Population Structure and Habitat Use of the Endangered Watercress Darter Free

Alabama has more fish species than any other state (Stein 2002), and is host to 20 endemic species (Boschung and Mayden 2004; Baker et al. 2013). However, ≥124 Alabama fish species are considered imperiled (Boschung and Mayden 2004), 16 are listed as endangered or threatened under the U.S. Endangered Species Act (ESA 1973, as amended; USFWS 2014), and 2 are extinct. Many of the state's most rare and endangered fishes are endemics associated with coldwater springs, habitats that have often been highly altered through hydrological alteration and water extraction (Fluker et al. 2010). One of these coldwater spring endemics is the watercress darter, Etheostoma nuchale, a small and colorful member of the clade Astatichthys within the subgenus Oligocephalus, which was not discovered until 1964 (Figure 1; Howell and Caldwell 1965; Near et al. 2011). The species is restricted to four springs (Roebuck, Glenn, Seven, Thomas) within the greater Birmingham metropolitan area of Jefferson County, AL (Figure 2). A fifth population was established at Tapawingo Spring in 1988 as an ark population outside its natural range (USFWS 2009). Some of these populations are believed to be stable, while others are in decline or their status is unknown (Fluker et al. 2008, 2009b; Duncan and Kuhajda 2012).

Initial surveys consistently found watercress darters associated with dense aquatic vegetation in deeper portions of spring pools and their run, which we define as the stream connecting a spring to the nearest tributary (USFWS 1992; Boschung and Mayden 2004; Stiles 2004). The darter has also been found in shallow-water, vegetated habitats with a moderate current (Duncan et al. 2010). The preference for vegetation seems related in part to its diet, which is believed to consist of small snails, aquatic insects, and crustaceans that abound within submerged vegetation (Howell and Caldwell 1965). Aquatic plants also provide shelter from predators and a substrate on which female darters lay eggs during the reproductive season, which peaks in March–July (Stiles 2004) but can occur year-round (Fluker et al. 2009b). Darters are thought to survive no more than 3 y based on discernible size classes present in most seasons across all five populations (Fluker et al. 2009b). Relatively little else is known about the darter's ecology, yet more detailed knowledge of its life history could inform protection and recovery efforts.

The watercress darter population at Roebuck Spring is believed to be the largest population (Fluker et al. 2008, 2009b; Duncan and Kuhajda 2012); is genetically distinct from all other natural populations (Fluker et al. 2010); and is the most accessible for study because of the spring's public ownership (City of Birmingham and State of Alabama), shallow waters, and relatively stable hydrology. The other springs are difficult to sample because of their depth, dense aquatic vegetation, and silt accumulation (Thomas); private ownership (Glenn); or the presence of North American beavers Castor canadensis and the frequent hydrologic changes they create (Seven and Tapawingo). Roebuck Spring has also been the subject of legal action by the U.S. Fish and Wildlife Service against the City of Birmingham on account of two dewatering events in recent years, one in the spring pool in 2008 (Fluker et al. 2009a) and another in the uppermost portion of the spring run in 2013 (G. Thornton, U.S. Department of the Interior, personal communication).

Given the importance of Roebuck Spring to the survival of the species and the spring's accessibility, we undertook a study to better understand the population demographics and habitat use by the watercress darter. Although most previous work had sampled Roebuck Spring pool (Fluker et al. 2008, 2009a, 2009b), the spring run was suspected to support a large darter population because of its length, diversity of habitat, abundance of aquatic vegetation, and relatively good water quality. Thus, we examined three main questions about the spring run: 1) To what extent, if any, does the watercress darter inhabit the run? 2) What habitats do the darters most prefer in the run? and 3) What characterizes the most preferred habitats? Answers to these questions will help in determining the importance of the run to the overall population at Roebuck Spring, and will guide conservation efforts on effective habitat restoration and creation throughout the darter's range.

