Southeastern Naturalist 163 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 22001177 SOUTHEASTERN NATURALIST 1V6o(2l.) :1166,3 N–1o8. 12 New Distributional Records of the Stygobitic Crayfish Cambarus cryptodytes (Decapoda: Cambaridae) in the Floridan Aquifer System of Southwestern Georgia Danté B. Fenolio1,*, Matthew L. Niemiller2, Andrew G. Gluesenkamp1,3, Anna M. McKee4, and Steven J. Taylor2 Abstract - Cambarus cryptodytes (Dougherty Plain Cave Crayfish) is an obligate inhabitant of groundwater habitats (i.e., a stygobiont) with troglomorphic adaptations in the Floridan aquifer system of southwestern Georgia and adjacent Florida panhandle, particularly in the Dougherty Plain and Marianna Lowlands. Documented occurrences of Dougherty Plain Cave Crayfish are spatially distributed as 2 primary clusters separated by a region where few caves and springs have been documented; however, the paucity of humanly accessible karst features in this intermediate region has inhibited investigation of the species’ distribution. To work around this constraint, we employed bottle traps to sample for Dougherty Plain Cave Crayfish and other groundwater fauna in 18 groundwater-monitoring wells that access the Floridan aquifer system in 10 counties in southwestern Georgia. We captured 32 Dougherty Plain Cave Crayfish in 9 wells in 8 counties between September 2014 and August 2015. We detected crayfish at depths ranging from 17.9 m to 40.6 m, and established new county records for Early, Miller, Mitchell, and Seminole counties in Georgia, increasing the number of occurrences in Georgia from 8 to 17 sites. In addition, a new US Geological Survey (USGS) Hydrologic Unit Code 8 (HUC8) watershed record was established for the Spring Creek watershed. These new records fill in the distribution gap between the 2 previously known clusters in Georgia and Jackson County, FL. Furthermore, this study demonstrates that deployment of bottle traps in groundwater-monitoring wells can be an effective approach to presence–absence surveys of stygobionts, especially in areas where surface access to groundwater is limited. Introduction Cambarus cryptodytes (Hobbs) (Dougherty Plain Cave Crayfish; Fig. 1) occurs in the Floridan aquifer system, which underlies the Marianna Lowlands and Dougherty Plain physiographic regions in southwestern Georgia, southeastern Alabama, and extends into the Florida panhandle (Hobbs 1981, 1989; Morris 2006). This cambarid crayfish is an obligate inhabitant of groundwater habitats (stygobiont) and has degenerate eyes and no pigmentation. Reported occurrences of Dougherty Plain Cave Crayfish are clustered in 2 primary groups: one comprising 8 localities in Georgia in Baker, Calhoun, Decatur, and Dougherty counties, and the second 1Department of Conservation and Research, San Antonio Zoo, San Antonio, TX 78212. 2Illinois Natural History Survey, Prairie Research Institute, University of Illinois Urbana– Champaign, Champaign, IL 61820. 3Texas Parks and Wildlife Department, 4200 Smith School Road, Austin, TX 78744. 4U.S. Geological Survey, South Atlantic Water Science Center, Norcross, GA 30093. *Corresponding author - dantefenolio@sazoo.org. Manuscript Editor: Hayden Matttingly Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 164 group in 29 localities in northwestern Florida, including 23 in Jackson County, 5 in Washington County, and 1 in Calhoun County (Fig. 2; Franz et al. 1994; Hobbs 1941, 1981; Hobbs et al. 1977; Morris 2006; Purvis and Opsahl 2005; Skelton 2008). Most occurrence records are from submerged freshwater limestone caves, but the species also has been documented in wells, sinks, and vadose caves (Hobbs 1981, 1989; Hobbs et al. 1977; Purvis and Opsahl 2005; Skelton 2008). It has been hypothesized that additional populations exist between these 2 groups (Purvis and Opsahl 2005, Skelton 2008). However, access to the Floridan aquifer system in this region is limited to groundwater-monitoring wells because few caves and springs are present in this region. Relatively little is known regarding the life history, demography, and ecology of this species (Fenolio et al. 2014). Dougherty Plain Cave Crayfish commonly co-occurs with Eurycea wallacei (Carr) (Georgia