The URL can be used to link to this page

Home My WebLink AboutJune_4,_2010_-_habitat_guilds_for_suitability_criteria_&_aquatic_life_decline_rateHello everyone. Glad that many of you continue to work on flow issues. I have two articles to share, both from SIFN members. Thank you to Donald Orth and Jeff Sole. These will also be posted on the SARP web site. Feel free to share them and use the information in them. 1. Research paper from Donald Orth of Virginia SIFN team, "Using habitat Guilds to Develop Habitat Suitability Criteria for a Warm water Stream Fish Assemblage." Orth is a co-author. Paper is attached. 2. From Jeff Sole of Kentucky SIFN team, a copy of a news article that he received by email. It is reproduced below. Center for Watershed Protection News Aquatic Life Declines at Early Stages of Urban Development The Center for Watershed Protection has been collaborating with the US Geological Survey's Effects of Urbanization on Stream Ecosystems research group to help interpret and disseminate the study results to local watershed managers and planners so they can base land use and management decisions on the best available science. A summary report of the EUSE study findings is now available, along with short video podcasts, as described in the summary below. A new USGS report explains the effects of urban development on stream ecosystem health. Surprisingly, aquatic insect communities show little, if any, initial resistance to low levels of urban development that were previously thought to be protective of aquatic life. The study showed, for example, that by the time a watershed reaches about 10 percent impervious cover in urban areas, aquatic insect communities are degraded by as much as 33 percent in comparison to aquatic insect communities in forested watersheds. The USGS determined the magnitude and pattern of the physical, chemical, and biological response of streams to increasing urbanization and how these responses vary throughout nine metropolitan areas: Portland, OR; Salt Lake City, UT; Birmingham, AL; Atlanta, GA; Raleigh, NC; Boston, MA; Denver, CO; Dallas, TX; and Milwaukee, WI. Comparisons among the nine metropolitan areas show that not all urban streams respond in a similar way. Land cover prior to urbanization can affect how aquatic insects and fish respond to urban development and is important to consider in setting realistic stream restoration goals in urban areas. Learn more about how stream ecosystems respond to urban development from USGS reports and video podcasts on the USGS website: http://water.usgs.gov/nawqa/urban/ The Center for Watershed Protection works to protect, restore, and enhance our streams, rivers, lakes, wetlands, and bays. We create viable solutions and partnerships for responsible land and water management so that every community has clean water and healthy natural resources to sustain diverse life Center for Watershed Protection | 8390 Main Street, 2nd floor | Ellicott City | MD | 21043 Have a great weekend. Marilyn O'Leary -- Marilyn Barrett-O'Leary Southeast Aquatic Resources Partnership SIFN Coordinator marilyno@southeastaquatics.net 225-892-7470 Persinger Habitat Guilds River Res Applic 2010.pdf 362K View Download USING HABITAT GUILDS TO DEVELOP HABITAT SUITABILITY CRITERIA FOR A WARMWATER STREAM FISH ASSEMBLAGE J. W. PERSINGER, *,y D. J. ORTH and A. W. AVERETT z Department of Fisheries and Wildlife Sciences, Virginia Polytechnic Institute and State University, 100 Cheatham Hall, Blacksburg, Virginia 24061-0321, USA ABSTRACT The diversity of ﬁsh species found in warmwater stream systems provides a perplexing challenge when selecting species for assessment of instream ﬂow needs from physical habitat analyses. In this paper we examined the feasibility of developing habitat suitability criteria (HSC) for the entire ﬁsh community of a warmwater stream using habitat guilds. Each species was placed a priori into a guild structure and habitat data were collected for depth, velocity, Froude number, distance to cover, embeddedness and dominantandsubdominantsubstrate.Correctguild classiﬁcationwastestedwithlinear discriminantanalysis foreach species.Correct classiﬁcation based on habitat-use data was highest for rifﬂe and pool-cover guilds, whereas the fast-generalist and pool-run classes, the broader nicheguilds, were more frequently misclassiﬁed. Variables most important for discriminating guilds were Froude number, velocity and depth in that order. Nonparametric tolerance limits were used to develop guild suitability criteria for continuous variables and the Strauss linear index was used for categorical variables. We recommend the use of a wide array of variables to establish more accurate habitat analysis. Additionally, guild HSC can be developed with similar effort to that needed to develop HSC for a small number of individual species. Results indicate that a habitat guild structure can be successfully transferred to another river basin and thathabitatsforadiverseﬁshassemblagecanbeadequatelydescribedbyasmallnumberofhabitatguilds.Thisapproachrepresentsan alternative for incorporating entire ﬁsh assemblages into habitat analyses of warmwater stream systems. Copyright #2010 John Wiley & Sons, Ltd. key words: habitat guilds; habitat suitability criteria; instream ﬂow; physical habitat analyses; warmwater streams; ﬁsh assemblages Received 11 May 2009; Revised 16 September 2009; Accepted 4 March 2010 INTRODUCTION Advances in modelling hydrodynamics of natural stream channels, including two- and three-dimensional models, provide more spatial resolution in habitat conditions. Yet these tools need to be integrated with habitat-use descriptors for resident ﬂora and fauna so ﬂow-habitat tradeoffs can be accurately described. Instream Flow Incremental Method- ology (IFIM), Physical Habitat Simulation (PHABSIM), MesoHABSIM (Parasiewicz, 2001), EVHA (Ginot, 1995) and similar systems are valuable tools for resource agencies to use when facing the difﬁcult challenge of managing stream resources. The effectiveness of these techniques depends on the accuracy of the data used to describe