A secondary focus of our study was the use of a throw trap, which was a novel sampling methodology for this species. Our previous experience suggested that seine nets were very poor at catching watercress darters, which led to inaccurate estimates of darter density in these heavily vegetated spring habitats. The most vexing problem was that seine nets were difficult to pull through dense aquatic vegetation, and the net bottom lifted off the substrate, which gave darters ample opportunity to escape. To sample darters in this study, we dropped a 1 × 1 m metal throw trap with tall sides on a distinct spring habitat, and then sampled the interior of the trap until no more darters were captured. We will present a summarized quantitative analysis of the effectiveness of this technique in a different paper.

Roebuck Spring (33°35.089′N, 86°42.583′W) is located within the Ridge and Valley ecoregion, which is defined by elongated mountains and intervening valleys (Griffith et al. 2001). The spring is located in Jones Valley, whose carbonate rocks are punctured by numerous springs (Osborne et al. 1989; Duncan 2013). The spring is within the eastern part of the City of Birmingham, and is surrounded by busy city streets, large shopping centers, and an interstate. A portion of the spring pool is owned by the State of Alabama as part of a juvenile detention facility. The remainder of the pool and the entire spring run are owned by the City of Birmingham and managed by the Parks and Recreation Board. The spring pool measures roughly 150 m long and 50 m wide and is impounded by a small dam. The first 54 m of the 807-m run are piped belowground under tennis courts. The run flows freely for 63 m, and is then piped for another 64 m under a drive. Immediately adjacent to this section of the run is a parking lot for Birmingham's Don A. Hawkins Recreation Center. After the parking area, the run flows through a public golf course before joining Village Creek, a tributary to the Locust Fork of the Black Warrior River drainage. The floodplain of the run is presently in an early phase of secondary succession that began in 2009 and now serves as a buffer between the golf course and spring run. Two low top-flowing dams are located along the run; these are the result of channel modifications during the past century.

We surveyed the habitats and their abundance within the spring run in March–April 2011 using 17 belt transects placed across the run at 40-m intervals. We measured stream width (between the wetted margins) and thalweg depth along each transect. We divided the area within each 1-m-wide transect into 0.25-m² square blocks. We classified the habitat for each block based on the dominant vegetation type or substrate cover. Most aquatic plants in the spring run grow in single-species patches, which made habitat classification relatively simple.

We measured the biomass of common vegetated habitats as an estimate of habitat quality for darters. Biomass may correspond to availability of foraging opportunities, egg-laying surfaces, and shelter from predators. However, it may not accurately reflect these and other measures of habitat quality, such as complexity. For any aquatic plant habitats that were ≥5% of all sample blocks, we took biomass samples within a 25 × 25 cm quadrat. We randomly chose 10 quadrats/species for sampling. We collected submergent vegetation in each quadrat and discarded emergent vegetation. We did not measure depth, so we scaled biomass measurements only to surface area. We removed debris and large macroinvertebrates from samples, but it was impractical to remove minute invertebrates and epiphytic algae. We allowed samples to drip dry, then weighed them (wet weight). We measured dry weights after complete drying in an oven at 50°C. Water can be a large proportion of tissue mass in aquatic plants; therefore, wet weights may better represent in situ habitat relative to dry weights, but dry weights do not include the water film adherent to stems when wet weight is measured. We compared wet weights among species with an Analysis of Variance and conducted post hoc analyses using Tukey's Honestly Significant Difference test; we set alpha level at 0.05 for all analyses

During May–June 2011, we conducted surveys of watercress darters in the five most abundant habitats as indicated by the habitat survey. This coincided with the breeding season so that we might understand how all age classes of darters use these habitats. Cumulatively, these five major habitats accounted for 70% of the surveyed blocks and included watercress Nasturtium officinale, coontail Ceratophyllum demersum, silt, American bur-reed Sparganium americanum, and alligator weed Alternanthera philoxeroides. Both watercress and alligator weed are nonnative invasive species known to disrupt aquatic ecosystems (Spencer and Coulson 1976; Benson et al. 2004). As a result of concerns about habitat damage from repeated sampling in this relatively small, but potentially important, portion of the darter's range, we conducted fish sampling only during this single season.

