Regular issues
Monographs
Special Issues



Southeastern Naturalist
    SENA Home
    Range and Scope
    Board of Editors
    Staff
    Editorial Workflow
    Publication Charges
    Subscriptions

Other EH Journals
    Northeastern Naturalist
    Caribbean Naturalist
    Neotropical Naturalist
    Urban Naturalist
    Prairie Naturalist
    Eastern Paleontologist
    Journal of the North Atlantic
    eBio

EH Natural History Home

A Database and Meta-Analysis of Ecological Responses to Stream Flow in the South Atlantic Region
Ryan A. McManamay, Donald J. Orth, John Kauffman, and Mary M. Davis

Southeastern Naturalist, Volume 12, Monograph Number 5 (2013): 1–36

Full-text pdf: complete with cover or manuscript only (Accessible only to subscribers.To subscribe click here.)

 

Site by Bennett Web & Design Co.
2013 SOUTHEASTERN NATURALIST 12(Monograph 5):1–36 A Database and Meta-Analysis of Ecological Responses to Stream Flow in the South Atlantic Region Ryan A. McManamay1,*, Donald J. Orth2, John Kauffman3, and Mary M. Davis4 Abstract - Generalized and quantitative relationships between flow and ecology are pivotal to developing environmental flow standards based on socially acceptable ecological conditions. Informing management at regional scales requires compiling sufficient hydrologic and ecological sources of information, identifying information gaps, and creating a framework for hypothesis development and testing. We compiled studies of empirical and theoretical relationships between flow and ecology in the South Atlantic region (SAR) of the United States to evaluate their utility for the development of environmental flow standards. Using database searches, internet searches, and agency contacts, we gathered 186 sources of information that provided a qualitative or quantitative relationship between flow and ecology within states encompassing the SAR. A total of 109 of the 186 sources had sufficient information to support quantitative analyses. Ecological responses to natural changes in flow magnitude, frequency, and duration were highly variable regardless of the direction and magnitude of changes in flow. In contrast, the majority of ecological responses to anthropogenic-induced flow alterations were negative. Fish abundance, diversity, reproduction, and habitat consistently showed negative responses to anthropogenic flow alterations, whereas other ecological categories (e.g., macroinvertebrates and riparian vegetation) showed somewhat variable responses and even positive responses (e.g., algal abundance). Fish and organic matter had sufficient sample sizes to stratify natural flow-ecology relationships by specific flow categories (e.g., high flow, baseflows) or by physiographic province (e.g., Coastal Plain, Piedmont). After stratifying relationships, we found that significant correlations existed between changes in natural flow and fish responses. In addition, a regression tree explained 57% of the variation in fish responses to anthropogenic and natural changes in flow. Altogether, our results suggested that the source of flow change and the ecological category of interest played primary roles in determining the direction and magnitude of ecological responses. Furthermore, our results suggest that developing broadly generalized relationships between ecology and changes in flow at a regional scale is unlikely unless relationships are placed within meaningful contexts, such as environmental flow components or geomorphic settings. Introduction Understanding the role of hydrology in structuring the biota of river environments has been a recent theme in stream ecology (Allan 1995, Gordon et al. 2004). The complexity in which moving water interacts with physical and chemical properties of a stream provides the template for ecosystem functions (Cuffney and Wallace 1989, Hornick et al. 1981, Newbold et al. 1982) and the habitats to 1Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6351. 2Department of Fish and Wildlife Conservation, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061. 3John Kauffman, LLC, PO Box 66, Free Union, VA 22940. 4Southern Instream Flow Network, Southeastern Aquatic Resources Partnership, Decatur, GA 30030. *Corresponding author - mcmanamayra@ornl.gov. 2 Southeastern Naturalist Vol. 12, No. 5 which organisms are adapted (Bunn and Arthington 2002, Poff and Allan 1995, Poff et al. 1997). Although the potential and realized effects of hydrological modifications on ecological systems has been clearly communicated (Carlisle et al. 2011, Jackson and Pringle 2010, Pringle et al. 2000), the prevailing questions related to ecologically sustainable flows, such as “How much?” and “How often?”, remain to be answered (Richter et al. 1997, 2003). Because of the extent and potential impact of anthropogenic disturbances on flowing environments, river managers require consistent guidelines to organize water policies geared towards water consumption and streamflow regulation. Unfortunately, the complexity of socio-economic, political, and ecological needs typically result in overly simplistic, static-flow guidelines that do not protect the natural flow variability of river systems (Arthington et al. 2006). Predictable and quantitative relationships between flow and ecology are pivotal to informing social processes (Poff et al. 2010), such as developing environmental flow standards based on socially acceptable ecological conditions (Richter et al. 2003). For example, the Ecological Limits of Hydrologic Alteration (ELOHA) framework, the consensus view of 19 international scientists, includes developing empirically testable relationships between flow alteration and ecological responses within different river types (i.e., groups of streams and rivers characterized by similar hydrologic regimes and geomorphic characteristics) (Poff et al. 2010). Presumably, different types of rivers may respond similarly to flow alterations (Arthington et al. 2006). Developing these “flow-ecology” relationships will obviously require compiling hydrologic and ecological databases to support regional quantitative analyses. Water policies are typically developed within state boundaries without regard to holistic basin needs (Sherk 1991, 1994, 2000). However, from a regulatory perspective, some state water resource programs have established some impressive research-based environmental flow protection standards. In Massachusetts, Armstrong et al. (2011) established quantitative relationships between fluvial fish relative abundance and species richness in response to % alteration in August median flows from groundwater withdrawals and % impervious cover. Michigan established legislation requiring a model to estimate the potential impacts of water withdrawals (Zorn et al. 2008). Using habitat suitability indices, Zorn et al. (2008) developed a model to predict fish responses to withdrawal scenarios and by association, summer temperatures. Results of the model were integrated into sophisticated impact assessment software (IWR 2008). Due to practical constraints and the desire to develop more region-specific guidelines, there is a need to compile sources of information at small operating scales, such as ecoregions or the US Fish and Wildlife Landscape Conservation Cooperatives (USFWS 2012). One approach is to use publically available spatial datasets, such as the USGS NAQWA program (USGS 2012), the EPA REMAP program (EPA 2010), or localized agency-derived databases. However, overlapping hydrologic and ecological spatial datasets are not available for many areas, which may preclude robust analyses for particular basins or geopolitical boundaries 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 3 (Knight et al. 2008) or may lead to coarse national analyses (Carlisle et al. 2011). A different approach includes conducting reviews and meta-analyses of the existing literature concerning flow-ecology relationships. Within the specific context of ecological responses to anthropogenic flow modifications, literature compilations to inform water policy are not unprecedented (e.g., Loyd et al. 2003, Poff and Zimmerman 2010). Loyd et al.(2003) reviewed 70 grey and peer-reviewed studies predominately conducted in Australia but also from the global literature and found that 86% of the studies reported ecological changes associated with anthropogenic flow modifications. Poff and Zimmerman (2010) conducted a global review of 165 peer-reviewed sources and found that 92% of the papers reported decreases in ecological responses to anthropogenic-induced flow alterations. Both studies reported a deficiency in the number of studies containing sufficient information to conduct quantitative analyses. Furthermore, one of the central and yet disconcerting conclusions from both studies was that “simple thresholds” (Loyd et al. 2003) or “general, transferable quantitative relationships” (Poff and Zimmerman 2010) could not be obtained from the existing literature. Given the limiting information at the global or continental scale, what is more disconcerting is the potential for inadequate information at smaller regional scales. In response to a possible information deficit, developing generalized relationships between flow and ecology has become a nation-wide research priority (Poff et al. 2010). For example, the Southern Instream Flow Network (SIFN), a program within the Southeastern Aquatic Resources Partnership (SARP), defined a comprehensive research agenda in 2010 to improve and provide new tools for instream flow protection and management (SIFN 2010). The research agenda identified five prioritized scientific needs, one of which included using existing information to develop commonalities or generalizations in ecological responses to hydrologic alterations. From this information, hypotheses regarding ecological responses to changes in flow can be developed, tested, and then used to support establishing regional environmental flow standards. As a part of this larger coordinated effort, the South Atlantic Landscape Conservation Cooperative desired to compile and summarize sources of information relating altered flow to ecology within the South Atlantic Region (SAR): Alabama, Georgia, Florida, North Carolina, South Carolina, and Virginia. The primary purpose of this project was to compile a body of knowledge on empirical and theoretical relationships between flow and ecology in order to aid managers in developing environmental flow standards in the SAR. Specifically, our goals were to 1) compile sources of information, 2) develop a database of meta-data associated with each source, and 3) conduct a qualitative and quantitative meta-analysis to determine any generalized relationships between flow and ecology in the SAR. Because of the potential for limiting availability of information at this scale, we broadened our search to not only peer-reviewed publications within databases but also grey literature and unpublished data through contacting representative federal and state agencies. Empirically based, tested relationships between natural flow variation and ecological responses 4 Southeastern Naturalist Vol. 12, No. 5 obviously aid in generating hypothetical ecological response to anthropogenic flow alterations (Arthington et al. 2006, Poff et al. 2010). Thus, we expanded our search to include any ecological response to changes in flow, regardless