the habitat needs of aquatic communities (Orth, 1995; Freeman et al., 1997; Mouton et al., 2007). A species’ habitat is described by stream-speciﬁc or previously established habitat suitability criteria (Bovee, 1986; Crance, 1987). If the habitat datum used is inaccurate then modelling efforts will fail to determine how changes in stream ﬂow affect the habitat available to a species or group of species (Waite and Barnhart, 1992; Bovee, 1994). While accurate habitat data is necessary for habitat analysis, the data also needs to represent the entire aquatic community (Moyle and Baltz, 1985; Orth, 1987; Gan and McMahon, 1990). Having habitat information for only one or two species in warmwater stream systems limits the usefulness of habitat model output. If only a small portion of the community is represented, then ﬂows thought to protect the integrity of the system may actually be detrimental to it (Bain et al., 1988; Lobb and Orth, 1991; Aadland, 1993). Using habitat guilds to represent the habitat needs of the aquatic community has been proposed as a solution to this problem (Orth, 1987; Leonard and Orth, 1988; Lobb and Orth, 1991; Aadland, 1993; Welcomme et al., 2006). Habitat guilds are treated as super species and their criteria are established from the data collected for all members of the guild (Gorman, 1988; Austen et al., 1994). This way all members of the guild are represented by the guild criteria. The drawback of this approach is the lack of RIVER RESEARCH AND APPLICATIONS River. Res. Applic.(2010) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rra.1400 *Correspondenceto:J.W.Persinger,MissouriDepartmentofConservation, Grassland Field Station, PO Box 368, Clinton, MO 64735, USA. E-mail: jason.persinger@mdc.mo.gov y Present address: Missouri Department of Conservation, Grassland Field Station, PO Box 368, Clinton, MO 64735, USA.z Virginia Department of Environmental Quality, 7705 Timberlake Road, Lynchburg, VA 24502, USA. Copyright #2010 John Wiley & Sons, Ltd. accurate habitat information for many species, which makes guild placement difﬁcult (Angermeier, 1987). The difﬁculty of guild placement is an important consideration because species placed incorrectly will create inaccurate habitat criteria. Therefore, the placing of species into the proper guild is a principal task in developing habitat criteria. Several different guild structures have been proposed for use in instream ﬂow studies with the number of guilds being used varying from 4–7 (Bain et al., 1988; Lobb and Orth, 1991; Aadland, 1993; Vadas and Orth, 1997; Vadas and Orth, 2000). The habitat guild structures proposed have always been developed in the system being studied. The establishment of site-speciﬁc guilds requires a large amount of data to be gathered on the ﬁsh community. This is at odds with the main purpose of using the guild approach, which is to reduce the data needed to establish habitat criteria. Guildstructuresneed tobeable totransfertosimilar systems with little or no alterations if this approach is to be useful. Fish habitat research has progressed little in determining the best way to use guilds in developing habitat suitability criteria (with the exception of Vadas and Orth, 2001) or if guild structures established in one system can be used elsewhere. In this study we evaluate the validity of using a habitat guild approach and examine alternative ways to establish habitat criteria for a warmwater stream ﬁsh community. METHODS Site description This study focused on the North Fork Shenandoah River basin located in the Ridge and Valley Physiographic Province of northwestern Virginia. Our sampling sites ranged from 30 to 130km upstream of the river’s conﬂuence with the South Fork Shenandoah River. The study sites for the summer of 2001 and 2002 were selected based on a mesohabitat assessment of the North Fork Shenandoah River, conducted during the fall of 1998 and spring of 1999 by Don Hayes and Peter Ruhl of the United States Geologic Survey (Krstolic et al., 2006). Seven ﬁsh sampling locations for the North Fork Shenandoah River were selected to proportionally represent the predominant habitat types (rifﬂe, run, pocket run and pool) found throughout the river. Flow levels during summer sampling (June–Septem- ber) were at historic lows, ranging from 3.5 to 228.5m 3 s 1 in 2001 and 2.0 to 22.8m 3 s 1 in 2002, at the Strasburg gagging station (USGS #01634000). Habitat sampling We conducted ﬁsh and habitat sampling using direct underwater observation and electroﬁshing using a throwable anode. The use of two separate sampling techniques allowed for a wide range of habitats and species to be sampled at a higher efﬁciency than either technique would have individually (Persinger et al., 2004). Snorkelling surveys were conducted using modiﬁed static-drop techniques described by Li (1988). After lanes were established, the snorkelers moved slowly upstream along the ropes and dropped a marker at each ﬁsh location. All ﬁsh were identiﬁedtospeciesexceptforthesatinﬁnshiner(Cyprinella analostana) and the spotﬁn shiner (Cyprinella spiloptera). These two species were identiﬁed to genus because of the difﬁculty involved in ﬁeld identiﬁcation and will be referred herein as Cyprinella spp. We used a stratiﬁed random sampling method to quantify the range of available habitat in the North Fork Shenandoah River (Grossman and Skyﬁeld, 2009). Prior to sampling, an initial distance within 6m of the starting point of the sample was chosen at random; a marker was placed at this point during the survey and at 6m intervals through the rest of the sampling area. After the survey was completed, we returned to each marked location and measured the habitat. The dominant substrate, subdominant substrate, embeddedness and cover were described within a 1m 2 area around all marked ﬁsh and random locations using a modiﬁed version ofthe Wentworthclassiﬁcation system(Bovee,1982).Water column depth (m), mean water column velocity (ms 1) and distance to nearest cover (m) were also measured. For electroﬁshing using a throwable anode the river was divided into ﬁve