled all four
sensors with a bubble level and placed a mesh barrier
upstream of the in- channel array to prevent debris from
collecting around the sensors. The Ebed sensor was disturbed
on 21 June, leaving 9 daily Ebed values. The Eo sensor
malfunctioned 18 -22 June, leaving only 5 daily E0 values.
We also monitored daily ES with PAR sensors (HOBO,
Onset) placed at two other transects in BSC: one located at
the sampling station (14 -23 June 2006) and the other 520 in
upstream of the sampling station (25 -26 June 2006).
6of16

G03022 JULIAN ET AL LIGHT IN RIVERS
Table 1 BLAM Input Parameters for Big Spring Creek (BSC) and Deep River (DR)
Temporal Spatial
Parameter BSC DR BSC DR
Ea mol m -z d -i
(504-61 23)
(7 10-6021)
3983
3968
S
0 17
078
(0 15 to 0 94)
(0 52 to 0 81 )
r
092
093
092
093
Y in
aQ
aQ
(023-1 26)
(034-3 55)
n
l 64
0 15
na
na
U
049
067
na
na
Kd m i
/3Q'
13V'
060
1 84
0
3240
0 18
na
na
W
369
1 31
na
na
Q
(030-051)
(3 17-7779)
na
na
Temporal parameters apply to the sampling station only Spatial parameters apply to baseflow only Parentheses
indicate the parameter is variable inside of which is the range of values for the study period Parameters that are not
applicable to the calculation of Ebed are labeled na
2 5 Model Assumptions and Linutations
[17] BLAM (i a equations (2) and (5)) is a one dimen
sional model that assumes the river is well mixed with no
lateral variation in optical water quality This assumption is
not valid for river sections with large dead water zones and
sections directly below confluences [Kenworthy and
Rhoads 1995] BLAM also does not take into account
shading by aquatic biota such as aquatic macrophytes
While we only assessed daily benthic PAR in the center
of the channel our approach can be used to assess light
availability at any wavelength depth lateral distance and
time step
[28] We performed all of our measurements and analyses
when Leaf Area Index (LAI) was greater than 90% of
annual maximum This period of >90% LAI was conserva
lively estimated from previous studies on seasonal leaf
dynamics in the study site s region central North Carolina
(1 May to 30 September [Ptflmroth et al 2005]) and central
Wisconsin (15 May to 15 September [E H Stanley
unpublished data 2005]) By confining our model results
to these periods of >90% LAI we effectively removed
seasonal variations in Ecan and ES and minimized seasonal
6
�b
S
O
G03022
variations in r and Kd BLAM can be used to investigate
seasonal variability in Ebed with additional measurements
but this analysis was beyond the scope of the present study
3 Results
3 1 Controlling Parameters
3 11 Hydrology and Channel Geometry
[29] BSC had a baseflow water surface width of 7 5 t 1 8 m
(mean f standard deviation) and depth of 0 6 t 0 2 m over
the 13 km study reach Its flow was relatively constant
(Figure 4A and Table 1) Average Q was 0 37 t 0 04 m3 /s
and only 4 stormflows with peaks greater than 0 40 m3 /s
(75th percentile) occurred during the study period Water
depth at the BSC sampling station ranged 0 9 -1 2 m and
averaged 1 0 f 0 1 m Spatially y was variable along the
sand bed channel of BSC fluctuating between 0 2 and
1 3 m during baseflow Channel onentation in BSC
changed often as a result of its high sinuosity (Figure 3)
[3o] DR had a much larger channel with a baseflow water
surface width of 35 0 t 4 7 m and depth of 12 f 0 6 m
Discharge at DR was greater and considerably more variable
5 a
0
4
O Id CMp R
° i110 80" Cr
—TMOO
2 o Kd = 017T„
� & r2 =088
° p <001
0
0 S 10 Is 20 as 30 35
Turbtdity, T. MU)
Figure 5 Diffuse attenuation coefficient (Kd) versus turbidity (T„) for Big Spring Creek and Deep
River
7of16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
30
er 25
e
20
Q 15
10
s
0
30
�25
e
20
qW� 1s
a 10
s
oa
0
Big Spring Creek
0 20 40 so BO
Above - canopy PAR, E. (mol m 2 d 1)
C
ee
oe�
e
e °e °'
r2=0116 684
° ee
°
AA A dA � a e, —4
4
30
N 25
e
�20
w� is
10
3
0
0 01 02 03 04 Os 06
Discharge, Q (m%)
Deep River
0 20 40 8o so
Above - canopy PAR. E. (mol m 2 d-t)
30
r25
e
20
W� 1s
10
S
0
0 20 40 80 so
Discharge, Q (m%)
Figure 6 Temporal compansons of benthic PAR (Ebed) with above canopy PAR (E,a„ A C) and
discharge (Q B D) for Big Spnng Creek and Deep River respectively On the basis of equation (5) linear
regression was used for Figures 6A and 6C and exponential regression was used for Figures 6B and 6D
(Figure 4B and Table 1) as average Q was 9 6 f 10 7 m3 /s
During the study penod DR experienced 11 stormflows with
peaks greater than 8 6 m3 /s (75th percentile) At the DR
sampling station y ranged 0 3 -2 9 m and averaged 0 6 t
0 4 m Spatially y was highly variable along the gravel bed
channel of DR ranging 0 3 -3 6 m dunng baseflow The low
sinuosity of DR resulted in relatively few changes in channel
onentation (Figure 3)
3 1 2 Terrestrial Shading
[31] Daily above canopy PAR (E,Q „) at both sites fluc
tuated considerably in response to varying degrees of
cloudiness (Figure 4 and Table 1) Between 15 May and
15 September 2006 E,Q„ at BSC averaged 42 0 f 13 9 mol
m -2 d -1 which agreed closely with the GLA prediction of
39 8 mol m -' d -1 Between 1 May and 30 September
2006 E,Q„ at DR averaged 412 f 12 4 mol m -2 d -1
which also agreed closely with the GLA prediction of
39 7 mol m d -1
[32] The proportion of E,a remaining after terrestnal
shading vaned widely along the BSC study reach due to
changes in the npanan community (Table 1) Spatially
averaged s was 0 51 f 0 25 1 e approximately 51 % of
the available daily PAR passed through the canopy and
reached the water surface over the entire reach s was more
consistent along DR due to a continuous and relatively