of whether the source was from natural variation or anthropogenic stressors. Here we define natural flow variation as changes in flow attributed to natural hydrologic processes (e.g., increases in flow magnitude due to flooding or decreases in flow magnitude due to drought). We classify changes to the natural flow regime attributed to some anthropogenic stressor (e.g., decreases in high-flow duration due to urbanization, or increases in daily magnitudes due to hydroelectric generation) as anthropogenic flow alterations. Methods Overview Ecological response categories in our analysis included fish, macroinvertebrates, algal and macrophytic vegetation, riparian vegetation, aquatic and terrestrial wildlife, organic matter, ecosystem metabolism, and nutrients. For fish, macroinvertebrates, and riparian vegetation, responses included changes in abundance, growth, survival, species richness, community composition, diversity (including biotic integrity indices and trophic diversity), reproduction, behavior, and habitat. Although habitat is only a surrogate for ecological responses, habitat-flow relationships have been used extensively and successfully to aid in developing environmental flow standards (Annear et al. 2004, Tharme 2003) and they can be linked to biological changes if used appropriately (Poff et al. 2010). Reproduction studies typically included measures of nesting success, spawning habitat, and egg abundance. Behavior studies included fish and macroinvertebrate drift, migration, or changes in use of different habitat types in relation to flow. For studies involving periphyton and algae, typical responses included abundance, community composition, or primary production. Ecosystem responses included alterations in metabolism, such as gross primary production. We also included studies documenting changes in organic matter, such as decomposition rates, fine and coarse-particulate export, and nutritional quality. Nutrient responses were included as export or concentration in relation to flow. For relevant ecological categories, habitat responses included habitat suitability indices (HSI), habitat area (e.g., weighted usable area—habitat availability), or specific habitat components of a particular category (e.g., fish migratory habitat, floodplain habitat). Flow-habitat ecology studies included defined areas of the river channel or floodplain important to various ecological groups, such as the aerial extent of fish nursery habitat, but also theoretical relationships, such as those found in Instream Flow Incremental Methodology (IFIM) studies. IFIM studies evaluate changes in habitat suitability or area in relation to incremental changes in flow. These studies typically are conducted by staging discharges of various magnitudes from dams or by measuring dischargehabitat relationships at different time periods in free-flowing rivers. IFIM studies 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 5 vary from simple 1-dimensional habitat relationships for individual species to multi-dimensional habitat relationships that incorporate biogenergetic models or ecosystem-level process (Tharme 2003). Typically, habitat area or suitability is calculated using combinations of depth, velocity, and substrate preferences for aquatic organisms. We excluded IFIM studies from our analysis that used predefined habitat-suitability models (e.g., Edwards et al. 1982). We only included IFIM or habitat-suitability studies where habitat area or HSI was empirically derived from field assessments of habitat preferences within the location of interest. Other habitat studies (non IFIM) were included only if they presented habitat as a defined area of the field site in relation to flow. For example, Townsend (2001) assessed pre- and post-dam flooding regimes in the Roanoke River at Roanoke Rapids, NC and reported associated changes in the area of floodplain inundation and consequently, riparian species composition. Database Search To find peer-reviewed published literature, we conducted literature searches using three databases (BioOne, CabDirect, and ISI Web of Science) because they seemed the most appropriate in finding relevant flow-ecology articles and provided the broadest scientific coverage. An exhaustive list of keywords were developed, reviewed, and then shortened to provide the most relevant keywords for searches. We searched the US Geological Survey database containing openfile reports or professional papers within the SAR relating to flow-ecology relationships. Searches were also conducted within the Southeastern Association for Fish and Wildlife Agencies (SEAFWA) conference proceedings database since SEAFWA has broad membership and encompasses the entire SAR. Manual Search Agency reports or unpublished data related to flow-ecology relationships were obtained by contacting agency and university personnel or by using search engines on the web. We compiled contact information for members of the Instream Flow Council, Warmwater Streams Committee within the Southern Division of the American Fisheries Society, and SIFN within the SARP program. Each of the organizations contained members who specifically have conducted management or research related to fluvial ecology. Individuals were contacted by email or phone to request reports or unpublished data that specifically addressed quantitative relationships between alterations in flow, due to natural or anthropogenic causes, and ecological responses. Members responded to the request by sending reports or unpublished data in electronic form. Internet searches were conducted using “Google” as a search engine to find agency reports, theses and dissertations, conference proceedings, and publications in open-access, narrow readership journals. Keywords used in internet searches were commonly associated with studies cited in other reports or publications, specific high-profile rivers, agencies, hydroelectric project names, or specific species listed as endangered, threatened, or species-of-concern. All reports and unpublished data were 6 Southeastern Naturalist Vol. 12, No. 5 reviewed to ensure that the sources included an association with a legitimate management agency or university and provided a date and location of report/data. Database Construction All journal articles, conference papers, reports, and unpublished data were reviewed to determine their relevance in terms of 1) the occurrence within a SAR state, and 2) the ability to isolate one qualitative or quantitative relationship between a primary flow component (magnitude, duration, frequency, timing, rate of change; Olden and Poff 2003) and an ecological response. After isolating references that met our criteria, we constructed a database that included attribute information useful for stratification and a summary of each study (see Supplemental File 1, available online at https://www.eaglehill.us/SENAonline/ suppl-files/mon5-1123-McManamay-s1, and, for BioOne subscribers, at http:// dx.doi.org/10.1656/S1123.s1). A subset of variables and their descriptions used to stratify studies is provided in Table 1. We recorded the latitude, longitude, province, and drainage area of each study location. If a study encompassed more Table 1. Subset of explanatory variables used in meta-database construction. “Y” in the Tree column indicates variables were used in the regression tree (for fi sh responses only). Tree Description Explanatory variable State Primary state location of each study. Time start Beginning of hydrologic period of record encompassing study. Time end End of hydrologic period of record encompassing study. Sites Number of sites (see methods for site designation). River type Categorizes rivers into free-flowing, dam-regulated, regulated by canal or diversion, reservoir (inflow studies), or a combination . Flow alteration Y Source of change in flow: general-natural variation, natural flooding, natural drought, reservoir operation improvements, withdrawal, urbanization, agriculture, or channelization. Study scale Y Primary scale of analyses. Spatial, Temporal, or Spatio-temporal. Latitude/longitude Latitude/Longitude of study location (see methods for multiple sites). USGS gage number USGS gage number for study location, if present (see methods for multiple sites). Drainage area (DA) Drainage area or range in drainage area of study site (see methods for multiple sites). DA category Y Categories for drainage area: small (less than 12 mi2 or mesocosims), small-med (21–150 mi2), med (203–1972 mi2), med-large (2058–10,102 mi2), large (14,700–86,771 mi2), and indeterminant (refers to sloughs, wetlands, or tidal influenced waterbodies where drainage area could not be determined). Province Y Physiographic province of study location (see methods for multiple sites). Quantitative Indicates if quantitative information exists for both the flow alteration and the ecological response. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 7 than one location, we provided coordinates for the centroid of the study area associated with the nearest USGS stream gage site, if applicable, and the average drainage area. In each study, we recorded the primary flow components used in each study and recorded any associated flow sub-components that would provide additional information (e.g., annual flow, baseflows, flood, rise rate). Because flow sub-components became complex, we created an additional variable that lumped sub-components into composite flow categories (e.g., low flows, highflood flows). We categorized each study by the source of change in flow, whether due to anthropogenic sources (e.g., withdrawal, urbanization, reservoir operations) or natural variation (e.g., flood, drought, variation). We also categorized studies by ecological categories (e.g., fish, macroinvertebrate, riparian vegetation) and by types of responses (e.g., abundance, growth, habitat, behavior). In order to determine the spatial and temporal extent required to determine flow-ecology relationships, we assessed each study’s period of record and Table 1 continued. Tree Description Data type Indicates whether study was a peer-reviewed publication (Pr), grey literature (Grey), or unpublished data (Up). Hydrologic resolution Y Temporal resolution of the hydrologic data used in the analysis (annual, monthly, daily, sub-daily). Flow component Y Primary flow component of central importance in analysis (magnitude, duration, frequency, timing, rate of change). From Olden and Poff (2003). Flow sub-component Sub-category for primary flow components. Sixteen sub-components: baseflow, low baseflows, annual flow, monthly flow, high flow/small flood, large flood, water level, drought-low flow, intermittency, hydroperiod, constancy, rise rate, range, daily variability. Flow category Y Categorizes flow sub-component in 7 broader categories for multivariate analyses: baseflows, low flows, flood-high, water level/ hydroperiod, constancy, rate of change, variability. Ecological category Y Primary ecological response group of central importance in analysis (see methods). Response type Y Type of ecological response within each group (see methods). Group Indicates whether analyses were based on individual species, genera, families, guilds, or assemblages. Flow-eco relationship Y Indicates whether flow-ecology relationship was direct or indirect. Other factors complicate flow-ecology relationships. If they were provided by authors, they were indicated (e.g., sediment, temperature, salinity, nutrients). Curve Indicates whether flow-ecology relationship was linear or curvelinear, if predictable relationship was presented in study. % flow alteration Y The percent alteration in the flow component. Response variable % ecological response Y The percent alteration in the ecological response. 