We distributed eight random samples of each of the five habitats across nine stream segments to provide spatial representation. Segments averaged 74 m in length (range = 31–127 m) and adjacent segments usually differed in their average width, current speed, depth, and prevalent aquatic vegetation. We chose sample locations within the segments at random from patches of the appropriate habitats. We sampled each habitat type in eight locations maximally spread across segments. Not all habitats were present in each segment, so a uniform distribution of sampling locations was not possible. Chosen sampling locations usually consisted entirely of the target habitat; however, in two cases we found another type of vegetation to be present during sampling, which totaled <10% of the sampled area.

We captured darters by dropping a heavy-gauge, aluminum-frame throw trap (Kushlan 1981; Jordan et al. 1997) with solid siding on each sample locations (Figure 3). The frame measured 1 m on each side and was 0.77 m tall. Immediately after we dropped the frame, we forced it into the sediment to prevent darters from escaping. We surrounded each sample location with a buffer of ≥0.5 m of the target habitat to help ensure that the sample was representative of that habitat. We extracted darters using a seine net and dip nets as part of the separate comparative study of sampling methods. First, we pulled through the frame a shortened 10 × 4 ft (3.05 × 1.22 m) seine, green dipped with 1/8-in (0.32-cm) Delta mesh (Memphis Net & Twine Co., Inc., Memphis, TN). This involved three people—one for each brail and a third to manually pull the lead-line along the substrate and through the dense vegetation. We completed a single seine pass for each frame. Second, three workers then sampled the contents of the frame using two dip nets measuring 15 in (38 cm) wide and 11–15 in (28–38 cm) long, and having a mesh size of 1/16 in (1.6 mm). Three overlapping sweeps of the dip nets completed a single “pass” through the frame. A pass began with two nets pushed simultaneously and unidirectionally along opposite sides of the frame. One net would then be swept in the opposite direction through the center of the frame to complete the pass. Nets were swept only across the substrate because darters lack a swim bladder and naturally rest along the bottom. Dip net contents were usually laden with organisms and silt, and the latter was flushed out by gently shaking the net with its contents in the stream's current while keeping the frame of the net above the surface. We continued passes through the frame until at least three passes were completed without catching a darter as detected during visual examination of the flushed (rinsed) contents of the dip nets.

We immediately placed captured darters in a 5-gallon bucket with clear spring water, while we placed vegetation and debris in large coolers with clear spring water. We then brought buckets and coolers to shore for processing. We carefully examined vegetation for darters by shaking small clumps over a water-filled bucket and then by floating handfuls of vegetation in a white enameled pan to facilitate the visual detection of small darters. We found it useful to sort through the floated material with white plastic spoons, which we also used to gently scoop up darters. We passed remnant water in the cooler and buckets through a fine-mesh aquarium net to capture remaining darters. We positively identified all watercress darters captured and recorded their length (mm, standard length), sex (if adults), and size categories (juvenile, 9-23 mm SL; medium, 24-31 mm SL; or large adult, 32–45 mm SL). Some samples with an abundance of vegetation required >1 h to process, with a team of 4–6 workers. During processing we regularly replaced water in coolers and buckets with new clean spring water from the spring run to minimize risk to captured organisms.

We used the data from the habitat and watercress darter survey to provide a simple estimate of the population size of the darter in Roebuck Spring's run. In short, we multiplied the estimated density of darters for habitats by an estimate of area occupied by each of these habitats in the run, and then summed the values. We used average stream width from the habitat surveys to represent stream width in our calculations. We estimated the length of the run to be 807 m using Google Earth and the ruler tool's path function. We subtracted the length of the run flowing through culverts and across the secondary dam, yielding 671 m of potentially inhabitable stream. We multiplied this and the stream width to estimate the total inhabitable area (7,046 m²) of the run. For each of the five major habitats, we multiplied its proportional coverage as estimated in the habitat survey by the total inhabitable area of the run to yield an estimate of each habitat's total area. Then, for each major habitat, we multiplied the average density of watercress darters (and its subcategories of sex and size) by the area occupied by that habitat, and then summed these values.