uniform riparian comdor (Table 1) and averaged 0 68 f
0 08 The fixed sampling stations at BSC and DR had an s
of 0 17 and 0 78 respectively
3 13 Aquatic Light Attenuation
[33] The proportion of E, remaining after reflection at the
air water interface (r) was relatively constant at both sates
averaging 0 92 t 0 03 at BSC and 0 93 f 0 03 at DR The
water at BSC was much clearer than DR where baseflow
Kd was 060± 0 09 m -1 and 1 84±039 m -i for BSC and
DR respectively The relationship between Kd and T„ for
both nvers was Kd = 0 17T„ (r = 0 88 Figure 5) The
8of16

G03022
025
02
0.14
01
0.05
0
015
02
g 0.15
c�
01
0.05
JULIAN ET AL LIGHT IN RIVERS
0 s 10 1s 20 25 30 36
Beatfuc PAR. Ebw (mol m 2 di)
0 a 10 is 20 25 30 35
Benduc PAR, Ew (md m-2 di)
Figure 7 Magnitude frequency distributions of benthic
light availability (Ebed) at (A) Big Spring Creek and (B)
Deep River Histograms were constructed from Monte Carlo
simulations using 10 000 iterations for each site
relationship between T„ and Q was T„ = 190 57Q3 69 (r2 =
0 54) for BSC and T„ = 1 04Q"' (r2 = 0 85) for DR
Together these relationships produced the rating parameters
between Kd and Q (equation (4) and Table 1)
3 2 Temporal Light Availability
3 2 1 Temporal BLAM Output
[34] Modeled benthic PAR (Eb,,d) at the BSC sampling
station vaned between 0 1 and 5 9 mot m 2 d -1 dunng 15
May to 15 September 2006 (Figure 4A) and average Ebed
dunng this period was 2 8± 1 3 mot m -2 d -1 Generally
Ebed at BSC was highest when E,.,„ was high and Q was low
(Figures 4A 6A and 613) Benthic PAR at the DR sampling
station vaned from 0 0 to 22 3 mot m -2 d -1 dunng 1 May
to 30 September 2006 (Figure 4B) The average Ebed dunng
this period was 8 2 ± 6 0 mot m -2 d -1 and Ebed was
typically highest when Q was low (Figures 4B and 6D)
Although the correlation was statistically significant E,u„
could account for only 11 % of the observed variation in
Ebed at DR (Figure 6C)
3 2 2 Magnitude- Frequency Distribution of Benthic
Light Availabibty
[35] The two temporally variable parameters in BLAM
assuming only summer conditions are E,,,, and Q (Table 1)
G03022
There was no dependence of Q on E a„ (i a no multi
collmeanty) for BSC (p = 0 57) or DR (p = 0 15) which
validated the use of Monte Carlo simulations at both sites
From these simulations and using equation (5) the possible
ran4e of Ebed was 0 -7 mot m -` d -1 for BSC and 0 -33 mot
m -- d -1 for DR (Figure 7) Ebed for BSC was approxi
mately nonmall� distributed (D = 0 04) with a peak at 3-
4 mot m-2 d- (Figure 7A) In contrast Ebed for DR was
nonnormally (D = 0 10) broadly distributed with two
modes one at 0 -1 and the other at 3 -4 mot m -2 d -1
(Figure 713) Most importantly for DR there were many
Ebed values with similar frequencies whereas for BSC
frequencies were dissimilar for the relatively few Ebed values
3 3 Spatial Light Availability
3 3 1 Spatial BLAM Output
[36] Benthic PAR along the 13 km reach of BSC vaned
between 3 2 and 25 1 mot m -2 d -1 during baseflow
(Figure 8A) with a reach average of 12 7 f 6 7 mot m -2
d- Generally Ebed was highest in unshaded sections where
s was high (Figure 9A) There was not a strong correlation
between Ebed and y along BSC (Figure 913) However when
divided into riparian groups correlations between Ebed and
y at BSC were stronger with r' values of 0 25 0 79 and
0 65 for forest mixed and grass respectively
[37] Benthic PAR along the 5 8 km DR study reach
vaned between 0 0 and 14 7 mot m -2 d -1 during baseflow
(Figure 813) with a mean of 4 4± 3 3 mot m -2 d -1 High
Ebed values usually occurred in shallow sections where y
was low (Figure 9D) The correlation between Ebed and s at
DR was relatively weak (Figure 9C) In sum Ebed along
BSC was well predicted by shading but not depth whereas
Ebed at DR was well predicted by depth but not shading
3 3 2 Channel Geometry and Canopy Structure
[38] The two spatially variable parameters in BLAM are s
and y (Table 1) both of which are influenced by channel
geometry Channel depth dictates while channel width and
orientation along with canopy structure dictates Canopy
structure was the major influence on s for BSC because of its
wide variation in riparian community forest (s = 0 26 f 0 10
n = 15) grass (s = 0 80 f 0 07 n = 13) and mixed (0 52 f
0 07 n = 11) These three groups were significantly different
with respect to s (p < 0 01) Width could only explain 21 % of
the variation in s along the entire BSC reach and explained
even less variation within riparian groups (r- = 0 01 0 03
and 0 02 for forest grass and mixed respectively) The
difference in s among the four axes of channel orientation
was only marginally significant (p = 0 06)
[39] Although y was the dominant control on Ebed along
DR s also affected Ebed because of its control on Eo
Compared to BSC DR had a relatively uniform forested
riparian corridor The correlation between s and channel width
was very weak (rz = 0 03) at DR and there was no significant
difference in s among the four axes of orientation (p = 0 79)
which suggests that variation in s here probably resulted from
the sum of independent variations in all three factors
[4o] The effect of channel orientation on s vaned for
different canopy structures For a transect at BSC with a
closed canopy channel orientation did not change s by more
than 0 06 (Figure 10) Similarly for a transect at BSC with
no canopy channel orientation did not change s by more
than 0 02 For a transect at BSC with a half canopy, channel
9 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
Lf
40
Zr,
8 30
I
v
120
a
10
50
A Big Sprmg Creek
r
• s
00 02 04 06 08 10 12 14
Distance from dam (]on)
B Deep River
40-� ---------------------------- E.