8 Southeastern Naturalist Vol. 12, No. 5 number of sites. We considered the period of record (POR) as the entire time in years in which hydrologic records were needed to formulate ecological relationships, not the time required to collect ecological information. In general, the hydrologic POR directly overlapped with the time period of ecological data collection. However, in some instances, a longer hydrologic record was used to generate relationships in which ecological data were collected within a shorter time span. We considered sites as areas representing independent samples (replicates) in spatial studies or as areas used to develop independent flow-ecology relationships. However, in the case of IFIM studies, although flow-ecology data is collected from multiple reaches, the data is typically compiled to generate composite flow-ecology relationships for a single river system; thus, these studies were represented by only one site. We also recorded whether each study provided a qualitative or quantitative flow-ecology relationship. The direction of the change in flow and ecological responses primarily depended on how the author(s) presented the results. For example, authors reporting a positive correlation between fish abundance and monthly flow resulted in both positive changes in ecological response and flow in our dataset. Although theoretically the same relationship could be used to generate negative changes in both flow and ecological responses, we only present the direction presented by the authors. The directionality of some ecological responses, such as abundance and richness, seem fairly straightforward to interpret. However, changes in community composition, organismal behavior, and organic matter responses are less intuitive. Generally, we interpreted alterations in community composition as positive or negative if native, endemic, or flowspecialists increased or decreased, respectively. Alterations in general habitat area, availability, or suitability due to some disturbance use were considered negative responses. We relied heavily on author interpretation to distinguish negative and positive behavior responses. We separately reviewed the database to ensure interpretation was as consistent as possible and to validate findings. Increases in dispersal or migration, if indicated beneficial to populations by authors, were considered positive. In contrast, forced movements due to disturbance (e.g., hydroelectric peaking) were considered negative. Increases in nutrients and organic matter export were considered positive, whereas decreases were considered negative. Increases in organic matter decomposition rates and increases in nutritional quality (e.g., increases in nitrogen or phosphorus content) were considered positive. Although our qualitative analysis supported behavioral and nutrients studies, we do not include any results from these studies in our quantitative analyses. Similar to Poff and Zimmerman (2010), for studies reporting quantitative relationships, we recorded changes in flow and ecological responses as percent changes. All flow-ecology relationships represented in each study were recorded as long as separate flow components or separate ecological categories were being evaluated. Studies frequently had more than one quantitative relationship. For anthropogenic flow alterations, we presented % changes as alterations from pre-disturbance (temporal) or reference conditions (spatial). For example, 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 9 increases in flow magnitudes due to hydroelectric peaking operations were presented as positive % changes from baseflow to peak flow magnitude, assuming that most flows in reference streams do not exhibit daily fluctuations of this magnitude and rate of occurrence. For changes in natural flow, we presented % changes as natural linear relationships, changes from median conditions, or changes from conditions prior to an acute hydrologic event. For example, if authors presented changes in flow magnitude as a result from the occurrence of a flood, we recorded positive % changes in flow from conditions prior to the occurrence of the flood. As another example, losses in flow magnitude due to droughts were recorded as negative % changes from flow magnitude during non-drought years. In contrast to acute hydrologic events, natural flow variation may occur along inter- and intra-annual scales, such as analysis within the same river system, or occur across spatial scales, such as differences in climate among basins. The majority of these studies were presented as linear relationships between gradients of flow and ecology. Because of the complexity of interpreting curvilinear trends, we did not include these results in the quantitative analysis. We calculated % changes in flow and ecological variables across the entire range of measured values and in the direction presented by the authors. For example, Rogers et al. (2005) reported positive linear relationships between minimum river stage and fish abundance in the Ocklawaha River, FL. In this case, % increase in river stage and % increase in fish abundance was calculated from the minimum to maximum values presented along the regressed line. In some cases, differences in natural flow were reported across different periods or different locations. For example, Lorenz (1999) assessed variation in hydroperiod magnitude and fish assemblage biomass across several sites in Florida Everglades sloughs. Percent changes in hydroperiod magnitude (e.g., water depth) and fish biomass were presented as changes from values at sites considered optimal, i.e., higher ecological values and associated hydrologic values, to values considered the least optimal. Although these studies are expected to present variable changes in flow and consequently, ecological responses, they still may provide results informative to generalizing flow-ecology relationships. Our interpretation of natural flow variation attempted to provide comparable results to that of anthropogenic flow alterations. For example, in the case of Lorenz (1999), we could determine whether anthropogenic reductions in hydroperiod magnitude and associated fish responses emulate patterns observed from natural variation. Qualitative analysis A central task of this paper was to summarize the impacts of changes in flow on ecology (positive or negative responses), according to different sources of change (e.g., withdrawals, flooding). We partitioned studies by source of change in flows and then summarized how each primary flow component was altered and any associated ecological responses. We compared anthropogenic-induced hydrologic disturbances to natural variation to determine if there was any evidence of overlap in ecological responses. 10 Southeastern Naturalist Vol. 12, No. 5 Quantitative and statistical analysis Another central purpose of this study was to provide quantitative relationships and to determine if flow-ecology relationships were similar with regard to similar changes in hydrology, regardless of whether the cause of change was from anthropogenic or natural sources. We expanded all flow-ecology relationships into separate data entries despite some being found within the same study (as long as each data entry represented a unique flow-ecology relationship). We tested whether % ecological responses were different among different sources of flow change, different study scales (e.g., spatial, temporal, spatio-temporal), and study hydrologic resolution (e.g., annual, monthly, daily, sub-daily) using Kruskal-Wallis tests followed up by post-hoc comparisons using a Tukey’s test (alpha = 0.05 significance level). We created separate plots of % ecological response versus % change in flow according to different primary flow components: magnitude, frequency, and duration. There were insufficient data to consider timing and rate of change. Prior to plotting results and statistical procedures, we transformed the data by taking the log(x + 1) for the absolute value in % alterations and % responses and then retaining the sign from the original value. However, we provide untransformed axis values for ease of interpretation. Analyses were separated into ecological responses to natural flow variation relative to anthropogenic flow alterations. We further explored relationships between fish metrics and flow magnitude because of sufficient data. For fish responses to natural changes in flow magnitude, we separated responses by physiographic province (i.e., Blue Ridge, Coastal Plain, Piedmont, and Ridge and Valley; Fenneman and Johnson 1946). Because of the limited number of natural flow studies in the Blue Ridge and Ridge and Valley provinces, we created a composite category as “Mountains”. We also stratified fish responses to natural flow by separating plots by flow categories (i.e., baseflows, low flows, water level, and variability). For fish responses to anthropogenic changes in flow magnitude, we observed that fish responses segregated into only two quadrats, leaving unclear patterns across the entire plotted space. Thus, we separated anthropogenic flow-fish relationships by quadrat and plotted responses according to the source of flow change. In addition to fish, organic matter had sufficient data to evaluate responses to changes in natural flow. Reservoir operation improvement studies were also evaluated in a separate analysis. We tested individual relationships using Spearman’s rank correlations. We questioned the relative importance of various factors, such as basin size, province, flow components, or type of hydrologic alteration, in determining percent ecological responses. We used flow categories as potential explanatory variables rather than flow sub-components. We focused solely on fish responses because they had a sufficient sample size for a multivariate analysis. We imported the % responses along with the variables (see Supplemental File 1, available online at https://www.eaglehill.us/SENAonline/ suppl-files/mon5-1123-McManamay-s1, and, for BioOne subscribers, at http://dx.doi.org/10.1656/S1123.s1) into the program R (ISM 2011) and developed regression trees using the rpart package (Therneau et al. 2011). The 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 11 rpart package uses recursive partitioning, which involves splitting the data on each sequential node using variables that maximize the between-groups sum-of-squares or maximizes the reduction in the total sum-of-squares. We used a parametric analysis-of-variance procedure since % responses were transformed and encompassed negative and positive values greater than 1. This procedure continues throughout subsequent nodes until the subgroups reach a specified minimal size or no further splits can be made (Breiman et al. 1998, Therneau and Atkinson 1997). Trees could become very complex; thus, the second step involves a pruning procedure that minimizes the cross validation error while also minimizing the