equal sized lanes and sampled using a modiﬁed version of the diamond-sampling pattern (Bovee, 1994). Even if no ﬁsh were captured, all sampled locations were marked to get available habitat data. Collected ﬁsh were identiﬁed to species and recorded. The group would then proceed to the next sampling site. After the run was completed, the same habitat variables recorded during the snorkelling surveys were measured at all marked locations. For a complete description of the ﬁsh sampling techniques see Persinger et al. (2004). Froude numberwascalculatedforallthedatapointsusing the measured site depth and velocity: F ¼v=ðgdÞ1=2 where v is equal to the mean water column velocity,d is equal to water column depth and g is equal to gravity (Gordon et al., 1992). Guild structure selection and testing The habitat guild structure used in this study is a modiﬁed version of one developed for the Roanoke River, VA (Vadas and Orth, 1997, 2000). The guilds used in this study are rifﬂe, fast generalist, pool-run and pool- cover (Figure 1). The rifﬂe, fast generalist and pool-cover Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra J. W. PERSINGER ET AL. guilds are each a combination of two guilds described by Vadas and Orth (2000). The pool-run guild was unchanged from Vadas and Orth (2000). Guilds were combined due to the similarity of the habitat described by the guilds, a lack of habitat that matched a guild or a lack of species that ﬁt into a guild. Prior to sampling, the adult life stages of all species were classiﬁed into the four habitat guilds based on Vadas and Orth (2000) or on habitat information taken from literature. Due to a lack of literature information on the habitat use of the juvenile and age-0 life stages, these life stages were not placed initially. The species placement and guild structure were tested using collected habitat measurements. Discriminant analysis was used to examine the existence of the guilds in the North Fork Shenandoah River, the species placement in the habitat guilds, and to develop a linear discriminant function for placing the other life stages. TheexistenceoftheguildswastestedusingtheMahalanobis squared distance betweentheguilds at a signiﬁcance levelof 0.05 and using a misclassiﬁcationmatrix. Speciesplacement was examined using the linear discriminant function to calculate guild placement for every data point of each species. SAS version 8e (SAS, 2000) was used for all statistical analysis. Guild habitat suitability criteria The data collected for all species assigned to a given guild were combined to form the data set used to develop the habitat suitability criteria. Criteria for each guild were developed for depth, velocity, Froude number, substrate, embeddedness, cover presence and distance to cover. Nonparametric tolerance limits were used to create criteria for the continuous variables depth, velocity, Froude number anddistancetocover(Newcomb etal.,2007).StraussLinear Index was used to create habitat suitability criteria for the categorical variables substrate, embeddedness, cover pre- sence, and distance to cover (Strauss, 1979). The substrate criteria included the dominant substrate and subdominant substrate. Cover presencewas based on whether or not cover could be found within a 1m area of the ﬁsh location. Distance to cover criteria was developed using both nonparametric tolerance limits and Strauss Linear Index because the data was collected as continuous data out to a distance of 0.5m, but all measurements greater than 0.5m were grouped into a single category (>0.5m). Criteria developed using both techniques were compared to determine if it should be treated as a categorical or continuous variable. The nonparametric tolerance limitswereusedtoconstruct type-III habitat suitability criteria for all species. The tolerance limits for the central 50% were used to establish the cutoffs for optimal habitat, which had a suitability value of one. The data located between the central 50% tolerance limits and the central 75% were given a suitability value of 0.5. The data located between the central 75% tolerance limits and the central 90% received a suitability value of 0.2. The data beyond the central 90% tolerance limit received a suitability of zero and were considered unsuitable habitat. The Strauss Linear Index (L) was used to develop habitat suitability criteria for the categorical variables. The linear index is the statistical difference in the proportion of species use versus the proportion of availability. The sampling variance of the linear index allows a statistical comparison between the calculated value and the Null-hypothesis value ofzero (Strauss,1979). Thelinear indexwas calculatedatan alpha of 0.05 for each level of the ﬁve categorical variables. Criteria were developed for each variable using the index values and the signiﬁcance tests. Values with positive signiﬁcance were considered optimal habitat and given a suitability of one. Negatively signiﬁcant values were considered unsuitable habitat and given a suitability level of zero. Non-signiﬁcant categories were considered usable habitat and given suitability values of 0.5 for positive values and 0.2 for negative values. Figure 1. Diagrams of the guild structures used for this study. The Vadas and Orth (2000) structure served as the basis for the creation and imple- mentation of the North Fork Shenandoah guild structure Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra HABITAT GUILDS FOR HABITAT SUITABILITY CRITERIA RESULTS Guild structure testing The data collected for the adult life stages of all species thatwere placed apriori intotheguild structurewere used to run a discriminant analysis of the guild structure. The Wilks Lambda statistic had a p-value of <0.0001 indicating that there were signiﬁcant differences among the groups. Mahalanobis distance test (Table I) found each guild was signiﬁcantly different from the other three guilds. The linear discriminant function was used to determine what habitat variablesweremostimportanttoeachguild;Froudenumber, velocity and depth, respectively, were the three most