We next factored in darters in the remaining minor habitats that comprise 31.4% of the spring run. For minor habitats that were vegetated, we used the average darter density from the four major vegetated habitats sampled for darters. We then multiplied this density by the estimated area of all minor vegetated habitats in the run to approximate the total darters in all minor vegetated microhabitats. For nonvegetated minor habitats, we used a similar approach by multiplying the density of darters in the one nonvegetated major habitat (silt) by the estimated area of all minor nonvegetated habitats in the run. This yielded an estimate of the total darters in all minor nonvegetated microhabitats. We expect this is a conservative estimate because darter density in the silt habitat was very low, and other nonvegetated minor habitats in the spring run (e.g., woody debris, rocks) provide more cover than silt and likely have greater densities of darters. For instance, Duncan et al. (2010) found that watercress darter densities were the second highest in detritus (woody debris) among eight habitats sampled at Seven Springs. Finally, we summed these three estimates of darter numbers in the spring run (all major microhabitats, vegetated minor microhabitats, and nonvegetated minor microhabitats) to yield the final estimate of the watercress darter population size in the spring run during May–June 2011. We determined confidence intervals of 95% by employing a parametric bootstrap (10,000 iterations) for our area-stratified abundance estimates, performed in R statistical language (version 3.2.0; R Core Team 2015).

The width of the run averaged 10.5 m (SD = 3.1) and ranged from 7.3 to 19.2 m, while maximal depth averaged 40.3 cm (SD = 23.3) and ranged from 11.8 to 95.0 cm. Across all transects, we surveyed 684 0.25-m² blocks (Table S1, Supplemental Material). Twelve of the 17 habitat categories were dominated by plants, and these categories comprised 74.7% of the surveyed area. The five most abundant habitats (hereafter, major habitats) were watercress (21% of surveyed blocks) followed by silt (16%), coontail (16%), bur-reed (8%), and alligator weed (7%). The other habitats included mixed species (7%), creeping primrose-willow Ludwigia repens (6%), rock (6%), cattail Typha sp. (5%), woody debris and leaf fragments (2%), terrestrial plant species (2%), fontinalis moss Fontinalis sp. (1%), green arrow arum Peltandra virginica (1%), sand (1%), green algae (1%), emergent sediment deposits (1%), and pondweed Potamogeton sp. (<1%).

Creeping primrose-willow had the greatest wet and dry weights, while cattail had the least wet and dry weights (Table S2, Supplemental Material; Figure 4). The relative rank order of mean weights was the same for wet and dry weights with the exception of watercress and alligator weed, which swapped position in rank order. We found significant differences between many of the plants when either wet or dry weights were compared. In general, creeping primrose-willow and cattail differed from more species than any others when comparing wet or dry weights.

Altogether, we captured 629 watercress darters across the five major habitats (Table S3, Supplemental Material). The majority were captured in coontail (53%), followed by watercress (23%), alligator weed (15%), bur-reed (5%), and silt (4%) habitats (Table 1). Average darter density in coontail was 10-fold higher than in bur-reed and silt habitats and roughly double that in watercress (the namesake plant for the darter's common name). Two-thirds (67%) of captures were juveniles, while 18% and 16% were adult females and males, respectively. There were significantly more juveniles in coontail than in bur-reed and silt habitats; juvenile densities were intermediate in watercress and alligator weed habitats. Females and males were significantly denser in coontail relative to alligator weed, bur-reed, and silt habitats, while watercress had an intermediate density for each sex. Medium-sized darters were significantly denser in coontail than in the other habitats, but the few large darters collected in this study were distributed similarly across habitats. Though silt and bur-reed habitats generally had very low densities across all categories, they were not found to be statistically different from most other habitats apart from coontail. The reason for this was that in each of these two habitats, one of the eight samples yielded darters at a density 4–6 times the mean of the seven other samples of each habitat.