mean-square error (increasing tree size or complexity). Subsets of the dataset are retained and used to “test” the tree to calculate an average cross-validation error (x-val error) across all nodes. The x-val error is then compared to the cost complexity factor (cp), a parameter that takes into account the residuals from the sum-of-squares of the tree in relation to tree size (number of nodes). The tree is pruned at the cp value that minimizes the x-val error within 1 SE (Breiman et al. 1998, Therneau and Atkinson 1997). Results A total of 186 sources were isolated that provided a qualitative or quantitative relationship between flow and ecology within states encompassing the SAR (Fig. 1). Studies had fairly broad geographic representation. Of the sources, 128 were peer-reviewed published articles, 55 were grey literature, and 3 were unpublished data. Fish were the predominant ecological category of interest in most studies (125), followed by macroinvertebrates (33) and riparian vegetation (16) (Table 2). In general, within most ecological categories, more sources were peer-reviewed literature than grey literature. Fish were the only ecological category with sources of unpublished data (3). The majority of studies were conducted or were focused within the Coastal Plain Province (114), followed by the Piedmont (41), Blue Ridge (13), and Ridge and Valley Table 2. Number of peer-reviewed, grey-literature, and unpublished studies within different ecological categories. Multiple ecological categories could be represented within each study. Ecological category Total Peer-reviewed Grey Unpublished Fish 125 78 44 3 Macroinvertebrate 33 24 9 - Riparian 16 15 1 - Organic 10 9 1 - Nutrient 9 5 4 - Algae 6 5 1 - Ecosystem GPP 4 2 2 - Macrophyte 4 1 3 - Mammal 2 - 2 - Bird 1 - 1 - Grand total 186 128 55 3 12 Southeastern Naturalist Vol. 12, No. 5 Provinces (7). Five of the studies were mesocosm experiments. Over 50% of studies used hydrologic information collected within 5 years or less, whereas over 80% used hydrologic information from 20 years or less. Over 50% of studies were conducted at only one site, and 80% of studies were collected at 5 or less sites. There was a higher number of studies that documented ecological Figure 1. Study sites in the South Atlantic Region within different physiographic provinces (Fenneman and Johnson 1946) and according to different types of sources. Each dot represents each study site or the approximate centroid of multi -site analyses. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 13 responses to natural flow variation (96) compared to studies documenting ecological responses to anthropogenic flow alterations (76) or reservoir operation improvements (e.g., flow restoration) (12). The majority of studies evaluated the effects of changes in flow magnitude, followed by duration and frequency. Fewer studies evaluated the influence of rate of change in flow or flow timing. Overall, we isolated 16 different flow sub-component categories (Table 1, see also variable descriptions in Supplemental File 1, available online at https://www.eaglehill.us/SENAonline/supplfiles/ mon5-1123-McManamay-s1, and, for BioOne subscribers, at http://dx.doi. org/10.1656/S1123.s1). High flows, flooding, average flow conditions (monthly, annual, baseflows), and low baseflows had the highest frequency of quantitative flow-ecology relationships. We created 7 broader composite flow categories (Table 1) to simplify flow sub-components and to use in multivaria te models. Qualitative analysis Studies evaluating relationships between natural flow variation and ecology showed a variety of responses (Fig. 2; see Supplemental File 2, available online at https://www.eaglehill.us/SENAonline/suppl-files/mon5-1123-McManamay-s2, and, for BioOne subscribers, at http://dx.doi.org/10.1656/S1123.s2). Changes in flow and ecological responses in drought and flood studies typically represented changes from baseline conditions along temporal scales (e.g., drought versus non-drought years). Because these studies represent acute changes in flow due to some disturbance, their results tend to be more intuitive. Interpreting changes in flow and ecological responses in response to general flow variation, however, becomes less clear since these studies were conducted over various spatial and temporal scales. Within each natural flow variation type, the number of studies reporting positive responses was very similar to those reporting negative relationships, with the exception of drought studies and reservoir operation improvement studies (Fig. 2; see Supplemental File 2, available online at https://www.eaglehill.us/ SENAonline/suppl-files/mon5-1123-McManamay-s2, and, for BioOne subscribers, at http://dx.doi.org/10.1656/S1123.s2). Studies concerning droughts reported more negative responses, whereas reservoir improvement studies reported more positive responses. For ease of representation, we included reservoir improvement studies in the natural flow variation summary (Fig. 2). Fish, macroinvertebrates, and riparian ecological categories showed variable and non-consistent responses to changes in average, high, and low flows as well as variability. For example, increases and decreases in the magnitude of flows led to increases and decreases in fish abundance and riparian tree growth (see Supplemental File 2, available online at https://www.eaglehill.us/SENAonline/supplfiles/ mon5-1123-McManamay-s2, and, for BioOne subscribers, at http://dx.doi. org/10.1656/S1123.s2). In contrast, organic matter export and decomposition, as well as algal and primary production responses, showed consistent responses to hydrologic alterations. For example, decreases in magnitude consistently led to decreases in organic matter export and increases in algal abundance/production. 14 Southeastern Naturalist Vol. 12, No. 5 For anthropogenic flow alterations, ecological responses were more consistently negative for most categories (Fig. 3; see Supplemental File 2, available online at https://www.eaglehill.us/SENAonline/suppl-files/mon5-1123-McManamay-s2, and, for BioOne subscribers, at http://dx.doi.org/10.1656/S1123.s2). Fish consistently showed negative responses regardless of the type of anthropogenic flow alterations, whereas algae tended to respond responded positively. Although macroinvertebrates and riparian ecological categories showed variable responses, the majority of studies reported negative responses to anthropogenic flow alterations. Quantitative and statistical analysis A total of 109 studies out of the 186 included quantitative flow-ecology relationships. We expanded the entire database into individual, quantitative flowecology relationships from each study. The resultant database totaled 213 unique flow-ecology relationships for all ecological categories combined, 140 of which were for fish. Figure 2. The number of negative (black) and positive (white) ecological responses to changes in flow within natural flow categories and reservoir operation improvement studies. No general patterns were apparent except the predominance of negative responses to drought. Each study could potentially report both positive and negative relationships; thus, the sum of positive and negative relationships may be greater than the total number of papers (see Supplemental File 2, available online at https://www.eaglehill.us/SENAonline/ suppl-files/mon5-1123-McManamay-s2, and, for BioOne subscribers, at http:// dx.doi.org/10.1656/S1123.s2 for more detailed information). 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 15 In general, ecological responses to natural flow variation were highly variable whereas ecological responses to anthropogenic flow alterations were predominately negative, regardless of the direction of flow change and the type of ecological response. Ecological responses were significantly different among sources of flow change (c2 = 25.14, df = 8, P = 0.0015). Ecological responses to floods, natural variation, and improving reservoir operations were significantly higher than ecological responses to withdrawals and reservoir operations, whereas they were not significant from other categories (Tukey’s test: P <0.05). Multiple comparisons among all other sources of flow change were not significantly different. Ecological responses were not significantly different among different study scales (c2 = 0.603, df = 2, P = 0.7399) or among different hydrologic resolutions (c2 = 3.036, df = 3, P = 0.3862). When considering all ecological categories, responses to natural changes in flow magnitude were highly variable (Fig. 4). However, once we placed fish responses into various contexts, patterns emerged, which explained more varia- Figure 3. The number of negative (black) and positive (white) ecological responses to types of anthropopogenic flow alterations. Responses for most ecological categories were predominately negative with the exception of algae. Each study could potentially report both positive and negative relationships; thus, the sum of positive and negative relationships may be greater than the total number of papers (see Supplemental File 2, available online at https://www.eaglehill.us/SENAonline/suppl-files/mon5-1123-McManamay-s2, and, for BioOne subscribers, at http://dx.doi.org/10.1656/S1123.s2 for more detailed information). 16 Southeastern Naturalist Vol. 12, No. 5 Figure 4. Relationships between % changes in flow magnitude and % ecological changes for natural flow variation (top) and anthropogenic flow alterations (bottom) according to different ecological groups. Ecological responses to natural flow variation were highly variable; however, responses to anthropogenic flow alterations were predominately negative. Lower-left quadrat (A) and bottom-right quadrat (B) refer to negative % ecological responses to decreases and increases in flow magnitude, respectively and are specifically addressed in Figure 7. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 17 tion. Fish exhibited significant positive correlations with natural flow magnitude within the coastal plain but showed no significant pattern within the mountain (BlueRidge, Ridge and Valley) or Piedmont provinces (Fig. 5A, B). After Figure 5. Positive correlation between natural % changes in flow magnitudes and % fish responses for sites occurring in the coastal plain (A). No correlation was observed for sites occurring in the upland provinces (Mountains, i.e., Blue Ridge and Ridge and Valley combined , and Piedmont) (B). ρ represents Spearman’s rank coefficients followed by P-values for statistical significance. 18 Southeastern Naturalist Vol. 12, No. 5 Figure 6. Relationships between % changes in the magnitude of various flow categories (A–D) versus % changes in fish responses according to different response types. ρ represents Spearman’s rank coefficients followed by P values for statistical significance. stratifying data by flow categories, fish had significant positive correlations with natural increases in moderate flow conditions (e.g., baseflow, monthly flow) and marginally significant positive correlations with natural increases in water levels (gage height) (Fig. 6A, B). Conversely, fish exhibited a significant negative correlation with increases in flow variation (e.g., range in flows, coefficient of variation; Fig. 6D). However, fish showed non-significant responses to natural decreases in low-flow conditions (Fig. 6C). In contrast to natural flows, most ecological responses to anthropogenic flow alterations were negative (Fig. 4). Fish consistently responded negatively to anthropogenic flow alterations. However, algae and macrophytes responded positively to decreases in flow magnitude due to anthropogenic sources. Although responses were at times unpredictable, macroinvertebrates typically responded negatively to anthropogenic changes in flow magnitude. Riparian vegetation responses were more variable. After separating fish responses by quadrat in Figure 4, we observed that fish responses showed significant decreases with decreases in flow magnitude (Fig. 7A). However, fish responses to increases in flow magnitude were not significant (Fig. 7B). 