%
V
v
20
a
10
0F —'—' -.
0 1
2 3 4 S 6
Dance from dam (!rm)
Figure 8 Longitudinal distribution of above canopy PAR (E,Q„) below canopy PAR (Es) and benthic
PAR (Ebed) during baseflow at (A) Big Spring Creek and (B) Deep River Note the different z axes
between the study sites
onentation changed s by as much as 0 39 For a typical
transect at DR with an open canopy channel orientation
changed s by as much as 0 20 with peaks at 90° and 270°
(Figure 10) In all four canopy scenanos the maximum s
occurred at an azimuth of 900 (West East) Thus given the
same canopy structure and channel width channel onenta
tion has the potential to alter s considerably
3 4 Comparisons Between Modeled and Actual
Benthic PAR
[41] BLAM (equation (5)) consistently over predicted
Ebed but did so within 39% on average for the period of
16 -25 June 2006 (Figure 11 and Appendix C) BLAM
predicted Ebed within 20% on four of the 9 d and the
greatest error was 92% A considerable portion of the error
resulted from the difference in s between the sensor and
modeled values GLA calculated an s of 0 67 at this site
while the sensors (E IE,,„) measured an s of 0 56 t 0 05
Substituting the actual s into equation (5) reduced the
average error of BLAM to 15% A PAR sensor placed at
the BSC sampling station showed similar error in s where
GLA calculated 0 17 and the sensors measured 0 08 f 0 01
(n = 7) However a PAR sensor placed at another transect
showed very little error in s where GLA calculated 0 79 and
the sensors measured 0 78 f 0 01 (n = 2)
[42] Differences in Kd between sensor and modeled values
also added model error Using Q and W from Table 1 BLAM
predicted a Kd of 0 58 f 0 05 m- i for the 9 d penod whereas
the sensors (In Eo In Ebed y i) measured 0 85 f 0 12 m -i
Substituting the actual Kd and s from Appendix C into
equation (5) reduced the average error of BLAM to 7%
(Figure 11) There were relatively minor differences in the
other parameters between modeled and measured values
10 of 16
'
E
2 3 4 S 6
Dance from dam (!rm)
Figure 8 Longitudinal distribution of above canopy PAR (E,Q„) below canopy PAR (Es) and benthic
PAR (Ebed) during baseflow at (A) Big Spring Creek and (B) Deep River Note the different z axes
between the study sites
onentation changed s by as much as 0 39 For a typical
transect at DR with an open canopy channel orientation
changed s by as much as 0 20 with peaks at 90° and 270°
(Figure 10) In all four canopy scenanos the maximum s
occurred at an azimuth of 900 (West East) Thus given the
same canopy structure and channel width channel onenta
tion has the potential to alter s considerably
3 4 Comparisons Between Modeled and Actual
Benthic PAR
[41] BLAM (equation (5)) consistently over predicted
Ebed but did so within 39% on average for the period of
16 -25 June 2006 (Figure 11 and Appendix C) BLAM
predicted Ebed within 20% on four of the 9 d and the
greatest error was 92% A considerable portion of the error
resulted from the difference in s between the sensor and
modeled values GLA calculated an s of 0 67 at this site
while the sensors (E IE,,„) measured an s of 0 56 t 0 05
Substituting the actual s into equation (5) reduced the
average error of BLAM to 15% A PAR sensor placed at
the BSC sampling station showed similar error in s where
GLA calculated 0 17 and the sensors measured 0 08 f 0 01
(n = 7) However a PAR sensor placed at another transect
showed very little error in s where GLA calculated 0 79 and
the sensors measured 0 78 f 0 01 (n = 2)
[42] Differences in Kd between sensor and modeled values
also added model error Using Q and W from Table 1 BLAM
predicted a Kd of 0 58 f 0 05 m- i for the 9 d penod whereas
the sensors (In Eo In Ebed y i) measured 0 85 f 0 12 m -i
Substituting the actual Kd and s from Appendix C into
equation (5) reduced the average error of BLAM to 7%
(Figure 11) There were relatively minor differences in the
other parameters between modeled and measured values
10 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
30
E 25
20
W� 15
aa. 10
c 5
0�
0
30
rl
25
a:
20
v
W 15
a
4. 10
v
5
0
Big Span C� reek
I1
0 0.2 04 06 08
SMdiiig coefficient, s
30
Deep River
0 02 04 06 08 1
Shading coefficient, s
30
25
l:
20
is
p 10
5
0
0 1 2 3 4
Water depth, y (in)
Figure 9 Spatial comparisons of longitudinal benthic PAR (Ebed) with shading coefficient (s A C) and
water depth (y B D) for Big Spring Creek and Deep River respectively Triangles = forest Squares =
mixed and Circles = grass On the basis of equation (2) linear regression was used for Figures 9A
and 9C and exponential regression was used for Figures 9B and 9D
E,Q„ (BLAM 43 0 t 10 6 mol m -2 d- i Sensors 40 7 f
10 9 mol m -` d -1) and r (BLAM 0 92 t 0 03 Sensors
0 88 f 0 04)
4 Discussion
4 1 Controls on Riverme Benthic Light Availability
411 Atmosphere
[43] Atmospheric constituents are the first order controls
on light availability in rivers Their enormous spatiotempo
ral variability and unpredictability [Kirk 1994] prevented us
from modeling E,Q„ as a dependent variable We therefore
used E,,,,, as the independent variable in BLAM While
solar simulation software (GLA) proved to be accurate
within 5% of the average daily E,Q„ the weather station
data were needed to derive actual frequencies of benthic
light availability (Figure 7) The weather station data are
also beneficial when correlations between ecological vana