Our estimate of the watercress darter population size in the five major habitats, comprising an estimated 68.6% of the spring run, was 85,114 darters (95% CI = 55,995–115,818). Using the average darter density from the four vegetated major habitats (18.9 darters/m²), we calculated there was an additional 29,711 watercress darters (95% CI = 20,269–40,066) in the remaining 22.4% of the run that was vegetated. We calculated the number of darters in the 9% of the run that was nonvegetated to be 2,107 (95% CI = 124–4,250). In sum, we estimated there were 116,932 (95% CI = 79,358–155,965) watercress darters present in the spring run during May–June 2011.

Our survey of watercress darters in the five most common habitats of Roebuck Spring yielded new insights into the ecology of this federally endangered darter. Our novel technique for sampling this species yielded densities far surpassing those of any previous survey of the species. For example, the maximum density of darters from a survey of Seven Springs yielded 18/m² (Duncan et al. 2010), whereas the present study's maximum was 87/m². However, inferences about habitat use by watercress darters must take into consideration that our study provides only a snap-shot during the springtime, and although this is a critically important time for the species on account of peak spawning, habitat use patterns may vary throughout the year. For example, as ambient temperatures rise in the summer, the distribution of watercress darters may contract toward the spring pool. Furthermore, at our location watercress dies back from late summer to midspringtime, and alligator weed dies back during the winter, forcing some darters to find alternative habitats during these times (RSD, personal observation). Thus, further studies addressing habitat use in other seasons would likely reveal new insights into the needs of the species.

For the first time, we documented watercress darters inhabiting coontail, a plant that appears to provide important habitat in the spring run. The plant species is not known to occupy the spring pool or other springs harboring watercress darters. Darter densities in coontail were more than double those in alligator weed and watercress, the habitats with the next highest densities. Alligator weed and watercress are nonnative invasives that may host fewer native, invertebrate prey species relative to coontail. In addition, alligator weed and watercress die back seasonally and invest in emergent growth that does not augment habitat for darters, whereas coontail is a submergent plant providing year-round cover. Though the biomass of these three species was comparable, coontail stems were thinner and seemed to be more numerous, thus potentially providing disproportionately more darter habitat. In fact, though coontail was mainly found where the current may be too swift for a small darter to occupy, coontail's thick growth slows the current speed down to half that of the thalweg (RSD, unpublished data). Despite their potential drawbacks as habitat, alligator weed and watercress seem to provide additional darter habitat along the shallow margins of the run where coontail does not grow.

Bur-reed had the lowest biomass and lowest average darter densities of any vegetated habitat. These plants have broad, unbranching, strap-like leaves that are submergent or float when young, but become stiff and emergent in late summer. Silt habitat had a very low average density of darters. Both the silt and, to a lesser degree, bur-reed seem to provide little cover for watercress darters and probably support relatively low amounts of prey. Darters captured in both habitats may have been in transit to more suitable habitats.

Surveys of the spring pool have found that fontinalis moss, which grows liberally there, shelters an abundance of darters (Fluker et al. 2008, 2009b; Duncan and Kuhajda 2012). As part of our separate and parallel study quantifying the effectiveness of the frame method for sampling darters, we completed a single sample of fontinalis moss habitat using the frame in Roebuck Spring pool. This sample yielded 21 darters, of which 5 were juveniles, 13 were medium-sized, and 3 were large; the adults included 6 males and 10 females. Though we caution against overinterpretation of this single sample, we note that these values are roughly half the average number of darters found per sample in coontail, but higher than the average densities in any of the other habitats studied. Both fontinalis moss and coontail are submergent native species whose numerous branching filaments form a structurally complex habitat. The presence of both species further diversifies and expands the habitats available for watercress darters. Because fontinalis moss and coontail do not develop roots and colonize new areas via stem fragments (Crum and Anderson 1981; Godfrey and Wooten 1981), both could be easy to introduce during habitat expansion or restoration projects for watercress darters. Creeping primrose-willow, while uncommon at Roebuck Spring, has been found to support densities of watercress darters as high as 8.7/m² in seining surveys elsewhere (RSD and BRK, personal observation) and seems to have a comparatively high biomass. Given that it is commonly propagated from cuttings by aquarists (Öztürk et al. 2004), creeping primrose-willow may be another submergent native plant that could be easily established to create new watercress darter habitat.