important variables for all four guilds (Table II). Embedd- edness was the fourth most important variable for the rifﬂe guild while distance to cover was the fourth most important for the other three guilds. A misclassiﬁcation matrix of all the data points (Table III) had an overall accuracy of 46.7%. For two guilds (rifﬂe 64.4% and pool-cover 58.9%) the majority of data points were assigned to the correct guild. For the fast generalist guild, the highest percentage (42.7) of data points was assigned to the proper guild, but a large percentage (32.6) of data points were assigned to the rifﬂe guild. The pool-run guild had only 8.9 per cent of the data assigned to the proper guild. The linear discriminant function was used to conﬁrm the a priori placement of species into the guilds and place the juvenile and age-0 life stage groups that were signiﬁcantly different in their habitat use from the adult life stages into guilds by calculating the proper guild for each data point. Several a priori placed species had a higher percentage of their data in a guild different from the one in which they were assigned. The two main reasons for this were a lack of observations for a species (e.g. tessellated darter,Etheos- toma olmstedi) or a lack of good habitat information on which to base guild placement for a species (e.g. river chub, Nocomis micropogon); however, no species was moved to another guild because the guilds themselves were signiﬁ- cantly differentfromeach other.Eachjuvenileandage-0life stage was placed into the guild with the highest percentage of its data points (Table IV). Guild habitat suitability criteria The rifﬂe and fast generalist guild’s habitat suitability criteria indicate they are using a smaller range of depth and shallower depths than the pool-run and pool-cover guilds (Figure 2). The velocity criteria (Figure 3) indicate that the rifﬂe guild is using the widest range and fastest velocities, pool-run and fast generalist guilds are usingthe intermediate velocities, and pool-cover guild is using the slowest velocities. The criteria developed for Froude number (Figure 4) shows a virtually identical pattern to that seen in the velocity criteria. Table I.p-values for a test of the Mahalanobis squared distance between the guilds. Null hypothesis being tested is no signiﬁcant difference between the guilds. A 0.05 p-value was used to deter- mine signiﬁcance Rifﬂe Fast Generalist Pool/ Run Pool- cover Rifﬂe 1.0000 Fast Generalist <0.0001 1.0000 Pool/Run <0.0001 <0.0001 1.0000 Pool-cover <0.0001 <0.0001 <0.0001 1.0000 Table II. Linear discriminant function for habitat guilds Variable Rifﬂe Fast Generalist Pool/ Run Pool- cover Constant 26.836 26.493 28.346 28.893 Depth 9.865 11.860 13.832 14.692 Velocity 13.661 12.556 16.009 18.056 Distance to Cover 5.220 7.909 7.867 8.358 Dominant substrate 2.400 2.345 2.376 2.421 Subdominant substrate 1.891 1.900 1.932 1.866 Embeddedness 7.404 7.272 7.726 8.050 Froude number 41.662 36.072 41.607 42.212 Table III. Numberofguildspeciesobservationsandpercentclassiﬁedintoeachguild.Thetotalcolumnistheoverallnumberofobservations for the guild species. Overall accuracy ¼46.7% Actual Guild Membership Number and Per cent of Guild Species Observed As: Rifﬂe Fast Generalist Pool/Run Pool-cover Total Rifﬂe 217 (64.4) 90 (26.7) 11 (3.3) 19 (5.6) 337 (100) Fast Generalist 120 (32.6) 157 (42.7) 22 (6.0) 69 (18.8) 368 (100) Pool/Run 59 (25.1) 82 (34.9) 21 (8.9) 73 (31.1) 235 (100) Pool-cover 47 (12.8) 83 (22.6) 21 (5.7) 216 (58.9) 367 (100) Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra J. W. PERSINGER ET AL. Optimal substrate habitat for rifﬂe guild and fast generalist guild ranges from small cobble to small boulder. Suitable habitat ranges from large gravel to ﬂat bedrock for the rifﬂe guild and from small gravel to ﬂat bedrock for the fast generalist guild (Figure 5). Pool-run guild optimal substrate ranges from large cobble to small boulder with suitable habitat ranging from small cobble to tilted bedrock (Figure 6). Small cobble is the optimal habitat for the pool- cover guild and all other substrate types except tilted bedrock is considered suitable (Figure 6). The rifﬂe, fast generalist and pool-run guilds all have the same criteria for embeddedness with 0–25% embeddedness being optimal habitat and anything more embedded considered unsuitable (Figure 7). The pool-cover guild has an optimal embeddedness of 25–50%, with anything greater than 25% being suitable (Figure 7). The cover presence criteria indicated that all four guilds preferred locations with cover (Figure 8). The distance to cover HSC was created using both nonparametric tolerance limits and Strauss linear index values. For the Table IV. Final guild placement of the North Fork Shenandoah River species Species Guild N R FG PR PC Greenside darter (Etheostoma blenniodies) R 1 1.00 0.00 0.00 0.00 Mottled sculpin (Cottus bairdi) R 237 0.57 0.33 0.04 0.06 Central stoneroller (Campostoma anomalum) R 29 0.76 0.17 0.03 0.03 Longnose dace (Rhinichthys cataractae) R 70 0.81 0.14 0.01 0.03 Bluehead chub (Nocomis leptocephalus) (J) R 23 0.52 0.35 0.04 0.09 River chub (Nocomis micropogon) (Y) R 9 0.44 0.11 0.11 0.33 Potomac sculpin (Cottus girardi) FG 1 0.00 1.00 0.00 0.00 Margined madtom (Noturus insignis) FG 81 0.27 0.47 0.07 0.19 Rosyface shiner (Notropis rubellus) FG 16 0.63 0.31 0.00 0.06 Comely shiner (Notropis amoenus) FG 13 0.23 0.54 0.08 0.15 Spotﬁn/satinﬁn shiner (Cyprinella spp.) (A) FG 211 0.29 0.42 0.06 0.23 Bull chub (Nocomis raneyi) FG 1 0.00 0.00 0.00 1.00 Bluehead chub (Nocomis leptocephalus) (A) FG 43 0.56 0.33 0.07 0.05 Blacknose dace (Rhinichthys atrastulus) FG 2 1.00 0.00 0.00 0.00 Fallﬁsh (Semotilus corporalis) (J&Y) FG 44 0.20 0.52 0.05 0.23 Tessellated darter (Etheostoma olmstedi) PR 1 0.00 1.00 0.00 0.00 Common shiner (Luxilus cornutus) PR 23 0.43 0.57 0.00 0.00 River chub (Nocomis micropogon) (A&J) PR 69 0.42 0.41 0.06 0.12 Fallﬁsh (Semotilus corporalis) (A) PR 51 0.18 0.51 0.12 0.20 Rosyside dace (Clinostomus funduloides) PR 1 0.00 0.00 0.00 1.00 Green sunﬁsh (Lepomis cyanellus) (A&J) PR 45 0.09 0.11 