Blind Salamander) and is known to prey on the salamander (Fenolio et al. 2013, Means 1992, Sutton and Relyea 1971). This crayfish species is listed as “least concern” by the International Union for Conservation of Nature (IUCN; Cordeiro et al. 2010) due to its broad distribution. However, Dougherty Plain Cave Crayfish is considered “imperiled” (G2) by NatureServe (2016), and “threatened” by the American Fisheries Society (Taylor et al. 2007), primarily because of existing threats to the Floridan aquifer system, such as groundwater withdrawal and pollution. At the state level, Dougherty Plain Cave Crayfish is designated as “threatened” in Georgia (Skelton 2008), but is not currently listed in Florida (Gruver and Coffey 2015). Figure 1. A Dougherty Plain Cave Crayfish, Cambarus cryptodytes, sampled from the Floridan aquifer system in Florida. Southeastern Naturalist 165 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 The Floridan aquifer system underlies more than 250,000 km2 in southern Alabama, southern Georgia, southern South Carolina, and all of Florida. The aquifer system represents the primary water-source for several large cities and is an important source of water for agriculture in the region. For these reasons, the Floridan aquifer system has been designated as an at-risk aquifer by the US Geological Survey (USGS) primarily due to contamination of groundwater from agricultural activities within the recharge basin of the system (Nolan et al. 1998). Overharvesting of groundwater, particularly for intensive center-pivot irrigation across the region, is also a potentially important threat (Fenolio et al. 2013). Unfortunately, no regular monitoring initiatives have been implemented to better delineate the distribution and assess health of populations of groundwater fauna that live in the Floridan aquifer system, including Dougherty Plain Cave Crayfish. In this study, we used an inexpensive trap design to sample for Dougherty Plain Cave Crayfish and other groundwater fauna in open-hole groundwater-monitoring wells that access the Floridan aquifer system in southwestern Georgia. In particular, we Figure 2. Distribution of Cambarus cryptodytes (Dougherty Plain Cave Crayfish) in relation to outcroppings of the Floridan aquifer system (USGS 2015), shaded gray. Occupied sites known prior to the current study are in white-filled circles; new occurrences documented during this study are shown in black-filled circles. Wells sampled with no C. cryptodytes detected are indicated with an X. Thin light-gray lines are county boundaries; medium darker gray lines are HUC8 watersheds. Inset map of southeastern USA shows Floridan aquifer system (gray shading) and study area (dotted line). Scale bar i s 50 km. Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 166 targeted wells located between 2 primary clusters of occurrences to better define the geographic extent of the target species. Methods Site selection and site descriptions We coordinated with the USGS to acquire permission to access and sample groundwater-monitoring wells in southwestern Georgia between population clusters of Dougherty Plain Cave Crayfish. We sampled 18 groundwater-monitoring well sites in 10 counties (Table 1, Fig. 3): Baker, Calhoun, Decatur, Dougherty, Early, Grady, Lowndes, Miller, Mitchell, and Seminole. All but 1 of the well sites were developed in the Floridan aquifer system. Well 11J011 (developed in the deeper Claiborne aquifer; Table 1) is located adjacent to Well 11J012 in Mitchell County, and was mistakenly sampled during the first round of trapping. Mean (± SD) well depth for well sites was 63.8 ± 34.6 m, and mean trap depth for those measured (Table 1) was 38.5 ± 20.1 m. Well casing diameters ranged from 10.2 cm to 50.8 cm (Table 1). Bait choice and trap design We deployed bottle traps baited with either cashews or frozen marine shrimp at each monitoring well (Figs. 4, 5). We chose nuts and shrimp to approximate potential natural food items of target organisms (roots, organic debris, and Malacostraca). We modeled our trap design after those used to detect and monitor groundwater Table 1. Summary of well sites sampled in the Floridan aquifer system of southwestern