Our findings of moderate to high densities of darters across a range of vegetated habitats in the Roebuck Spring system supports the assessment that watercress darters primarily associate with densely vegetated habitats, but that otherwise, it is a versatile species. These habitats can be deep, with little to no current (e.g., fontinalis moss); moderately deep with swift current (coontail); shallow with a moderate current (alligator weed and watercress); and vegetation can be native or nonnative, and submergent or emergent. The proportions of all size and sex classes of darters were nearly identical across the coontail, alligator weed, and watercress habitats, suggesting the relative importance of these habitats to the darter is independent of these population-structure categories. However, the darter's dependency on densely vegetated habitats may be its greatest vulnerability at this time. Such habitats can no longer be supported throughout most of its range because of channel modification and stormwater runoff from the surrounding urban landscape. At all five of the springs supporting watercress darters, aquatic vegetation is scarce to nonexistent in the receiving stream (Fluker et al. 2008, 2009b; Duncan and Kuhajda 2012).

Similar studies of the relationship between the geomorphology and habitat use and availability of other springs harboring watercress darters would significantly advance the knowledge needed to help the species recover. In the Roebuck Springs run, our findings suggest that geomorphological diversity helps sustain a diversity of aquatic vegetation. This plant diversity may be critical for sustaining the run's robust population by ensuring there is appropriate habitat available over the seasons and during episodic disruptions such as drought and extreme freezes. Any perturbation reducing the geomorphological diversity and, hence habitat diversity, is a threat to the population, including high levels of stormwater runoff, anthropogenic channel modification, or even extensive hydrological alteration by beavers.

To our knowledge, this study provides the most complete picture of the structure and habitat use of a watercress darter population to date. This population was probably near its annual maximum because peak spawning occurs in springtime. Males and females were found in similar proportions, and there were very few large adults, as could be expected for this short-lived species. Juveniles comprised two-thirds of the population in the spring run, suggesting that prolific reproduction is occurring. Given that one-third of the population was mature, we expect that approximately half the juveniles present in May–June 2011 would survive to adulthood over the next year if conditions in the population remain stable. Though Roebuck Spring's run harbors a large population of darters, this population faces threats presented by urban stormwater; beavers and muskrats Ondatra zibethicus that destroy aquatic vegetation used by the darter; and the nonnative invasive virile crayfish Orconectes virilis, which is a probable predator of, and competitor with, the watercress darter (Savino and Miller 1991; Bryan et al. 2002; Dorn and Mittelbach 2004; Taylor and Soucek 2010). However, this new knowledge about the darter's habitat preferences may help managers restore darter habitat after disturbances and expand darter habitat when opportunities arise.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Data on vegetation cover and stream geomorphology collected from Roebuck Spring run during March–April 2011 using 17 belt transects placed across the run at 40-m intervals. Located in Birmingham, Alabama, the spring is one of five known locations sheltering the endangered watercress darter Etheostoma nuchale.

Found at: DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S1 (48 KB XLSX).

Table S2. Dry weights (above) and wet weights (below) of common aquatic plants (and plant debris) collected during April 2011 from among 17 transects spanning Roebuck Spring run, Birmingham, Alabama. The spring is one of the five known locations for the endangered watercress darter Etheostoma nuchale. Dry and wet weights were used as measures of habitat quality for the darter. Data linking dry and wet weights for individual samples were lost, so each is reported here in separate tables.

Found at: DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S2 (18 KB DOCX).