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 19 Sample sizes for ecological responses to natural and anthropogenic changes in flow duration and flow frequency were not as large as that for flow magnitude, which precluded statistical analyses. However, we observed the following apparent trends in the data. Fish responded negatively to both natural and anthropogenic Figure 7. Negative fish responses stratified by anthropogenic losses (A) and increases (B) in flow magnitude. Graphs A and B refer to fish responses in Figure 4A and 4B. Fish responses decreased predictably with losses in flow magnitude (A) whereas no relationship was observed for increases in flow magnitude (B). ρ represents Spearman’s rank coefficients followed by P values for statistical significance. 20 Southeastern Naturalist Vol. 12, No. 5 Figure 8. Ecological responses to % changes in flow duration due to natural flow variation (top) and anthropogenic flow alterations (bottom). Fish displayed negative responses to changes in flow duration regardless of whether the source was natural or anthropogenic whereas responses by riparian vegetation and macroinvertebrates were variable. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 21 changes in flow duration (Fig. 8). Fish responses to changes in flow frequency were negative for anthropogenic sources and variable to natural sources (Fig. 9). Figure 9. Ecological responses to % changes in flow frequency due to natural flow variation (top, fish only) and anthropogenic flow alterations (bottom). Fish responses were variable to natural changes in flow and consistently negative to anthropogenic changes in flow. 22 Southeastern Naturalist Vol. 12, No. 5 Again, macroinvertebrates and riparian vegetation showed variable responses to changes in flow duration or frequency, regardless of whether the source was anthropogenic or natural (Figs. 8, 9). Organic matter export exhibited a significant positive correlation with natural changes in flow associated with baseflow magnitudes, flooding magnitudes, and flooding frequency, whereas organic decomposition and nutritional content did not show any pattern (Fig. 10). Reservoir improvement studies typically (7 of 8 responses) reported increases in flow magnitude, which generally led to positive fish responses and inconclusive responses for macroinvertebrates; however, there was no significant correlation for fish (Fig. 1 1). We used regression trees to determine the relative importance of predictor variables in determining the direction and magnitude of % ecological responses for fish (n = 140). Cross-validation error minimized at a tree size of 5 branches. The tree for fish explained 57% of the variation in % ecological responses and included the source of flow change, % change in flow, flow category, and drainage area size as predictor variables (Fig. 12). Anthropogenic flow alterations and flooding led to negative fish responses, whereas drought, reservoir operation improvements, and natural flow variation led to variable responses (Fig. 12). Changes in flow less than 19.5% led to negative fish responses within intermediate-sized river systems but more neutral fish responses in small and large river systems. Changes in average flow conditions (baseflow, monthly Figure 10. Positive relationship between percent changes in organic matter export, decomposition, and nutrient composition in relation to % changes in flood-frequency (FF), flood-duration (DF), flood magnitude (MF), and baseflow magntiude (MBFL). 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 23 Figure 11. Percent ecological responses to % changes in flow magnitude due to reservoir operation improvements. ρ represents Spearman’s rank coefficients followed by P values for statistical significance. Figure 12. Regression tree predicting % fish responses to natural and anthropogenicinduced flow alterations using primary splitting variables (in box directly below each node). Categories or values for each splitting variable are represented along branches. Top value in each circle or box represents the average % ecological response at that specific node or branch, respectively, whereas the bottom number represents the sample size. 24 Southeastern Naturalist Vol. 12, No. 5 flows, water levels) greater than 19.5% led to positive fish responses, whereas 19.5% or greater increases in fall rates, high flows, and ranges in flow led to negative fish responses. Discussion From our examination of responses to flow variation in 186 separate studies in the SAR, we found that ecological responses to flow changes were primarily dependent upon the source of change and the ecological category of interest. Furthermore, over-generalized flow-ecology relationships, at least with regard to primary flow components, were not supported unless relationships were placed in appropriate contexts. For example, fish responses to natural changes in primary flow components were highly variable and inconsistent; however, once responses were stratified by flow categories and by geomorphic context, quantitative and predictable relationships could be extracted (e.g., Figs. 5, 6). In addition, our multivariate analysis using fish responses suggested that if sufficient meta-data are included in database compilations, the predictive capabilities of these models can be dramatically increased. Qualitative analysis The evidence of common, directional (positive or negative) flow-ecology relationships among studies in natural and anthropogenic-modified systems, if present, would suggest that the results of flow-ecology studies can be used interchangeably in supporting environmental flow standards, regardless of temporal, spatial, or disturbance contexts. For example, losses in flow magnitude, whether anthropogenic or naturally induced, led to predictable decreases in fish metrics (abundance, diversity, and habitat) and increases in algal abundance. However, the majority of ecological responses to natural changes in flow were highly different than those to anthropogenic flow alterations, despite similar directional changes in primary flow components. Most meta-analyses in ecology, including ours, are designed to generalize ecological responses, and thus have limitations. Determining what constitutes a negative or positive response when considering response types or the focal scale of the study (e.g., community, population, or individual) can be subjective and influenced by interpretation. Although we attempted to be sensitive to how the authors interpreted the ecological response, our results highlight types of ecological categories that may be more prone to judgment-related uncertainty. Generalizing studies conducted across various spatial and temporal scales can also lead to ambiguous results. For example, the response of fish abundance to changes in flow magnitude across sites with varying daily flow averages may be very different compared to a single site with varying hourly flows due to hydroelectric generation. Natural flow variation. The majority of studies reported negative and positive ecological responses to natural flow variation, regardless of similarities in the direction of changes in flow. One exception was droughts, 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 25 which consisted of decreased magnitudes and high-flow frequencies, and in turn, negative ecological responses, such as decreased fish and macroinvertebrate metrics and alterations to trophic community structure. However, Grossman et al. (1998) and Grossman et al. (2010) reported increases in fish abundance and diversity during drought because stream conditions in a North Carolina mountain stream became more suitable to “drought immigrants”. In addition, Averett et al. (2004) found that lower flows during drought conditions increased algal abundance and habitat. In contrast to droughts, ecological responses to flooding were less predictable. For example, a 500-yr flood event and associated debris-flows resulted in, on average, 76% declines in Brook Trout populations in 3 mountainous river systems in Virginia (Smith and Atkinson 1993). However, increases in the magnitude of flooding in the coastal plain streams were associated with increased fish diversity (Strong and Nagid 2006). Floods typically have many beneficial ecological responses, such as increased organic matter export (Atkinson et al. 2009) and increases in riparian seed germination (Pierce and King 2007); however, the short- and long-term ecological responses may depend on whether floods occur as habitat creation or destruction. In rivers systems draining mountainous or piedmont provinces, floods may be more disturbance-related, whereas in coastal plain rivers, floodplain inundation occurs for major portions of the year and provides important habitats and refuge for a host of fish, macroinvertebrates, plants, and wildlife (Light et al. 1998, Pearsall et al. 2005, Rehage and Loftus 2007, Walsh et al. 2009). Thus, additional information, such as geomorphic structure of river channels (e.g., confined versus unconfined), is essential to predicting ecological responses to natural flow variation across broad regions (McCargo and Peterson 2010). Anthropogenic flow alterations. In comparison to natural flows, ecological responses to anthropogenic flow alterations were consistently negative. Withdrawals were predominately associated with losses in flow magnitude, and although most ecological responses were negative, there was some degree of variability. Reservoir operations caused many more negative ecological responses than any other source of hydrologic alteration. The ecological consequences of dam operations are well documented within the SAR (Freeman et al. 2001, Hupp et al. 2009, Travincheck and Maceina 1994, West et al. 1988). Fish tend to respond predictably to anthropogenic flow alterations, whereas riparian and macroinvertebrate responses tend to be inconsistent (Poff and Zimmerman 2010). The majority of studies that evaluated the effects of urbanization- induced flow alterations on fish reported negative associations (Adams et al. 2009, Helms et al. 2009a, Roy et al. 2005). In contrast, reported macroinvertebrate responses to urbanization were, at times, positive (Helms et al. 2009b), and the algal responses were typically positive (Taulbee et al. 2009). Interestingly, Brown et al. (2009) conducted a multi-regional study on the ecological effects of urbanization-induced hydrologic alterations and reported that macroinvertebrates, as opposed to fish and algae, showed the most consistent 26 Southeastern Naturalist Vol. 