Georgia. Trap depths were estimated based on the length of line used for each trap. All wells were finished in the Floridan aquifer system with the exception of 11J011 in Mitchell County, which was finished in the Claiborne Group. Trap depth was not measured (nm) at 2 sites. Well Casing Casing Trap County Well site Latitude Longitude depth (m) depth (m) diameter (cm) depth (m) Baker 10H009 31.2333 -84.4986 61.0 28.0 10.2 28.1 Baker 12K014 31.4383 -84.1850 41.8 21.0 10.2 40.3 Calhoun 10K005 31.4814 -84.4642 42.1 12.2 10.2 40.6 Decatur 08E038 30.7866 -84.6661 45.1 39.2 14.3 36.7 Decatur 08E039 30.8019 -84.6781 19.7 11.0 15.2 13.1 Decatur 09F520 30.9617 -84.5961 76.5 39.6 40.6 36.8 Dougherty 13L012 31.5181 -84.1119 66.4 16.5 10.2 39.6 Dougherty 13L180 31.5464 -84.0139 94.5 67.1 15.2 49.2 Early 06G006 31.0742 -84.9864 37.5 17.7 10.2 36.1 Early 08K001 31.3772 -84.6547 38.1 18.6 10.2 17.9 Grady 12F036 30.8764 -84.2144 142.3 139.6 15.2 47.9 Lowndes 19E009 30.8308 -83.2828 104.2 61.0 50.8 104.2 Miller 07H002 31.1689 -84.8317 22.9 19.5 10.2 nm Miller 08G001 31.1142 -84.6789 68.6 39.6 30.5 32.7 Mitchell 10G313 31.0853 -84.4394 62.8 26.5 30.5 39.4 Mitchell 11J011 31.3006 -84.3231 127.1 121.0 10.2 nm Mitchell 11J012 31.3006 -84.3231 68.6 18.9 15.2 24.5 Seminole 06F001 30.8969 -84.8986 30.0 18.9 10.2 29.0 Southeastern Naturalist 167 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 fauna in Texas (Gluesenkamp and Krejca 2007); consequently, our trap design differed from that previously used by Purvis and Opsahl (2005) to collect Dougherty Plain Cave Crayfish in southwestern Georgia. Recovery of traps was a primary consideration, and we designed traps to reduce the risk of trap entanglement and loss due to obstruction, abrasion, and changing flow conditions, while providing large volume and ease of sampling. Trap construction We constructed traps from disposable 1.5-L polycarbonate water bottles with an elongate, cylindrical body and short neck (Glacéau Smartwater®; Fig. 5). This shape allowed construction of traps with a relatively large volume relative to their diameter and reduced the likelihood of them becoming lodged in well bores (Hutchins and Orndorff, 2009). These bottles also have thicker, more rigid plastic than most sports-drink or soda bottles. We melted approximately 150 small holes (~1–2 mm in diameter) through the sides of the bottle using an 18-W soldering iron to create a thickened “grommet” of plastic around each hole, which increased Figure 3. Location of USGS groundwater-monitoring wells sampled during the current study in 9 Georgia counties in relation to where the Floridan aquifer system is unconfined (USGS 2015) (gray-shaded area). Wells 11J011 and 11J012 in Mitchell County are located immediately adjacent to each other and appear as 1 point on this map. A well in Lowndes County was also sampled during the study and is not shown here due to its location east of Thomas County. Inset map of southeastern USA shows Floridan aquifer system (gray shading) and study area (dotted line). Scale bar is 50 km. Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 168 strength and rigidity. Numerous perforations increased diffusion of water across the trap wall to reduce the risk of fouled bait impacting trapped organisms. These perforations terminated 3 cm above the bottom of the bottle to allow retention of small organisms when the trap was retrieved from the well. We attached weights (usually, a small mesh bag of marbles or 1–4 steel lug-nuts, small enough not to obstruct the well bore or pump) to the bottom of the bottle via #300 braided-nylon seine-twine passed through a hole in the center of the trap bottom with a knot to plug the hole. Braided-nylon seine-twine is abrasion-resistant and easier to handle during trap deployment and retrieval than smaller-diameter products such as fishing line or survey twine. Monofilament line used in previous studies (Purvis and Opsahl 2005) is unsuitable due to its lack of abrasion resistance and propensity to cut Figure 4. (A) A well-pipe bottle trap ready to be deployed at a (B) U S G S g r o u n d - water monitoring well (08K001) in Early County, GA. Note the mop head and cashews used as bait at the bottom of the trap. (C