0.13 0.67 White sucker (Catostomus commersoni) PR 6 0.00 0.00 0.00 1.00 Northern hog sucker (Hypentelium nigricans) PR 39 0.18 0.23 0.10 0.49 Banded killiﬁsh (Fundulus diaphanus) PC 1 0.00 0.00 0.00 1.00 Swallowtail shiner (Notropis procne) PC 1 0.00 1.00 0.00 0.00 Spottail shiner (Notropis hudsonius) PC 14 0.14 0.36 0.07 0.43 Bluntnose minnow (Pimephales notatus) PC 54 0.26 0.31 0.06 0.38 Common carp (Cyprinus carpio) PC 1 0.00 0.00 0.00 1.00 Pumpkinseed (Lepomis gibbosus) PC 1 0.00 0.00 0.00 1.00 Bluegill (Lepomis macrochirus) PC 1 0.00 0.00 0.00 1.00 Redbreast sunﬁsh (Lepomis auritus) PC 206 0.06 0.15 0.04 0.75 Largemouth bass (Micropterus salmoides) PC 7 0.00 0.00 0.00 1.00 Smallmouth bass (Micropterus dolomieu) PC 138 0.17 0.29 0.04 0.50 Rock bass (Ambloplites rupestris) PC 71 0.14 0.27 0.07 0.52 Brown bullhead (Ameiurus nebulosus) PC 4 0.00 0.25 0.00 0.75 Yellow bullhead (Ameiurus natalis) PC 35 0.08 0.29 0.06 0.57 Channel catﬁsh (Ictalurus punctatus) PC 1 0.00 0.00 0.00 1.00 Spotﬁn/satinﬁn shiner (Cyprinella spp.) (J&Y) PC 46 0.13 0.28 0.04 0.54 Bluehead chub (Nocomis leptocephalus) (Y) PC 11 0.18 0.27 0.00 0.55 Green sunﬁsh (Lepomis cyanellus) (Y) PC 27 0.11 0.26 0.00 0.63 R, rifﬂe; FG, fast generalist; PR, pool-run; and PC, pool-cover.N is the number of observations for each species or life stage collected. Species that showed differencesinhabitatusebasedonlifestagewereassignedtoguildsseparatelyforeachlifestage.Theletterinparenthesisfollowingthenamerepresentsthelife stage. A, adult; J, juvenile; Y, young of year and no letter means all life stages grouped together. The decimal fraction of data assigned to a guild is listed in the guild-speciﬁc columns. The percentage of data assigned to a guild is listed in the guild-speciﬁc columns. Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra HABITAT GUILDS FOR HABITAT SUITABILITY CRITERIA rifﬂe guild and the fast generalist guild the Strauss criteria indicate that they need to be much closer to cover than is indicated with the tolerance limits (Figure 9). Although both sets of criteria cover similar ranges, the suitability of the Strauss criteria declines at a faster rate for the pool-run and pool-cover guilds than the tolerance limits criteria (Figure 9). Figure 2. Depth habitat suitability criteria created using nonparametric tolerance limits for the four guilds sampled in the North Fork Shenandoah River during the summers of 2001 and 2002. Line width gradients are the central 50% (thickest line), central 75% (medium line), and the central 90% (thinnest line). For each guild the sample totals are: rifﬂe n¼338, fast generalist n¼351, poolrun n ¼194, and pool-cover n¼415 Figure 3. Velocity habitat suitability criteria created using nonparametric tolerance limits for the four guilds sampled in the North Fork Shenandoah River during the summers of 2001 and 2002. Line width gradients are the central 50% (thickest line), central 75% (medium line), and the central 90% (thinnest line). For each guild the sample totals are: rifﬂe n¼338, fast generalist n ¼351, pool-run n¼194, and pool-cover n ¼415 Figure 4. Froude number habitat suitability criteria created using nonpara- metric tolerance limits for the four guilds sampled in the North Fork ShenandoahRiverduringthesummers of2001and2002.Linewidthgradients are the central 50% (thickest line), central 75% (medium line), and the central 90% (thinnest line). For each guild the sample totals are: rifﬂe n¼338, fast generalist n¼351, pool-run n¼194, and pool-cover n¼415 Figure 5. Rifﬂeguildandfastgeneralistguildsubstratehabitatcriteria.The guild bars represent the frequency that the substrate category was used by members of the guild and the available bars represent the frequency that substrate was found in all sampled locations in the North Fork Shenandoah River. The number of observations used were: rifﬂe guild N¼676, fast generalist guild N¼702, and available N¼3176. Suitability values were based on the signiﬁcance of the Strauss linear index values calculated Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra J. W. PERSINGER ET AL. Overall the rifﬂeguild prefers shallow,fast waterwith low embeddedness, cobble sized substrate and nearby cover.The fast generalist guild prefers locations with medium depths and velocities, cobble to boulder-sized substrate, low embeddedness and nearby cover. The pool-run guild prefers locations with deeper depths, medium velocities, cobble to boulder-sized substrate, low embeddedness and nearby cover. The pool-cover guild prefers deeper, slower water with embedded substrate and nearby cover. DISCUSSION The lessons learned during this study focus mainly on ways to include the elements of ﬁsh habitat diversity into habitat criteria:(1)variablesotherthan depth,velocityandsubstrate are important to habitat choice in species, (2) data can be gathered on multiple species at a time without much more effort than gathering data on one species, (3) habitat guild structurescanworkinothersystems,(4)guildcriteria canbe createdwithoutanymoreeffortthan creating species criteria and (5) Strauss linear index appears to provide a reasonable approach for developing criteria for categorical variables. Traditional habitat analysis has focused on a few individual variables such as depth, velocity and substrate. Habitat suitability criteria have traditionally been developed for individual variables separately and then the habitat is evaluated based on a combination of these independently developedcriteria.Recentworkhastriedtomoveawayfrom the individual variable approach to include complex hydraulic variables because species make habitat choices based on multiple factors at the same time (Brooks et al., 2005; Lamouroux and Jowett, 2005). Froude number was used to address multiple aspects of a species’ habitat choice with one variable. Froude number is a complex hydraulic variablethataccountsfordepthandvelocitysimultaneously; therefore, it may demonstrate a species’ habitat selection more accurately than either depth or velocity individually. Previous work has found the Froude number was signiﬁcantly related