Table S3. Data on watercress darters Etheostoma nuchale captured in surveys of five major habitat types in Roebuck Spring run during May–June 2011. Each habitat was sampled in eight locations. Fish were trapped in a 1 × 1 × 0.77 m aluminum-frame throw trap, extracted with the single pass of a seine net, then sampled to depletion with dip nets. Fish were measured (standard length) and classified by size and sex.

Found at: DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S3 (219 KB DOCX).

Reference S1. Benson AJ, Jacono CC, Fuller PL, McKercher ER, Richerson MM. 2004. Summary report of nonindigenous aquatic species in U.S. Fish and Wildlife Service Region 5. Report to the U.S. Fish and Wildlife Service, Arlington, Virginia.

Found at: DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S4 (4.6 MB PDF); also available at https://nas.er.usgs.gov/publications/R5finalreport.pdf (4.6 MB PDF).

Reference S2. Duncan RS, Kuhajda BR. 2012. Status survey of the endangered watercress darter, Etheostoma nuchale. Report to U.S. Fish and Wildlife Service, Jackson, Mississippi.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S5 (4.2 MB PDF).

Reference S3. Fluker BL, Kuhajda BR, Harris PM. 2008. Status and distribution survey and genetics of the endangered watercress darter, Etheostoma nuchale. Report to the Alabama Department of Conservation and Natural Resources (Montgomery) and the U.S. States Fish and Wildlife Service, Jackson, Mississippi.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S6 (7.4 MB PDF).

Reference S4. Fluker BL, Kuhajda BR, Harris PM. 2009b. Status and distribution survey and genetics of the endangered watercress darter, Etheostoma nuchale. Report to the Alabama Department of Conservation and Natural Resources, Montgomery, and U.S. Fish and Wildlife Service, Jackson, Mississippi.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S7 (3.6 MB PDF).

Reference S5. Griffith GE, Omernik JM, Comstock JA, Lawrence S, Martin G, Goddard A, Hulcher VJ, Foster T. 2001. Ecoregions of Alabama and Georgia, (color poster with map, descriptive text, summary tables, and photographs). Reston, Virginia: U.S. Geological Survey.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S8 (6.0 MB PDF); also available at https://archive.epa.gov/wed/ecoregions/web/html/alga_eco.html (6.0 MB, pdf).

Reference S6. [USFWS] United States Fish and Wildlife Service. 1992. Watercress darter (Etheostoma nuchale) recovery plan. Jackson, Mississippi: USFWS.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S9 (1.0 MB PDF).

Reference S7. [USFWS] United States Fish and Wildlife Service. 2009. Watercress darter (Etheostoma nuchale) 5-year review: summary and evaluation. Jackson, Mississippi: USFWS.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S10 (154 KB PDF).

Reference S8. [USFWS] United States Fish and Wildlife Service. 2014. Listing and occurrences statewide for Alabama. Daphne: Alabama Ecological Services Field Office.

Found at DOI: http://dx.doi.org/10.3996/072015-JFWM-062.S11 (66 KB PDF); also available at: http://ecos.fws.gov/tess_public/reports/species-listed-by-state-report?state=AL&status=listed (66 KB, pdf).

This study was supported in part by a research contract from the U.S. Fish and Wildlife Service (USFWS contract 40181AM387). We would like to thank the City of Birmingham, the Alabama Department of Conservation of Natural Resources, and USFWS for permission to conduct this study. We are grateful for additional support from the University of Alabama and Birmingham-Southern College. We are very indebted to J. Heath Howell, Karen Marlowe, Zach Napier, Rebecca P. Parker, Matthew S. Piteo, Josh Rabbit, Marty Schulman, and Nicole White for their assistance in conducting the fish sampling. Special thanks to Josh Ennen for assistance with statistics and Sarah C. Hazzard for assistance with cartography (Figure 2). We extend our appreciation to two anonymous reviewers and a Subject Editor, whose feedback improved our manuscript. Permits to collect the species were issued to coauthor B.K. Kuhajda (USFWS Permit TE137403-0 and State of Alabama Protected Species Scientific Collecting Permit 2011000061868680).

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.