bles and E,Q, are sought [e g Mulholland et al 200 11
Daily E,a„ is likely to vary considerably in response to
cloud cover (Figure 4) and therefore correlations of this
nature require accurate measurements which can only be
acquired from a local weather station or user installed PAR
sensor
4 1 2 Terrestrial Controls
[44] Before solar irradiance enters the water column
its intensity is reduced by the terrestrial controls of
topography riparian vegetation and channel geometry
Topography was not an effective control on light attenua
tion at either river due to their limited relief Topography is
however capable of being a dominant control on light
availability in mountainous streams canyon rivers and
heavily incised rivers [Davies Colley and Payne 1998
Yard et al 2005]
11 of 16
B
9*
EZa
eb
� 0 0
o
�
a 10
° MD
5
° °o
° ° r2=005
01
0
0
03 06 09 12 15
Water depth, y (in)
Deep River
0 02 04 06 08 1
Shading coefficient, s
30
25
l:
20
is
p 10
5
0
0 1 2 3 4
Water depth, y (in)
Figure 9 Spatial comparisons of longitudinal benthic PAR (Ebed) with shading coefficient (s A C) and
water depth (y B D) for Big Spring Creek and Deep River respectively Triangles = forest Squares =
mixed and Circles = grass On the basis of equation (2) linear regression was used for Figures 9A
and 9C and exponential regression was used for Figures 9B and 9D
E,Q„ (BLAM 43 0 t 10 6 mol m -2 d- i Sensors 40 7 f
10 9 mol m -` d -1) and r (BLAM 0 92 t 0 03 Sensors
0 88 f 0 04)
4 Discussion
4 1 Controls on Riverme Benthic Light Availability
411 Atmosphere
[43] Atmospheric constituents are the first order controls
on light availability in rivers Their enormous spatiotempo
ral variability and unpredictability [Kirk 1994] prevented us
from modeling E,Q„ as a dependent variable We therefore
used E,,,,, as the independent variable in BLAM While
solar simulation software (GLA) proved to be accurate
within 5% of the average daily E,Q„ the weather station
data were needed to derive actual frequencies of benthic
light availability (Figure 7) The weather station data are
also beneficial when correlations between ecological vana
bles and E,Q, are sought [e g Mulholland et al 200 11
Daily E,a„ is likely to vary considerably in response to
cloud cover (Figure 4) and therefore correlations of this
nature require accurate measurements which can only be
acquired from a local weather station or user installed PAR
sensor
4 1 2 Terrestrial Controls
[44] Before solar irradiance enters the water column
its intensity is reduced by the terrestrial controls of
topography riparian vegetation and channel geometry
Topography was not an effective control on light attenua
tion at either river due to their limited relief Topography is
however capable of being a dominant control on light
availability in mountainous streams canyon rivers and
heavily incised rivers [Davies Colley and Payne 1998
Yard et al 2005]
11 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
No Canopy 0WO
oe
b
$ Open Canopy (DR)
0e
0
Half Canopy O3=
02
Gored Canopy (11=
0
0 as 90 05 180 225 270 31S 380
Azimuth ( &-Vva)
Figure 10 Effect of channel orientation (i a compass
direction) on shading coefficients (s) which were derived
from Gap Light Analyzer software and canopy photos from
Big Spring Creek (BSC) and Deep River (DR) Canopy
photos were rotated in 45° increments to obtain the entire
range of channel orientations No Canopy is a grassed
transect at BSC (width = 7 m) Open Canopy is a
forested transect at DR (width = 34 m) Half Canopy is a
transect at BSC with the east bank grassed and the west
bank forested (width = 8 m) Closed Canopy is a forested
transect at BSC (width = 6 m)
[45] Riparian vegetation was a dominant control on Ebed
at BSC because of its relatively narrow channel In forested
sections of BSC riparian vegetation shaded as much as
85% of the incoming PAR In contrast riparian vegetation
accounted for only a 32% reduction of E,Q„ at the wider
DR This trend supports the common expectation that
terrestrial shading decreases with increasing channel width
[Davies Colley and Quinn 1998 Yannote et al 19801
[46] Channel orientation can also mitigate or exaggerate
the effect of terrestrial shading The relative change in s
caused by channel orientation was greatest at DR which
had an open canopy and in sections of BSC with a half
canopy (Figure 10) In river sections with either a closed
canopy or no canopy the orientation of the channel does
not significantly alter s because of the uniform distribution
of canopy gaps relative to the sunpath For river sections
with an open canopy riparian shading is the most exag
gerated (lowest s) by North South orientations because
of the higher opacity of the channel margins and the
smaller window for direct solar radiation transmission (see
Figure 2B for context) Conversely East West orientations
provide a larger window for direct solar radiation trans
mission and orient the sunpath over the upper canopy
which has more gaps than the lower canopy Our finding
that maximum s occurred at an azimuth of 90° for all
riparian vegetation scenarios confirmed this relation The
relationship between channel orientation and topographic
shading follows a similar trend as long as the sun angle is
higher than the local topography however when the sun
angle is lower than the local topography (e g deep
canyons or during winter) an inverse relationship is more