to macroinvertebrate abundance (Brooks et al., 2005) and reach habitat value (Lamouroux and Jowett, 2005; Schweizer et al., 2007). Similarly, our study supports the importance of Froude number for discriminating between ﬁsh habitat guilds. Distance to cover, although traditionally not examined in ﬁsh habitat studies, may also be an important habitat variable. Multiple studies have found that trout and salmon species select habitat closely related to cover although they showed no preference between cover types (Quinn and Kwak, 2000; Banish et al., 2008; Holecek et al., 2009). Additionally, while many ﬁsh species do not spend much time actually using cover they often remain near cover in case they need to use it (Groshens, 1993). Because of the way distance to cover was measured in this study, criteria were developed using both tolerance limits and the Strauss index. The two methods resulted in different criteria. The differences in the results are a cause for concern and warrant further investigation into the correct approach for evaluating distance to cover. Tolerance limits will probably result in the mostaccurate criteria fordistancetocoveras longasenough distance is considered when measurements are taken. By including variables such as Froude number and distance to cover, a more complete analysis of habitat selection was developed with little additional time spent in the ﬁeld. Because species select habitat based on a range of variables it is important to include multiple habitat variables and variables that combine multiple aspects of the habitat, such as Froude number, so that criteria represent a more realistic picture of how species and guilds select habitat. While distance to cover remains a cumbersome variable to incorporate into one-dimensional models such as PHAB- SIM, the development of two-dimensional models has made it easier to account for distance to cover in physical habitat analysis studies. Furthermore, advances in two-dimensional Figure 6. Pool-runguildandpool-coverguildsubstratehabitatcriteria.The guild bars represent the frequency that the substrate category was used by members of the guild and the available bars represent the frequency that substrate was found in all sampled locations in the North Fork Shenandoah River.Thenumberofobservationsusedwere:pool-runguildN¼388,pool- cover guild N ¼830, and available N¼3176. Suitability values were based on the signiﬁcance of the Strauss linear index values calculated Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra HABITAT GUILDS FOR HABITAT SUITABILITY CRITERIA modelling make it possible to incorporate spatially explicit variables in habitat analysis (Crowder and Diplas, 2006; Shen and Diplas, 2008). Typically instream ﬂow studies examine just one or two species; however, if these studies are going to provide useful information for making decisions in stream systems with diverse ﬁsh communities then criteria development needs to look at the entire aquatic community (Orth, 1987; Braaten and Berry, 1997). Collecting ﬁeld data on multiple species requires relatively little extra effort than data collection for a few speciﬁc species. This is particularly true of the guild approach which relies on the collectivedata for all species in a guild, thus reducing the amount of data needed for any particular species. A guild approach is most useful if the guilds are transferable from one river system to another with only minor modiﬁcations. This study took a guild structure previously established for the Roanoke River, Virginia (Vadas and Orth, 2000) and modiﬁed it for use on the North Fork Shenandoah River, Virginia. Initial modiﬁcations were made to the guild structure to account for different species between the two rivers. After initial species placement into the guilds, no species were moved despite some species having a higher percentage of their data in guilds other than the one they were assigned to. This was done because the guilds themselves were different from each other and any species movement would automatically change the deﬁ- nitions of the guilds. This could lead to endless changes as the guilds were constantly redeﬁned each time a species was moved; therefore, the species were all left in place. The results of this study conﬁrmed that the four guilds were signiﬁcantly different from each other in another river; therefore, the habitat guild structure used in one riversystem was successfully applied, with minor modiﬁcations, to another river system. While transferring a guild structure between two river systems with similar ﬁsh communities worked for this study, more research is needed to determine ifaguildstructurecanbeappliedtoawiderrangeofsystems successfully. Theprocessforcreating guild criteriawas identical tothat used in creating single species criteria, except data from multiple species were combined into a single set of criteria. Figure 7. Habitat criteria for embeddedness. The guild bars represent the frequency that the embeddedness category was used by members of the guild and the available bars represent the frequency that embeddedness was found in all sampled locations in the North Fork Shenandoah River. The number of observations usedwere:rifﬂeguildN¼338,fastgeneralistguildN¼351,pool-runguildN¼194,pool-coverguildN ¼415,andavailableN ¼1588.Suitabilityvalueswere based on the signiﬁcance of the Strauss linear index values calculated Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra J. W. PERSINGER ET AL. Figure 8. Habitatcriteriaforcover.TheX-axiscategoriesarecoverpresenceorabsence.Theguildbarsrepresentthefrequencythatcoverwaspresentorabsent at guild locations and available bars represent the frequency that cover was present or absent in all sampled locations in the North Fork Shenandoah River. The number of observations used were: rifﬂe guild N ¼338, fast generalist guild N ¼351, pool-run guild N¼194, pool-cover guild N¼415, and available N¼ 1588. Suitability values were based on the signiﬁcance of the Strauss linear index values calculated Figure 9. Comparison of habitat criteria created using nonparametric tolerance limits