12, No. 5 negative associations. Quantitative and statistical analysis The advantage of quantitative analyses is that they can provide more digestible pieces of information, stratify analyses into appropriate contexts, and finally, organize predictions into quantitative predictions. Prior to segregating studies into context-specific analyses, the results of our quantitative and statistical analysis reiterated the results of the qualitative analysis: ecological responses to natural flow variation were highly variable, whereas responses to anthropogenic flow alterations were predominantly negative. Natural flow variation. Once we stratified fish responses to natural flow variation by flow category and by province, predictable patterns emerged that provided opportunities for statistical analysis. Fish showed positive correlations with increases in natural flow magnitudes in the coastal plain, whereas no relationship was apparent in upland areas. Geomorphology may influence how river communities respond to changes in flow magnitudes. For example, simply organizing stream reaches on the basis of their ability to migrate may increase the predictive accuracy of assessing ecological responses to flow variation (Liermann et al. 2011, McCargo and Peterson 2010). Streams within the Virginia piedmont commonly show parabolic relationships between Micropterus dolomieu Lacepède (Smallmouth Bass) recruitment and spring/ summer flow magnitudes (Copeland et al. 2006, Smith et al. 2005), which suggests low and high flows during spawning and young-of-year growth periods in these systems may be detrimental to populations. In contrast, multiple sources report positive linear relationships between fish abundance and flow magnitudes in coastal environments for estuarine fish (Greenwood et al. 2008, Peebles 2005), Morone saxatilis Walbaum (Striped Bass; James Bulak, South Carolina Department of Natural Resources, Eastover, SC, unpubl. data), and Micropterus salmoides Lacepède (Largemouth Bass; Bob Greenlee, Virginia Department of Game and Inland Fisheries, Suffolk, VA, unpubl. data). Thus, the occurrence of high flows or flooding in an unconstrained coastal plain stream may be less of a disturbance to river communities than in a floodplainconstrained upland stream. Interestingly, fish responses were positively correlated with changes in average flow magnitudes and water levels, but negatively correlated with increases in flow variation. We cannot say with certainty that this is a trend applicable to all systems and all fish species within the region. However, the various types of responses (e.g., abundance, growth, diversity) in these analyses were fairly diverse, which indicates our results are doubtfully influenced by certain taxa. Anthropogenic flow alterations. Fish had significant, negative responses to reductions in flow magnitude as a result of withdrawals or reservoir operations. These studies suggest that losses in fish habitat accompany losses in flow magnitude, and the relationship can be predicted with some certainty. Given 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 27 these results, we predicted fish responses would be inversely correlated to anthropogenic increases in flow magnitude; however, no trend was observed. Unlike decreased flows, increases in flow magnitude may not directly translate to losses in habitat, especially due to reservoir operations. For example, increases in magnitude may result from increased daily discharges from dams during hydroelectric generation or flow inflation from seasonal storage. In a nationwide assessment, Carlisle et al. (2010) found that predicting ecological impairment or fish life-history traits from flow inflation was more difficult compared to that of decreasing flow magnitudes. Fish and organic matter responded consistently with alterations in some primary flow components, regardless of the source of flow change. For example, fish responded fairly consistently to changes in the duration of flows, whereas organic matter responded consistently to changes in the magnitude, frequency, and duration of flows. Increases or decreases the in the frequency, duration, or magnitude of flows generally led to increases or decreases in organic matter export and decomposition, respectively (Battle and Golladay 2001, Cuffney and Wallace 1989, Wallace et al. 1995). Reservoir improvement studies. Most responses of fish to reservoir operation improvements were positive, whereas insufficient data existed to form conclusions for macroinvertebrates or algal communities. Of the seven reservoir improvement studies with quantitative information, five were associated with increases in minimum flows. The degree of ecological benefits associated with increased minimum flows will most likely depend on the extent of disturbance conditions prior to restoration or the prevalence of other limiting factors. For example, Travnicheck et al. (1995) reported a 138% increase in fish species richness (8 to 19 species) following a 340% increase in minimum flows below a dam. In contrast, Bednarek and Hart (2005) found that the effect of increases in minimum flows below dams on macroinvertebrate richness, although positive, was more effective if used in conjunction with increased dissolved oxygen levels. We also found one study documenting the use of pulsed releases from a hydroeletric facility to control nuisance periphyton growth in a regulated river (Flinders and Hart 2009). Similar to the other quantitative analyses, results of the regression tree suggested that the source of flow change was the most important predictor of the magnitude of fish responses. The combined results from our analyses suggest that fish tend to respond negatively to changes in the extremities of the hydrograph. For example, fish responded negatively to increases in flood magnitude/frequency, high flows, fall rates, and ranges whereas responses to increases in average conditions (baseflows, hydroperiod, monthly flows) were typically positive (e.g., Fig. 6 and 12). Flow categories are summaries of combinations of primary flow components; thus, we expected that flow categories would explain more variation in predictive models. For example, low flows suggest not only a direction in magnitude but also duration whereas flooding suggests high magnitudes and a frequency for particular river systems (assuming flooding is predictable). The 28 Southeastern Naturalist Vol. 12, No. 5 regression tree also suggested that the size of a particular basin may influence fish responses, depending on the degree of flow change. It is important to note that the regression tree only explained 57% of the overall variation in fish responses. Obviously , many factors, such as sediment and temperature, may play a role in influencing the nature of ecological responses and may confound the effects of flow or interact with flow to influence ecology. A total of 22% of the studies in our analysis mentioned other variables that may interact with flow to influence ecological responses. For example, losses in flow magnitude in coastal rivers were many times associated with encroachment of salt wedges or the alteration of salinity levels in estuaries, both of which affected fish populations (Bachman and Rand 2008, Greenwood et al. 2008, Van Den Avyle and Maynard 1994). We also found evidence that relationships between flow and fish were complicated by flow interactions with sediment (Roy et al. 2005, Smith and Atkinson 1999) and temperature (Baker and Jennings 2005, Krause et al. 2005). Although we included the nature of flow-ecology relationships (direct or indirect) as a variable in our analyses, it was excluded as an important predictor. Implications for future studies and environmental flow standards Documenting the mechanisms responsible for structuring ecological patterns in relation to hydrologic variability has become increasingly important to developing water policy standards at regional scales (Poff et al. 2010). Unfortunately, analyses at smaller regional scales may be constrained due to insufficient overlapping hydrologic and ecological data (Knight et al. 2008). Because sufficient quantitative evidence may be unavailable, or only available for individual ecological categories, developing mechanisms to efficiently use the best available information may be required to support decision making (Sullivan et al. 2006). Of the 186 sources within our study, approximately 30% were gathered from grey literature. We find this extremely important since the grey literature might be often overlooked as a legitimate source of data. Adopting environmental flow standards are typically executed through regional or statewide agencies and water-supply planning offices, who are informed through agency and consultant reports, i.e., grey literature. As one of many examples, Florida statewide law and water policy, as documented in the Florida Statutes (Chapter 373.042), requires that regional water-supply planning offices develop minimum flow levels (MFLs) for specified water bodies (TFS 2011). Personnel in the regional office or outside agencies are then contracted to conduct technical field analyses, suggest sufficient MFL levels, and produce technical reports, which are rigorously peer-reviewed by a local water supply district panel for legitimacy as acceptable research. In essence, grey literature can be a valuable resource that may inform water policy at regional scales. Analytical approaches may fail to maximize information resources at regional levels because of strict search criteria (e.g., only sources with quantitative information available). For example, previous studies have suggested there is insufficient quantitative information to document generalized relationships between 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 29 altered flow and ecology at continental (Loyd et al. 2003) and global scales (Poff et al. 2010). In addition, analytical approaches may underrepresent complex ecological dynamics by measuring ecological responses as negative/positive changes or as percent changes. For example, a loss in river flows may lead to an increase in riparian coverage through encroachment, i.e., positive response (Brandt et al. 2000); however, alterations in riparian taxa abundance from changes in flow are also likely to alter community structure, i.e., negative response (Burke et al. 2003). In response to the limiting and potentially confusing information, one approach could be to organize responses as supportive or harmful to ecological goods and services (Daily et al. 1997). An alternative approach is the “eco evidence” method, which uses weighted evidence from literature to support a series of a priori hypotheses about a cause-effect relationship (Norris et al. 2012). Because cause-effect relationships may be limited within the ecological literature, eco evidence approaches are convenient in that they provide strength of evidence without diminishing the complexity of underlying ecological rel ationships. We suggest that in order for flow-ecology relationships to become relevant to management and reliable in creating environmental flow standards, analyses should either be stratified into appropriate contexts or include other associated predictors in multivariate models. Armstrong et al. (2011) increased the accuracy of models predicting fish responses to flow alteration