likely where maximum s occurs in North South onenta
tions [Yard et al 2005]
4 13 Aquatic Controls
[47] While the boundary conditions of rivers (terrestrial
controls) create spatial variation in light availability within a
season the aquatic controls of hydrologic regime and
optical water quality create temporal variation In small
spring fed rivers such as BSC this temporal variation may
not be large due to a relatively constant hydrologic regime
and optical water quality Further temporal variation of Ebed
in small rivers is likely to be suppressed by the influence of
terrestrial shading However for most rivers the variation in
benthic light availability is likely to be quite large due to the
variability in Q which dictates the temporal variability in y
[Leopold and Maddock 1953] and Kd [Davies Colley
1990] While the correlation between y and Q was strong
the correlation between Kd and Q was far more variable
This greater variation is largely the result of mterstorm and
seasonal effects on optical water quality [Julian et al
2008]
[48] The use of T„ as an intermediate regressor also added
variation to the correlation However we found a strong and
similar correlation between Kd and T„ at both study sites
(Figure 5), and therefore suggest T„ as a predictor of Kd in
nontidal freshwater rivers with low chromophoric dissolved
organic matter (CDOM) Indeed Davies Colley and Nagels
[2008] found that T„ explained 95% of the variation in Kd
for a wide range of rivers across New Zealand For rivers
with highly colored water or very low turbidity including
the effect of CDOM on Kd would likely improve Kd
estimates [Davies Colley and Nagels 2008] Additionally
their study showed that the relation between Kd and T„ is
more likely to follow a power law trend (exponent —0 5)
when a wider range of Kd values is considered In all
temporal variation of benthic light availability within river
30 T 1 i
25
v
N
E 20
O Oi
15 O
$ O
10 0 wWw tn-m data
C6
wan heiw data
5
0
0 S 10 1s 20 25 30
Actual Ebd (mol m'2 d't)
Figure 11 Predicted daily benthic PAR (Ebed) versus
actual Ebed at BSC for 16 -25 June 2006 Predicted Ebed was
modeled using equation (5) Actual Ebed was measured with
an underwater PAR sensor In situ data include s and Kd
derived from the sensor array (Appendix C)
12 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
E
0
E
rx
a
Stream order
10
Figure 12 Benthic light availability along the river continuum of an idealized 10th order river with a
continuous forested npanan corndor Benthtc PAR (Eked) was derived from equation (2) where EW „ is
above canopy PAR E, is the intensity of PAR at the water surface after terrestrial shading and e —� is
the inverse exponential product of depth (y) and the diffuse attenuation coefficient (Kd) The product
of a —K`-v is a dimensionless proportion with a value of —1 in the headwaters and a value of —0 at the
outlet The influence of aquatic attenuation thus increases in the downstream direction whereas the
influence of terrestrial shading decreases in the downstream direction
reaches is likely to be substantially and predominantly
driven by variability in nver depth and optical water
quality
4 2 Small Versus Large Rivers
[49] Overall DR had less spatial variability (Figure 8)
but greater temporal variability (Figures 4 and 7) in Ebed
than BSC The magnitude frequency distnbution of benthic
light availability in rivers is affected by all the parameters
in equation (5) but it is mostly governed by the temporal
distributions of E,u„ and Q Because of the dominating
influence of s on Ebed for small rivers such as BSC their
temporal variability in Ebed is likely to follow the trend of
E,Q„ In basins with frontal weather patterns this trend is
characterized by an approximately normal distribution in
which most days have an intermediate E un and few days
have very low or very high E,a For large nvers such as
DR Q is likely to be the dominant influence on Ebed
however E,Q„ also affects the temporal distribution of Ebed
because it is the first order control on light availability
Therefore the magnitude frequency distribution of benthic
light in large rivers is likely to have a broad and more
bimodal distribution in which one peak is set by E,,,,, and
the other by Q For example the left peak in Figure 7B
was caused by the high frequency of floods in DR which
lead to elevated turbidity for long periods [Julian et al
2008] This elevated turbidity attenuates most of the
underwater light before it reaches the bed The right peak
in Figure 8B was caused by the distribution of E,,,, which
is similar to that of BSC Overall s sets the maximum
potential Ebed while Q sets the potential range and
frequency of Ebed
[5o] Along the river continuum (from headwaters to
mouth) the influence of shading on Ebed decreases due
to the mitigating effect of width on s Conversely the
effect of y and Kd on Ebed increases with increasing river
size due to the increase in depth and turbidity in the
downstream direction Using equation (2) and assuming a
continuous forested riparian corridor the combined effect
of terrestrial shading and aquatic attenuation produces a
longitudinal distribution of Ebed where it is low in the
headwaters high in the middle reaches and essentially zero
in the higher orders (Figure 12) In general s is the dominant