and D) An adult Cambarus cryptodytes captured in a well-pipe bottle trap deployed at a USGS groundwater monitoring well (10G313) in Mitchell County, GA. Southeastern Naturalist 169 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 through plastic under tension. We suspended weights below the trap (as opposed to affixing them to the outside of the trap or placing them inside) to aid in deployment while reducing risk of entanglement or mechanical injury to trapped organisms. We cut the top of the bottle at the base of the neck and inverted it such that the small opening was oriented toward the interior of the bottle, creating a funnel. We then tied several strands of cotton mop head to the weight line inside the trap to provide refuge for smaller crayfish and other invertebrates. We created a tether by threading 0.75 m of braided-nylon twine through 1 eye of a heavy-duty barrel swivel (size 1/0) and tied the other eye to a primary line consisting of the same nylon material. We passed the trap tether through 2 pairs of aligned holes in the top and body of the trap that served at attachment points and tied a knot in the tether near the swivel to create secure double loops that allowed the trap to be checked and emptied without complete disassembly of funnel and body. We added plastic zip ties to additional Figure 5. Left illustration: Well-pipe bottle traps used for collecting groundwater organisms in this study were constructed from 1.5-L Smartwater® plastic bottles.The upper tether is tied in a loop with a knot just below a swivel. Note the holes in the bottle traps allowing for water passage through the traps. Scale bar is 15 cm. Right illustration: A partially disassembled bottle trap showing trap construction. Bait is attached to the inner end of lower line (inside the bottle trap; bait can be contained within the tip of a nylon stocking) and a weight is attached to the other end of the same line (outside of the bottle trap; weight not depicted in the illustration but attaches to the loop). Illustrations by Matthew Stephens. Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 170 attachment points in some cases to further secure the funnel to the body. We provided enough primary line (line connecting trap and surface) such that traps could be lowered to the bottom of each well. However, some monitoring wells were obstructed partway down; thus, we lowered traps as deep as possible into these wells (Table 1). We secured all traps by tying the line of f at the top of the well casing. Trap deployment We deployed traps from September 2014 through August 2015 for a total of 10 rounds of sampling, baiting the traps with unsalted raw cashew nuts for the first 5 (September 2014–February 2015) and shrimp for the last 5 (March 2015–August 2015) rounds of trapping. We baited traps with cashews at 18 well sites for a total of 716 trap days and shrimp at 10 well sites for a total of 648 trap days. Deployment depths and baits used varied among well sites (Tables 1, 2). We estimated deployment depths based on the length of line used to set the trap, minus the height of the groundwater monitoring-well housing to which the line was tied, minus an additional 0.5 m to account for line used to tie the line to the housing. When this estimate exceeded the groundwater-monitoring well depth, we assumed that the deployment depth was the same as the groundwater-monitoring well depth. We checked traps by slowly pulling them up so that crayfish were not injured and the traps did not become snagged inside the well casing. We attempted to check traps within 14 days of deployment for any given trapping period. However, logistical constraints resulted in trapping-event durations of 6–36 days (mean ± SD = 13.8 ± 8.2 days). The 99 trapping events (Table 2) represent 1364 trap days. The number of trapping events per well site ranged from 1 to 10 (mean ± SD = 5.5 ± 2.7 trapping events). We sampled well 07H002 only once because of a collapse below the well casing. Statistical analyses Summary statistics and other statistical analyses were conducted in the R statistical computing environment (v.3.1.2; R Core Team 2015). When applicable, we report mean values ± standard deviation. We