by adding % impervious surfaces as a secondary variable. Despite great variation in the direction and magnitude of changes in flow, anthropogenic flow alterations within our analysis led to ecological losses. Ecological relationships with natural flow variation were highly variable; however, the inclusion of additional explanatory variables increased model predictive capacity. We found that as little as a 10% change in flow can result in very large ecological responses. We observed common directional responses to changes in flow by particular ecological groups, regardless of the type of response or whether the source was natural or anthropogenic. Fish, algae, and organic matter responses to changes in flow were consistent and thus, may be good candidates for future meta-analyses. For example, losses in flow magnitude, anthropogenic or naturally induced, led to predictable decreases in fish metrics and increases in algal abundance. Similarly, relationships between organic matter decomposition and flow magnitude/freqency were predictable. Our results also suggest that acute extremities in natural hydrographs, such as droughts and flooding, show the most altered and predictable ecological responses; thus, anthropogenic alterations that mimic these types of flow changes, such as withdrawals and reservoir operations, may induce the most extensive and predictable ecological impacts. We recommend that great care should be taken in interpreting ecological responses, such as abundance measures and community composition, since these factors may be highly influenced by subjectivity. Primary flow components (e.g., magnitude, frequency) were relatively unimportant in predicting most ecological responses in our analyses. Intuitively, these results may not come as a surprise; however, primary flow components 30 Southeastern Naturalist Vol. 12, No. 5 have proved to be an efficient organizational template of hydrologic indices and ecological responses (Bunn and Arthington 2002, Olden and Poff 2003, Poff et al. 1997). More descriptive hydrologic units, such as flow categories, may provide more relevance in forming predictive flow-ecology relationships. For example, environmental flow components (EFCs) provide an additional level of detail (low, high, variability) to hydrologic metrics, which has increased their utility in environmental flow settings and determining the extent of hydrologic alteration (Mathews and Richter 2007). Although we attempt to provide an overview of results associated with our meta-analysis, we view the database compilation as an ongoing iterative process where hypotheses are updated as new data is acquired. More research is needed to isolate quantitative relationships, refine existing hypotheses, and generate new hypotheses through future literature searches, meta-analyses, regional and basin-specific studies, as well as mesocosm studies. Approximately one half of our studies in our database were conducted at only one site. In addition, approximately one half were conducted in five years or less. Thus, we conclude that there is a considerable need for short-term studies conducted at small spatial scales, i.e., case studies, to provide important information relevant to developing environmental flow standards. Acknowledgments This project was funded by the South Atlantic Landscape Conservation Cooperative and was managed by the Southeastern Aquatic Resources Partnership (SARP) and the Southern Instream Flow Network, a program within SARP. We thank Scott Robinson for providing technical support through SARP and Marilyn O’Leary for producing email and internet notices for data requests. We thank the following individuals for contributing data, publications, or references to incorporate in the literature review: Brian Alford, James Bulak, Chuck Cichra, Jason Dockendorf, Todd Ewing, Mary Freeman, Bob Greenlee, Chris Goodreau, Gary Grossman, Paul Johnson, Paul Leonard, Graham Lewis, Scott Longing, Eric Nagid, Jeff Powell, Jeff Quinn, Bobby Reed, Roger Rulifson, Scott Smith, Christopher Taylor, Theresa Thom, Jack Webster, and Jerry Ziewitz. Literature Cited Adams, A.J., R.K. Wolfe, and C.A. Layman. 2009. Preliminary examination of how human-driven freshwater flow alteration affects trophic ecology of juvenile Snook (Centropomus undecimalis) in estuarine creeks. Estuaries and Coasts 32:819–828. Allan, J.D. 1995. Stream Ecology: Structure and Function of Running Waters. Kluwer Academic Publishers, Boston, MA. Annear, T., I. Chisholm, H. Beecher, A. Locke, P. Aarrestad, C. Coomer, C. Estes, J. Hunt, R. Jacobson, G. Jobsis, J. Kauffman, J. Marshall, K. Mayes, G. Smith, R. Wentworth, and C. Stalnaker. 2004. Instream Flows for Riverine Resource Stewardship. Revised Edition. Instream Flow Council, Cheyenne, WY. Armstrong, D.S., T.A. Richards, and S.B. Levin. 2011. Factors influencing riverine fish assemblages in Massachusetts. US Geological Survey Scientific-Investigations Report 2011-5193. Available online at http://pubs.gov/sir/2011/5193/. Accessed 29 March 2013. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 31 Arthington A.H., S.E. Bunn, N.L. Poff, and R.J. Naiman. 2006. The challenge of providing environmental flow rules to sustain river systems. Ecological Applications 16:1311–1318. Atkinson, C.L., S.W. Golladay, S.P. Opsahl, and A.P. Covich. 2009. Stream discharge and floodplain connections affect seston quality and stable isotopic signatures in a coastal plain stream. Journal of the North American Benthological Society 28:360–370. Averett, A.W., J. Persinger, M.D. Chan, J.L. Lozinski, and D.J. Orth. 2004. Stream habitat modeling to support water management decisions for the North Fork Shenandoah River, Virginia. Final Report to Northern Shenandoah Valley Regional Commission, North Fork Instream Flow Technical Advisory Committee. December 2004. Available online at http://fishwild.vt.edu/North_Fork_Shenandoah/final_report/NFSR%20 MIF%20Final%20Report%20VT.pdf. Accessed 10 February 2010. Bachmand, P.M., and G.M. Rand. 2008. Effects of salinity on native estuarine fish in South Florida. Ecotoxicology 17:591–597. Baker, T.L., and C.A. Jennings. 2005. Striped Bass survival in Lake Blackshear, Georgia during drought conditions: Implications for restoration efforts in Gulf of Mexico drainages. Environmental Biology of Fishes 72:73–84. Battle, J.M., and S.W. Golladay. 2001. Hydroperiod influence on breakdown of leaf litter in cypress-gum wetlands. American Midland Naturalist 146:128–145. Bednarek, A.T., and D.D. Hart. 2005. Modifying dam operations to restore rivers: Ecological responses to Tennessee River dam mitigation. Ecological Applications 15:997–1008. Brandt L.A., K.M, Portier, and W.M. Kitchens. 2000. Patterns of change in tree islands in Arthur R. Marshall Loxahatchee National Wildlife Refuge from 1950 to 1991. Wetlands 20:1–14. Breiman, L., J.H. Friedman, R.A. Olshen, and C.J Stone. 1998. Classication and Regression Trees. Chapman and Hall/CRC, New York, NY. Brown, L.R., T.F. Cuffney, J.F. Coles, F. Fitzpatrick, G. McMahon, J. Steuer, A.H. Bell, and J.T. May. 2009. Urban streams across the USA: Lessons learned from studies in 9 metropolitan areas. Journal of the North American Benthological Society 28:1051–1069. Bunn S.E., and A.H. Arthington. 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30:492–507. Burke, M.K., S.L. King, D. Gartner, and M.H. Eisenbies. 2003. Vegetation, soil, and flooding relationships in a blackwater floodplain forest. Wetlands 23:988–1002. Carlisle, D.M., D.M. Wolock, and M.R. Meador. 2010. Alteration of streamflow magnitudes and potential ecological consequences: A multi-regional assessment. Frontiers in Ecology and the Environment 9:264–270. Copeland, J.R., D.J. Orth, and G.C. Palmer. 2006. Smallmouth Bass management in the New River, Virginia: A case study of population trends with lessons learned. Proceedings of the Annual Conference of the Southeastern Association of the Fish and Wildlife Agencies 60:180–187. Cuffney, T.F., and J.B. Wallace. 1989. Discharge-export relationships in headwater streams: The influence of invertebrate manipulations and drought. Journal of the North American Benthological Society 8:331–341. 32 Southeastern Naturalist Vol. 12, No. 5 Daily, G.C., S. Alexander, P.R. Ehrlich, L. Goulder, J. Lubchenco, P.A. Matson, H.A. Mooney, S. Postel, S.H. Schneider, D. Tilman, and G.M. Woodwell. 1997. Ecosystem services: Benefits supplied to human societies by natural ecosystems. Issues in Ecology 2:1–16. Edwards, E.A., M. Bacteller, and O.E. Maughn. 1982. Habitat suitability index models: Slough Darter. US Fish and Wildlife Service. FWS/OBS-82/10.9. February 1982. Available online at http://www.nwrc.usgs.gov/wdb/pub/hsi/hsi-009.pdf. Accessed 29 March 2013. Environmental Protection Agency (EPA). 2010. Regional environmental monitoring and assessment program. Available online at http://www.epa.gov/emap/remap/index. html. Accessed 14 May 2012. Fenneman N.M., and D.W. Johnson. 1946. Physiographic divisions of the conterminous US. US Geological Survey special map series, scale 1:7,000,000. Available online at http://water.usgs.gov/GIS/metadata/usgswrd/XML/physio.xml. Accessed 23 June 2010. Flinders, C.A., and D.D. Hart. 2009. Effects of pulsed flows on nuisance periphyton growths in rivers: A mesocosm study. Rivers Research and Applications 1330:1320–1330. Freeman, M.C., Z.H. Bowen, K.D. Bovee, and E.R. Irwin. 2001. Flow and habitat effects on juvenile fish abundance in natural and altered flow regimes. Ecological Applications 11:179–190. Gordon, N.D., T.A. McMahon, B.L. Finlayson, C.J. Gippel, and R.J. Nathan. 2004. Stream Hydrology: An Introduction for Ecologists. 2nd Edition. John Wiley and Sons, Ltd., West Sussex, UK. Greenwood, M.F. D., E.B. Peebles, S.E. Burghart, T.C. MacDonald, R.E.J. Matheson, and R.H.J. McMichael. 2008. Freshwater inflow effects on fishes and invertebrates in the Chassahowitzka River and Estuary. Report to the Southwest Florida Water Management District. Brooksville, FL. Available online at http://www.swfwmd.state.fl.us/ projects/mfl/reports/Chass_Appendices-section10.pdf. Accessed 10 February 2010. Grossman, G.D., R.E. Ratajczak, M. Crawford, and M.C. Freeman. 1998. Assemblage organization in stream fishes: Effects of environmental variation and interspecific interactions. Ecological Monographs 68: 395–420. Grossman, G.D., R.E. Ratajczak, M.D. Farr, C.M. Wagner, and J.T. Petty. 2010. Why there are fewer fish upstream. American Fisheries Society Symposium 73:63–81. Helms, B.S., J.E. Schoonover, and J.W. Feminella. 2009a. Assessing influences of hydrology, physicochemistry, and habitat on stream fish assemblages across a changing landscape. Journal of the American Water Resources Association 45:157–169. Helms, B.S., J.E. Schoonover, and J.W. Feminella. 2009b. Seasonal variability of landuse impacts on macroinvertebrate assemblages in streams of western Georgia, USA. Journal of the North American Benthological Society 28:991–1006. Hornick, L.E., J.R. Webster, and E.F. Benfield. 1981. Periphyton production in an Appalachain Mountain trout stream. American Midland Naturalist 106:22–36. Hupp, C.R., A.R. Pierce, and G.B. Noe. 2009. Floodplain geomorphic processes and environmental impacts of human alteration along coastal plain rivers, USA. Wetlands 29:413–429. Institute for Statistics and Mathematics (ISM). 