control on Ebed in small rivers and y is the dominant
control on Ebed in large rivers (Figure 9) The influence
of y on Ebed increases with increasing turbidity These
above relations were developed from reach scale compar
isons and expected longitudinal patterns In order to
venfy the trends in Figure 12 basin scale surveys of
light availability are needed
4 3 Applications of BLAM
4 3 1 Required Data and Accuracy
[51] BLAM incorporates the six mayor controls on light
availability in rivers and allows for both temporal and
spatial vanation in these controls Using our approach the
minimum information needed to charactenze light avail
ability at one location in a river is a canopy photo and
some measure of optical water quality (e g T,) Applying
our method to an entire reach would require measures of
depth and additional canopy photos Temporal character
ization of light availability would require knowledge of the
hydrologic regime and its relationship with y and Kd For
any application of BLAM the extent of data collection
would be determined by the desired precision
[52] Overall BLAM provided fairly accurate estimates
of Ebed (Figure 11) Most of the model error was in s and
Kd We denved the model value of s from GLA using
generalized and average configuration parameters These
parameters are highly variable in both space and time
13 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
[Kirk 1994] and are the primary control on canopy light
transmission [Song and Band 2004] To better characterize
s one would therefore need more spatiotemporally explicit
values of the configuration parameters used in GLA
[53] We derived the model value of Kd from measure
ments taken in unshaded locations during midday and full
sun conditions Variations in the ambient light field are
expected to only minimally affect Kd in most nvers due to
their high scattering to absorption ratios [Zheng et al
2002] however in optically clear rivers such as BSC
increased zenith angles (early morning and late afternoon)
and reduced direct irradiance (cloudy and shaded) are
likely to decrease Kd [Gordon 19891 Our model value
was therefore probably more characteristic of the minimum
Kd than the daily average Kd Obtaining a daily average Kd
for varying levels of cloudiness and streamside shade
would involve greater sampling and more sophisticated
techniques [e g Davies Colley et al 1984] than we used
especially for optically clear rivers
4 3 2 River Ecosystem Dynamics
[54] BLAM can be used to characterize spatial and
temporal trends in river light regimes however its greater
utility is as a tool to investigate river ecosystem dynamics
Light is a first order control on both abiotic (via the
hydrological cycle temperature and photochemical reac
tions) and biotic (via temperature photosynthesis and
visual perception) processes in rivers [Wetzel 2001] Fur
ther it is the only control that exhibits a strong correlation to
gross primary production over a wide range of streams
[Mulholland et al 2001] Yet light budgets are rarely
developed for river ecosystem studies BLAM provides a
fairly simple inexpensive (time and money) and precise
tool for creating these budgets
[55] If we can quantify the amount of solar radiation
entering a river we have a first approximation of one of
the mayor components of ecosystem energy which can then
be used to assess metabolism [see Brown et al 2004] One
of the mayor metabolic processes in rivers is photosynthesis
(or primary production) by algae and submersed aquatic
macrophytes All aquatic plants have a compensation
irradiance which is the amount of PAR required for
photosynthesis to exceed respiration [Kirk 1994] and
these compensation points can be determined relatively
easily for naturally occurring assemblages of macrophytes
or algae Thus by knowing how much PAR reaches the
benthos we can approximate net photosynthesis For
example assuming a compensation irradiance of 3 mol
m -z d-1 for a typical macrophyte population [Kirk 1994
p 278] benthic photosynthesis would occur 46% of the
days during the summer at BSC and 77% of the days
during the summer at DR (Figure 7) Relations such as
these calculated with BLAM can be used to investigate
spatiotemporal trends in riverme vegetation primary pro
ductivity and metabolism Other potential applications of
BLAM include riparian zone management [Kiffney et al
2004] nutrient budgets [Doyle and Stanley 2006] envy
ronmental maintenance flows [Baron et al 2002] stream
restoration [Scarshrook and Halliday 1999] biotic behav
ioral adaptations [Kelly et al 2003] and feedbacks be
tween geomorphology and ecology [Bott et al 2006]
Although the above references establish the ecological
importance of light in rivers the role of light in each of
these areas has largely been underappreciated and not fully
demonstrated
5 Conclusions
[56] Compared to other aquatic ecosystems rivers argu
ably possess the greatest spatiotemporal vanability and
complexity This complexity has up to now prevented the
development of a general framework in which to assess
light regimes in rivers By combining previously verified
optical and hydrological methods we were able to generate
the benthic light availability model (BLAM) which calcu
lates the intensity of PAR at the riverbed BLAM links river
hydrogeomorphology and benthic light availability by in
corporating the light attenuation of topography riparian