and distance groupings using Strauss linear index values. Line width gradients are the 100% suitable (thickest line), 50% suitable (medium line), and 20% suitable (thinnest line) Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra HABITAT GUILDS FOR HABITAT SUITABILITY CRITERIA The HSC were created using all the data for every species assigned to the guild. When multiple species assigned to the same guild shared a common data point, that data point was only counted one time when creating the criteria. The main issue surrounding the development of guild criteria involves how each species was included in the guild criteria. In this study every data point collected for every species was included in the ﬁnal guild criteria. As a result, the most common species were weighted more heavily than rare species. A way to counteract the effect of common species would be to weight all species equally in the guild criteria. The problem with this approach is getting enough data points for rare species. In some cases we were only able to get one or two observations for a species such greenside darter (Etheostoma blenniodies) in the rifﬂe guild or bull chub (Nocomis raneyi) in the fast generalist guild. That means these species made little contribution to the guild criteria, which could be problematic if they are important species such as a threatened or endangered species. The best approach for dealing with rare species when developing guild criteria needs further study. Nonparametric tolerance limits were used for creating habitat suitability criteria for the continuous variables for several reasons. Tolerance limits provide a consistent and repeatable way to create criteria when compared to the more arbitrary nature of curve ﬁtting techniques (Newcomb et al., 1995). With tolerance limits anyone can take the same data set and create identical criteria. The use of the Strauss linear index also provided a consistent and repeatable method for creating criteria for categorical habitat variables. The Strauss index is a statistical method for evaluating the ratio of per cent categorical variable utilization to its availability in the environment (Strauss, 1979). The use of this index reduces some of the subjectivity often associated with substrate and cover criteria development. Guild-based criteria development may improvethe ability of habitat suitability criteria to represent the habitat needs of adiverseaquaticcommunity.Ifﬁshhabitatguildsareusedin conjunction with habitat guilds for stream macroinverte- brates and other species groups then for the ﬁrst time instream ﬂow studies might come close to accounting for the habitat requirements of the entire aquatic community (Gore et al., 2001; Orth and Newcomb, 2002). This research represents the ﬁrst attempt to test habitat guild typology in another river basin. The results indicate that there is potential for guilds to transfer between stream systems. Additionally, the study results suggest several methods for including habitat diversity into ﬁsh habitat criteria. Variables, such as Froude number, combine multiple aspects of the habitat in order to represent species habitat choices. Though much research is needed to determine the best way to use a habitat guild approach in developing habitat suitability criteria in a stream system, the approach appears to have some deﬁnite beneﬁts for studies melding hydrodynamic models to habitat needs for ﬁsh assemblages. Further work is needed to test guild structures andtheirassociatedcriteriainotherrivers,thetransferability of guilds between river systems, and the best approach to including rare species into habitat guilds. ACKNOWLEDGEMENTS The authors would like to thank the numerous individuals who contributed to this study including T. Newcomb, C. A. Dolloff, D. Hayes, J. Krstolic, S. Reeser, J. Kauffman, J. Milam, J. Harris, C. Holbrook, L. Scarbourgh, V. Eaton, M. Anderson, J. Kilpatrick, T. Smith and M. Chan. They would also like to thank the NFSR basin landowners who allowed them access to the study sites. Funding for this study was provided by the Northern Shenandoah Valley Regional Commission and the Virginia General Assembly. They are grateful to Ken Bovee and one anonymous reviewer whose insightful comments helped improve this paper. REFERENCES Aadland LP. 1993. Stream habitat types: their ﬁsh assemblages and relationship to ﬂow.North American Journal of Fisheries Management 13: 790–806. Angermeier PL. 1987. Spatial temporal variation in habitat selection by ﬁshes in a small Illinois stream. In Community and evolutionary ecology of North American stream ﬁshes. Matthews WJ, Heins DC (eds). Univ. Oklahoma Press: Norman; 52–60. Austen DJ, Bayley PB, Menzel BW. 1994. Importance of the guild concept to ﬁsheries research and management.Fisheries 19: 12–20. Bain MB, Finn JT, Booke HE. 1988. Streamﬂow regulation and ﬁsh community structure.Ecology 69: 382–392. Banish NP, Peterson JT, Thurow RF. 2008. Physical, biotic, and sampling inﬂuences on diel habitat use by stream-dwelling bull trout.North American Journal of Fisheries Management 28: 176–187. Bovee KD. 1982. A guide to stream habitat analysis using the instream ﬂow incremental methodology.Instream Flow Information Paper 12. U.S. Fish and Wildlife Service FWS?OBS-82/26: Washington, D. C. Bovee KD. 1986. Development and evaluation of habitat suitability criteria for use in the instream ﬂow incremental methodology.Instream Flow Information Paper 21. U.S. Fish and Wildlife Service Biological Report 86 (7): Washington, D. C. Bovee KD. 1994.Data Collection Procedures for the Physical Habitat Simulation System. National Biological Service, RSM: Fort Collins, CO; 159. Braaten PJ,Berry CR Jr. 1997.Fish associationswith fourhabitat typesin a South Dakota prairie stream.Journal of Freshwater Ecology 12: 477– 489. Brooks AJ, Haeusler T, Reinfelds I, Williams S. 2005. Hydraulic micro- habitats and distribution of macroinvertebrate assemblages in rifﬂes. Freshwater Biology 50: 331–344. Crance JH. 1987. Guidelines for using the Delphi technique to develop habitat suitability index curves.U.S. Fish and Wildlife Service Biological Report: 82(10.134). Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra J. W. PERSINGER ET AL. Crowder DW, Diplas P. 2006. Applying spatial hydraulic principles to quantify stream habitat.River Research and Applications 22: 79–89. Freeman MC, Bowen ZH, Crance JH. 1997. Transferability of habitat suitability criteria for ﬁshes in warmwater streams.North American Journal of Fisheries Management 17: 20–31. Gan K, McMahon T. 1990. Variability of results from the use of PHABSIM in estimating habitat area.Regulated Rivers: Research & Management 5: 233–239. Ginot V. 1995. EVHA, un logiciel d’evaluation de l’habitat du poisson sous windows.Bulletin franc¸ais de la peˆche et de la pisciculture337/338/339: 303–308. Gordon ND, McMahon TA, Finayson BL. 1992.Stream Hydrology: An Introduction for Ecologists. John Wiley & Sons: Winchester, England. Gore JA, Layzer JB, Mead JIM. 2001. Macroinvertebrate instream ﬂow studies after 20 years: a role in stream management and restoration. Regulated Rivers 17: 527–542. Gorman OT. 1988. The dynamics of habitat use in a guild of Ozark minnows.Ecological Monographs 58: 1–18. Groshens TP. 1993. An assessment of the transferability of habitat suitability criteria for smallmouth bass,Micropterus dolomieu.Master’sThesis. Vir- ginia Polytechnic Institute and State University, Blacksburg, Virginia. Grossman GD, Skyﬁeld JP. 2009. Quantifying microhabitat availability: stratiﬁed random versus constrained focal-ﬁsh methods.Hydrobiologia 624: 235–240. Holecek DE, Cromwell KJ, Kennedy BP. 2009. Juvenile Chinook salmon summer microhabitat availability, use, and selection in a central Idaho wilderness stream.Transactions of the American Fisheries Society 138: 633–644. Krstolic JL, Hayes DC, Ruhl PM. 2006. Physical habitat classiﬁcation and instream ﬂow modeling to determine habitat availability during low-ﬂow periods, North Fork Shenandoah River.U.S. Geological Survey Scientiﬁc Investigations Report: 2006-5025: Virginia; 63 p. Lamouroux N, Jowett IG. 2005. Generalized instream habitat models. Canadian Journal of Fisheries and Aquatic Sciences 62: 7–14. Leonard PM, Orth DJ. 1988. Use of habitat guilds of ﬁshes to determine instream ﬂow requirements.North American Journal of Fisheries Man- agement 8: 399–409. Li SK. 1988. Measuring microhabitat in swift water. In Proceedings of a workshop on the development and evaluation of habitat suitability criteria. Bovee KD, Zuboy JR (eds). U. S. Fish and Wildlife Service Biological Report 88 (11): Washington, D.C.; 183–193. Lobb MD III, Orth DJ. 1991. Habitat use by an assemblage ofﬁshin a large warmwater stream.Transactions of the American Fisheries Society 120: 65–78. Mouton AM, Schneider M, Depestele J, Goethals PLM, De Pauw N. 2007. Fish habitat modelling as a tool for river management.Ecological Engineering 29: 305–315. Moyle PB, Baltz DM. 1985. Microhabitat use by an assemblage of California stream ﬁshes: developing criteria for instream ﬂow determi- nations.Transactions of the American Fisheries Society 114: 695–704. Newcomb TJ, Perry SA, Perry WB. 1995. Comparison of habitat suitability criteria for smallmouth bass (Micropterus dolomieu) from three West Virginia rivers.Rivers 5: 170–183. Newcomb TJ, Orth DF, Stauffer DF. 2007. Habitat Evaluation. In Analysis and Interpretation of Freshwater Fisheries Data. Brown ML, Guy CS (eds). American Fisheries Society: Maryland; 843–886. Orth DJ. 1987. Ecological considerations in the development and appli- cation of instream ﬂow-habitat models.Regulated Rivers: Research and Management 1: 171–181. Orth DJ. 1995. Food web inﬂuences on ﬁsh population response to instream ﬂow.Bulletin Fr. Peche Piscic 337/338/339: 317–328. Orth DJ, Newcomb TJ. 2002. Certainties and uncertainties in deﬁning essential habitats for riverine smallmouth bass. In Black Bass: Ecology, Conservation, and Management. Ridgway MS, Philipp DP (eds). American Fisheries Society: Bethesda, Maryland; 251–264. Parasiewicz P. 2001. MesoHABSIM: a concept for application of instream ﬂow models in river restoration planning.Fisheries 29: 6–13. Persinger JW, Orth DJ, Newcomb TJ. 2004. A comparison of snorkeling versus throwable anode electroﬁshing for evaluating stream ﬁsh habitat use.Journal of Freshwater Ecology 19: 547–557. Quinn JW, Kwak TJ. 2000. Use of rehabilitated habitat by brown trout and rainbow trout in an Ozark tailwater river.North American Journal of Fisheries Management 20: 737–751. SAS. 2000.SAS/STAT User’s Guide, Version 8e. SAS Institute: Cary, North Carolina. Schweizer S, Borsuk ME, Jowett IG, Reichert P. 2007. Predicting joint frequency distributions of depth and velocity for instream habitat assess- ment.River Research and Applications 23: 287–302. Shen Y, Diplas P. 2008. Application of two- and three-dimensional com- putational ﬂuid dynamics models to complex ecological stream ﬂows. Journal of Hydrology 348: 195–214. Strauss RE. 1979. Reliability estimates for Ivlev’s electivity index, the forage ratio, and a proposed linear index of food selection.Transactions of the American Fisheries Society 108: 344–352. Vadas RL Jr, Orth DJ. 1997. Species associations and habitat use of stream ﬁshes: the effects of unaggregated-data analysis.Journal of Freshwater Ecology 12: 27–37. Vadas RL Jr, Orth DJ. 2000. Habitat use of ﬁsh communities in a Virginia stream system.Environmental Biology of Fishes 59: 253– 269. Vadas RL Jr, Orth DJ. 2001. Formulation of habitat-suitability models for stream-ﬁsh guilds: do the standard methods work?Transactions of the American Fisheries Society 130: 217–235. Waite IR, Barnhart RA. 1992. Habitat criteria for rearing steelhead: a comparison of site-speciﬁc and standard curves for use in the instream ﬂow incremental methodology.North American Journal of Fisheries Management 12: 40–46. Welcomme RL, Winemiller KO, Cowx IG. 2006. Fish environmental guilds as a tool for assessment of ecological condition of rivers.River Research and Applications 22: 377–396. Copyright #2010 John Wiley & Sons, Ltd.River. Res. Applic.(2010) DOI: 10.1002/rra HABITAT GUILDS FOR HABITAT SUITABILITY CRITERIA