2011. The R Project for statistical computing. Available online at http://www.r-project.org/. Accessed 12 October 2011. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 33 Institute of Water Research (IWR). 2008. Michigan’s water withdrawal assessment tool. Available online at http://www.miwwat.org/. Accessed 14 May 2012. Jackson, C.R., and C.M. Pringle. 2010. Ecological benefits of reduced hydrologic connectivity in intensively developed landscapes. BioScience 60:37 –46. Knight, R.R., M.B. Gregory, and A.K. Wales. 2008. Relating streamflow characteristics to specialized insectivores in the Tennessee River Valley: A regional approach. Ecohydrology 1:394–407. Krause, C.W., T.J. Newcomb, and D.J. Orth. 2005. Thermal habitat assessment of alternative flow scenarios in a tailwater fishery. River Research and Applications 21:581–593. Lackey, R.T. 2007. Science, scientists, and policy advocacy. Conservation Biology 21:12–17. Liermann, C.A. R., J.D. Olden, T.J. Beechie, M.J. Kennard, P.B. Skidmore, C.P. Konrad, and H. Imaki. 2011. Hydrogeomorphic classification of Washington State rivers to support emerging environmental flow management strategies. River Research and Applications. Available online at DOI:10.1002/rra.1541. Light, H.M., M.R. Darst, and J.W. Grubbs. 1998. Aquatic habitats in relation to river flow in the Apalachicola River floodplain, Florida. US Geological Survey Professional Paper 1594. Reston, VA. Lloyd N., G. Quinn, M. Thoms, A. Arthington, B. Gawne, P. Humphries, and K. Walker. 2003. Does flow modification cause geomorphological and ecological response in rivers? A literature review from an Australian perspective. CRC for Freshwater Ecology. Technical Report 1 ⁄ 2004. ISBN 0-9751642-02. Available online at http://live.greeningaustralia. org.au/nativevegetation/pages/pdf/Authors%20L/13_Lloyd_et_al.pdf. Lorenz, J.T. 1999. The response of fishes to physiochemical changes in the mangroves of northeast Florida Bay. Estuaries 22:500–517. Mathews, R., Richter, B.D. 2007. Application of the indicators of hydrologic alteration software in environmental flow setting. Journal of the American Water Resources Association 436:1400–1413. McCargo, J., and J. Peterson. 2010. An evaluation of the influence of seasonal base flow and geomorphic stream characteristics on Coastal Plain stream fish assemblages. Transactions of the American Fisheries Society 139:29–48. Newbold, J.D., R.V. O’Neill, J.W. Elwood, and W. Van Winkle. 1982. Nutrient spiralling in streams: Implications for nutrient limitation and invertebrate activity. American Naturalist l20:628–652. Norris, R.H., J.A. Webb, S.J. Nichols, M.J. Stewardson, and E.T. Harrison. 2012. Analyzing cause and effect in enviromental assessments: Using weighted evidence from the literature. Freshwater Science 31:5–21. Olden J.D., and N.L. Poff. 2003. Redundancy and the choice of hydrologic indices for characterizing streamflow regimes. River Research and Applications 19:101–121. Pearsall, S.H., B.J. McCrodden, and P.A. Townsend. 2005. Adaptive management of flows in the lower Roanoke River, North Carolina, USA. Environmental management 35:353–367. Peebles, E. 2005. An analysis of freshwater inflow effects on the early stages of fish and their invertebrate prey in the Alafia River Estuary. Report to the Southwest Florida Water Management District. Brooksville, FL. September 2005. Available online at http://www.marine.usf.edu/documents/Peebles/AlafiaPlanktonReport_090705.pdf. Accessed 14 May 2012. 34 Southeastern Naturalist Vol. 12, No. 5 Pierce, A.R., and S.L. King. 2007. The effects of flooding and sedimentation on seed germination of two bottomland hardwood tree species. Wetlands 27:588–594. Poff, N.L. 1996. A hydrogeography of unregulated streams in the United States and an examination of scale-dependence in some hydrological descriptors. Freshwater Biology 36:71–91. Poff, N.L., and J.D. Allan 1995. Functional organization of stream fish assemblages in relation to hydrologic variability. Ecology 76:606–627. Poff, N.L., and J.K.H. Zimmerman. 2010. Ecological responses to altered flow regimes: A literature review to inform the science and management of environmental flows. Freshwater Biology 55:194–205. Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D. Richter, R.E. Sparks, J.C. Stromberg. 1997. The natural flow regime: A paradigm for river conservation and restoration. BioScience 47:769–784. Poff, N.L., B.D. Richter, A.H. Arthington, S.E. Bunn, R.J. Naiman, E. Kendy, M. Acreman, C. Apse, B.P. Bledsoe, M.C. Freeman, J. Henriksen, R.B. Jacobson, J.G. Kennen, D.M. Merritt, J.H. O’Keeffe, J.D. Olden, K. Rogers, R.E. Tharme, and A. Warner. 2010. The ecological limits of hydrologic alteration (ELOHA): A new framework for developing regional environmental flow standards. Freshwater Biology 55:147–170. Pringle, C.M., M.C. Freeman, and B.J. Freeman. 2000. Regional effects of hydrologic alterations on riverin macrobiota in the New World: Tropical-temperate comparisons. BioScience 50:807–823. Rehage, J.S., and W.F. Loftus. 2007. Seasonal fish community variation in the headwater mangrove creeks in the Southwestern Everglades: An examination of their role as dry-down refuges. Bulletin of Marine Science 80:625–645. Richter, B.D., J.V. Baumgartner, R. Wiginton, and D.P. Braun. 1997. How much water does a river need? Freshwater Biology 37:231–249. Richter, B.D., R. Mathews, D.L. Harrison, and R. Wiginton. 2003. Ecologically sustainable water management: Managing river flows for ecological integrity. Ecological Applications 13:206–224. Rogers, M.W., M.S. Allen, and M.D. Jones. 2005. Relationship between river surface level and fish assemblage in the Ocklawaha River, Florida. River Research and Applications 21:501–511. Roy, A.H., M.C. Freeman, B.J. Freeman, S.J. Wenger, W.E. Ensign, and J.L. Meyer. 2005. Investigating hydrologic alteration as a mechanism of fish assemblage shifts in urbanizing streams. Journal of the North American Benthological Society 24:656–678. Sherk, G.W. 1991. Meetings of waters: The conceptual confluence of water law in the eastern and western States. Natural Resources and Environment 5 :47–49. Sherk, G.W. 1994. Resolving interstate water conflicts in the eastern United States: The re-emergence of the federal interstate compact. Water Resources Bulletin 30:397–408. Sherk, G.W. 2000. Dividing the Waters: The Resolution of Interstate Water Conflicts in the United States. Kluwer Law International, Cambridge, MA. Smith, P.P., and J.B. Atkinson. 1999. Flood and debris-flow effects on Virginia Brook Trout populations. Proceedings of the Annual Conference of the Southeastern Association of the Fish and Wildlife Agencies 53:91–99. Smith, S., J. Odenkirk, and S. Reeser. 2005. Smallmouth Bass recruitment variability and its relation to stream discharge in three Virginia rivers. North American Journal of Fisheries Management 25:1112–1121. 2013 R.A. McManamay, D.J. Orth, J. Kauffman, and M.M. Davis 35 Southern Instream Flow Network (SIFN). 2010. Southern Instream Flow research agenda: A proposal to a proposal to advance the science for instream flow protection in southern rivers. Report prepared by the Southern Instream Flow Network, Southeast Aquatic Resources Partnership. March 2010. Available online at http://www.southeastaquatics. net/programs/sifn. Accessed 10 February 2012. Strong, W.A., and E.J. Nagid. 2006. Relations between hydrologic variables and fish assemblages in the Ocklawaha and St. Johns Rivers, Florida. Final Report prepared for St. Johns River Water Management District. Contract No. SD170AA. Gainesville, FL. Sullivan, P.J., J.M. Acheson, P.L. Angermeier, T. Faast, J. Flemma, C.M. Jones, E.E. Knudsen, T.J. Minello, D.H. Secor, R. Wunderlich, and B.A. Zanetell. 2006. Defining and implementing best available science for fisheries and environmental science, policy, and management. Fisheries 31:460–465. Taulbee, W.K., C.T. Nietch, D. Brown, B. Ramakrishnan, and M.J. Tompkins. 2009. Ecosystem consequences of contrasting flow regimes in an urban-effects stream mesocosm study. Journal of the American Water Resources Association 45:907–927. The Florida Statutes (TFS). 2011. The 2011 Florida Statutes. Title XXVIII. Natural resouces; conservation; reclamation; and use. 373.042. Minimum flows and levels. Available online at http://www.leg.state.fl.us/Statutes/index.cfm?App_mode = Display_ Statute&Search_String = &URL = 0300-0399/0373/Sections/0373.042.html. Accessed 14 May 2012. Tharme, R.E. 2003. A global perspective on environmental flow assessment: Emerging trends in the development and application of environmental flow methodologies for rivers. River Research and Applications 19:397–441. Therneau, T.M., and E.J. Atkinson. 1997. An introduction to recursive partitioning using RPART routines. Mayo Foundation Publication. September 1997. Available online at http://www.mayo.edu/hsr/techrpt/61.pdf. Accessed 12 October 2011. Therneau, T.M., B. Atkinson, B. Ripley. 2011. Rpart: Recursive partitioning. Available online at http://cran.r-project.org/web/packages/rpart/index.html. Accessed 12 October 2011. Townsend, P.A. 2001. Relationships between vegetation patterns and hydroperiod on the Roanoke River floodplain, North Carolina. Plant Ecology 156:43–5 8. Travnichek, V.H., and M.J. Maceina. 1994. Comparison of flow regulation effects on fish assemblages in shallow and deep water habitats in the Tallapoosa River, Alabama. Journal of Freshwater Ecology 9:207–216. Travnichek, V.H., M.B. Bain, and M.J. Maceina. 1995. Recovery of a warmwater fish assemblage after the initiation of a minimum-flow release downstream from a hydroelectric dam. Transactions of the American Fisheries Society 124:836–844. US Fish and Wildlife Service (USFWS). 2012. Landscape conservation cooperatives. Available online at http://www.fws.gov/science/shc/lcc.html. Accessed 14 May 2012. US Geological Survey (USGS). 2012. National Water-Quality Assessment (NAWQA) Program. Available online at http://water.usgs.gov/nawqa/. Accessed 14 May 2012. Van Den Avyle, M.J., and M.A. Maynard. 1994. Effects of saltwater intrusion and flow diversion on reproductive success of Striped Bass in the Savannah River Estuary. Transactions of the American Fisheries Society 123:886–903. Wallace, J.B., M.R. Whiles, S. Eggert, T.F. Cuffney, G.H. Lugthart, and K. Chung. 1995. Long-term dynamics of coarse particulate organic matter in 3 Appalachian Mountain streams. Journal of the North American Benthological Society 14:217–232. 36 Southeastern Naturalist Vol. 12, No. 5 Walsh, S.J., E.N. Buttermore, O.T. Burgess, and W.E.I. Pine. 2009. Composition of Age-0 Fish Assemblages in the Apalachicola River, River Styx, and Battle Bend, Florida. US Geological Survey Open-File Report 2009-1145. Reston, VA. West, J.C., A. G. Eversole, and R.W. Christie. 1988. Influence of river discharge on blueback herring abundance. Proceedings of the Annual Conference of the Southeastern Association of the Fish and Wildlife Agencies 42:166–174. Zorn, T.G., P.W. Seelbacn, E.S. Rutherford, T.C. Wills, S.T. Cheng, and M.J. Wiley. 2008. A landscape-scale habitat suitability model to evaluate impacts of groundwater withdrawals on fish communities in Michigan streams. Michigan Department of Natural Resources Fisheries Research Report 2089. Ann Arbor, MI. Available online at http:// www.michigandnr.com/PUBLICATIONS/PDFS/ifr/ifrlibra/Research/reports/2089/ RR2089.pdf. Accessed 8 May 2012.