vegetation channel geometry optical water quality and
hydrologic regime
[57] The accuracy of BLAM is largely dependent on the
accuracy of the techniques used to obtain s and Kd We
recommend that future studies assess the validity of these
techniques especially for varying degrees of cloudiness and
shading Further we encourage testing on a wide vanety of
rivers thereby improving upon the accuracy and range of
empirical coefficients used in BLAM
[58] We used BLAM to demonstrate how the spatiotem
poral vanations in hydrogeomorphic controls dictate ben
thic light availability in a small optically clear river versus
a large turbid river In addition to assessing the dominant
controls on nvenne light regimes BLAM is a tool that can
be used to investigate the role of light in nver ecosystem
dynamics and establish light availability targets in water
resource management BLAM also provides a framework
for future models that characterize spatiotemporal vana
tions of ultraviolet radiation and water temperature in
rivers Our ultimate objective in developing BLAM is that
it will be a catalyst for more investigations and applications
of the vital role of light in rivers
Appendix A
[59] Table Al
Appendix B
[6o] Table BI
Appendix C
[61] Table Cl
Table Al GLA User Defined Parameters
Parameter
Value
Projection
polar
Orientation
horizontal
Time step
l min
Azimuth regions
36
Zenith regions
9
Solar constant
1367 W/m
Cloudiness Index
050
Spectral traction
045
Beam fraction
050
Sky region brightness
UOC model
Clear sky transmission coefficient
060
14 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
Table B1 Turbidity Sampling at Big Spring Creek (BSC) and Deep River (DR)
21 -30 May 14— 16 Jun 11 -17 Jul
2006 2006 2006
29 Aug to
11 Sep 2006
24 -26 Apr
2006
15 -24 Jun
2006
24 Jun 2005 to
18 Sep 2006
Location DR DR DR
DR
BSC
BSC
BSC
Method automated manual automated
automated
automated
automated
manual
Flow baseflow flood baseflow
flood
baseflow
baseflow
baseflow /flood
Sample interval h 12 —24 6
6
4
6
discrete
Sample number 22 3 23
54
11
33
22/2
Automated samples were collected with a Teledyne ISCO 6712
15 37
033
4774
1757
Table C1 Predicted Versus Actual Benthic PAR in Big Spring Creek WI
6/17/2006
4639
2647
Source
PAR Sensor
PAR Sensor
PAR Sensor
PAR Sensor
Stage Recorder
Weather Station
BLAM
Date
(M/D/Y)
E (mol /m2 /d)
E (mol /m2 /d)
E0 (moUmZ /d)
Eb d (moUmZ /d)
Q (m' /s)
E (moUmZ /d)
Ei d (moUmZ /d)
Eh d /Ei d
6/16/2006
4332
2543
2247
15 37
033
4774
1757
1 14
6/17/2006
4639
2647
22 16
1444
032
4399
1681
l 16
6/1 8/2006
4004
2121
na
97
033
4163
1524
l 57
6/19/2006
4645
2243
na
1034
033
4723
1706
1 65
6/20/2006
3057
1449
na
644
033
3399
1233
1 92
6/22/2006
4031
25 59
na
10 16
034
4981
1695
1 67
6/23/2006
5931
3541
3051
1945
034
60 18
2049
1 05
6/24/2006
3980
23 13
2209
13 71
034
3952
13 35
097
6/25/2006
20 19
1192
10 30
5 65
035
2245
724
128
Actual values were collected with PAR sensors at a transect 175 in downstream of the sampling station during 16 -25 June 2006 Values for 21 June are
not reported because the Eb t sensor was disturbed on that day The shading coefficient (s) for this site as derived by GLA was 0 67 All other temporal
parameters used in BLAM are listed in Table Al Data not available due to equipment malfunction are labeled na Ebed is the predicted benthic PAR
according to equation (5) and Eb t is the actual benthic PAR measured with a PAR sensor
[62] Acknowledgments The project was supported by the National
Research Initiative of the USDA Cooperative State Research Education
and Extension Service (CSREES grant 2004 35102 14793) Special thanks
to Bill Gmsler and the town of Big Spring WI for site access Derek
Anderson Richelle Baroudi Zack Feiner Victoria Julian Rebecca Manners
Caihn Orr Steve Powers Hollis Rhinelander Adam Riggsbee Daisy Small
Matt Smith and Sarah Zahn assisted with field and laboratory work We are
grateful to three anonymous reviewers whose comments helped fine mne the
manuscript
References
Baker K S and R C Smith (1979) Quasi inherent characteristics of the
diffuse attenuation coefficient for irradiance Soc Photo opt Instrum
Eng 208 60 -63
Baron J S et al (2002) Meeting ecological and societal needs for fresh
water Ecol Appl 12 1247 -1260
Bott T L et al (2006) Ecosystem metabolism in Piedmont streams
Reach geomorphology modulates the influence of riparian vegetation
Ecosystems 9 398 -421
Brown J H et al (2004) Toward a metabolic theory of ecology Ecology
85 1771 -1789
Chazdon R L and R W Pearcy (199 1) The importance of sunflecks for
forest understory plants BioScience 41 760 -766
Davies Colley R J (1987) Optical properties of the Waikato River
New Zealand Mitt Geol Paleontol Inst Univ Hamb SCOPE /UNEP
Sonderband 64 443 -460
Davies Colley R J (1990) Frequency distributions of visual water clanty
in 12 New Zealand rivers N Z J Mar Freshwater Res 24 453 -460
Davies Colley R J and M E Close (1990) Water colour and clanty of
New Zealand rivers under baseflow conditions N Z J Mar Freshwater
Res 24 357 -365
Davies Colley R J and J W Nagels (2008) Predicting light penetration
into river waters J Geophjs Res doi 10 1029/2008JG000722 in press
Davies Colley R J and G W Payne (1998) Measuring stream shade
J North Am Benthol Soc 17 250 -260
Davies Colley R J and J M Quinn (1998) Stream lighting in five
regions of North Island New Zealand Control by channel size and
riparian vegetation N Z J Mar Freshwater Res 32 591 -605
Davies Colley R J and J C Rutherford (2005) Some approaches for
measuring and modelling riparian shade Ecol Eng 24 525 -530
Davies Colley R J et al (1984) Optical charactensation of natural waters
by PAR measurement under changeable light conditions N Z J Mar
Freshwater Res 18 455 -460
Davies Colley R J et al (1992) Effects of clay discharges on streams
part I Optical properties and epilithon Hydrobiologia 248 215 -234
Davies Colley R J et al (2003) Colour and Clarity of Natural Waters
310 pp Ellis Horwood New York
DeNicola D M et al (1992) Influences of canopy cover on spectral
irradiance and penphyton assemblages in a prairie stream J North Am
Benthol Sac 11 391 -404
Doyle M W and E H Stanley (2006) Exploring potential spatial
temporal links between fluvial geomorphology and nutrient penphyton
dynamics in streams using simulation models Ann Assoc Am Geogr
96 687 -698
Evans G C and D E Coombe (1959) Hemispherical and woodland
canopy photography and the light climate J Ecol 47 103 -113
Frazer G W et al (1999) Gap Light Analyzer (GLA) version 2 0 Ima
grog software to extract canopy structure and gap light transmission
indices from true colour fisheye photographs users manual and program
documentation 40 pp Simon Fraser Univ and Inst of Ecosystem Stud
Burnaby B C
Gordon H R (1989) Can the Lambert Beer law be applied to the diffuse
attenuation coefficient of ocean water? Limnol Oceanogr 34 1389-
1409
Gordon N D et al (2004) Stream Hydrology An Introduction for Ecol
ogists 429 pp John Wiley Chichester
Jerlov N G (1976) Marine Optics 231 pp Elsevier Amsterdam
Julian J P M W Doyle S M Powers E H Stanley and A Riggsbee
(2008) Optical water quality in rivers Water Resour Res dot 10 1029/
2007WR006457 in press
Kelly D J et al (2003) Effects of solar ultraviolet radiation on stream
henthic communities An mtersite comparison Ecology 84 2724 -2740
Kenworthy S T and B L Rhoads (1995) Hydrologic control of spatial
patterns of suspended sediment concentration at a stream confluence
J Hvdrol 168 251 -263
Kiffney P M et al (2004) Establishing light as a causal mechanism
structuring stream communities in response to experimental manipulation
of riparian buffer width J North Am Benthol Soc 23 542 -555
Kirk J T O (1994) Light and Photosynthesis to Aquatic Ecosystems 509
pp Cambridge Umv Press New York
15 of 16

G03022 JULIAN ET AL LIGHT IN RIVERS G03022
Koch R W et al (2004) Phytoplankton growth in the Ohio Cumberland
and Tennessee Rivers USA Inter site differences in light and nutrient
limitation Aquat Ecol 38 17 -26
Leopold L B and T Maddock Jr (1953) The hydraulic geometry of
stream channels and some physiographic implications US Geol Sury
Prof Pap 252 1 -57
Mobley C D (1994) Light and Water Radiative Tiansfer in Natwal
Waters 592 pp Elsevier San Diego Calit
Mulholland P J et a] (200 1) Inter biome comparison of tactors control
ling stream metabolism Freshwater Biol 46 1503 -1517
Palmroth S et al (2005) Contrasting responses to drought of forest floor
CO2 efflux in a Loblolly pine plantation and a nearby Oak Hickory
forest Global Change Biol 11 421 -434
Phlips E J et al (2000) Light availability and vanations in phytoplankton
standing crops in a nutrient rich black-water river Limnol Oceanogr 45
916 -929
Poff N L et al (1997) The natural flow regime BioScience 47 769-
784
U S Department of Agriculture (USDA) (2007) UV B Monitoring and
Research Program (Available at http / /uv mel colostate edu/UVB/)
Scarsbrook M R and J Halliday (1999) Transition from pasture to native
forest land use along stream continua Effects on stream ecosystems and
implications for restoranon N Z J Mar Freshwater Res 33 293 —3 10
Smith D G et al (1997) Optical characteristics of New Zealand rivers in
relation to flow J Am Water Resour Assoc 33 301 -312
Song C H and L E Band (2004) MVP A model to simulate the spatial
patterns of photosynthetically active radiation under discrete forest cano
pies Can J For Res 34 1192 -1203
Strayer D L et al (2006) Using geophysical information to define
benthic habitats in a large river Freshwater Biol 51 25 -38
Taylor S L et a] (2004) Catchment urbanisation and increased benthic
algal biomass in streams Linking mechanisms to management Fresh
water Biol 49 835 -851
Vannote R L et al (1980) The river continuum concept Can J Fish
Aquat Sci 37 130 -137
Westlake D F (1966) The light climate for plants in rivers m Light as an
Ecological Factor edited by R Bambridge et al pp 99-118 Blackwell
Set Oxford
Wetzel R G (2001) Limnology Lake and River Ecosystems 1006 pp
Elsevier San Diego Calif
Yard M D et al (2005) Influence of topographic complexity on solar
insolation estimates for the Colorado River Grand Canyon AZ Ecol
Modell 183 157 -172
Zheng X et al (2002) Variability of the downwelhng diffuse attenuation
coefficient with consideration of inelastic scattering Appl Opt 41
6477 -6488
M W Doyle Department of Geography University of North Carolina
205 Saunders Hall Chapel Hill NC 27599 3220 USA (mwdoyle@
email unc edu)
J P Julian Appalachian Laboratory University of Maryland Center
for Environmental Science 301 Braddock Road Frostburg MD 21532
USA (juhan@al umces edu)
E H Stanley Center for Limnology University of Wisconsin 680 N
Park Street Madison WI 53706 USA (ehstanley @wise edu)
16 of 16