conducted Welch’s t-tests to determine if well depth and trap depth differed between sites where cave crayfish were present and those where cave crayfish were absent. We employed logistic regression with Bayesian model averaging (BMA) to examine potential relationships of crayfish mortality with the following predictor variables: trap duration (length of days a trap was deployed), deployment depth of trap, bait type (cashews or shrimp), and the number of crayfish captured in a trap. Bait types were not randomized; therefore, detection of a significant bait-type effect may be confounded with another effect. BMA incorporates model selection and parameter- estimation uncertainties into inference and prediction (Hoeting et al. 1999, Raftery et al. 1997, Wintle et al. 2003), which yields more-accurate predictions than approaches that select a single optimal model (Thomsen et al. 2007, Wintle et al. 2003). BMA was conducted using the bic.glm function in the BMA package in R (Raftery et al. 2009), using Bayesian information criterion (BIC) to compare and identify the most-probable models. Under the BMA approach, the posterior probability that a predictor variable has a nonzero coefficient in the model is used Southeastern Naturalist 171 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 Table 2. Summary of trapping events at 18 well sites in southwestern Georgia from September 2014 to August 2015. CPUE = catch per unit effort defined as number of crayfish captured per trap day. TTFC = time to first capture defined as the number of trap days from the when a trap was first deployed at a well site to when a crayfish was captured, along with the number of sampling events until the first capture of a crayfish (in par entheses). Mean event Bait (number of Crayfish County Well site Total trap days Trap events duration (days ± SD) trap events) Captured CPUE TTFC Baker 10H009 129 9 14.3 ± 8.8 Cashews (4), shrimp (5) 1 0.008 79 (5) Baker 12K014 65 4 16.3 ± 12.8 Cashews (4) 2 0.031 30 (3) Calhoun 10K005 64 4 16.0 ± 12.9 Cashews (4) 4 0.063 29 (3) Decatur 08E038 75 6 12.5 ± 4.3 Cashews (1), shrimp (5) 0 0.000 Decatur 08E039 74 6 12.3 ± 4.2 Cashews (1), shrimp (5) 0 0.000 Decatur 09F520 76 6 12.7 ± 4.0 Cashews (1), shrimp (5) 3 0.039 61 (5) Dougherty 13L012 73 6 12.2 ± 3.5 Cashews (1), shrimp (5) 9 0.123 53 (4) Dougherty 13L180 61 5 12.2 ± 4.0 Cashews (1), shrimp (4) 0 0.000 Early 06G006 136 10 13.6 ± 8.4 Cashews (5), shrimp (5) 0 0.000 Early 08K001 64 4 12.9 ± 12.9 Cashews (4) 4 0.063 8 (1) Grady 12F036 140 10 14.0 ± 8.5 Cashews (5), shrimp (5) 0 0.000 Lowndes 19E009 63 4 15.8 ± 13.8 Cashews (4) 0 0.000 Miller 07H002 7 1 7.0 Cashews (1) 0 0.000 Miller 08G001 62 4 15.5 ± 13.9 Cashews (4) 6 0.097 7 (1) Mitchell 10G313 63 4 15.8 ± 13.8 Cashews (4) 2 0.032 14 (2) Mitchell 11J011 8 1 8.0 Cashews (1) 0 0.000 Mitchell 11J012 130 9 14.4 ± 8.7 Cashews (4), shrimp (5) 0 0.000 Seminole 06F001 74 6 12.3 ± 4.2 Cashews (1), shrimp (5) 1 0.014 37 (3) Total 18 sites 1364 99 13.8 ± 8.2 Cashews (50), shrimp (49) 32 0.023 Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 172 as a measure of the influence of that variable on the response. Predictor variables with high posterior-probabilities contribute most to model fit, whereas explanatory variables with low posterior-probabilities contribute the least (Nally et al. 2008, Selwood et al. 2009, Thomson et al. 2007). We considered explanatory variables with posterior probabilities > 0.75 to be key variables in predicting the response variable (Nally et al. 2008, Selwood et al. 2009, Viallefont et al. 2001). We used hierarchical partitioning to calculate the amount of variance of the response variable that could be attributed independently to each key predictor variable using the hier.part package in R (Mac Nally 2000, Walsh and Mac Nally 2015). Results We captured 32 Dougherty Plain Cave Crayfish from 9 well sites in 8 counties (Table 2, Fig. 2): Baker (2 wells), Calhoun (1 well), Decatur (1 well), Dougherty (1 well), Early (1 well), Miller (1 well), Mitchell (1 well), and Seminole (1 well). At the 9 wells where we detected Dougherty Plain Cave Crayfish, the total number of crayfish captured ranged from 1 to 9 per well, with a mean of 3.6 ± 2.6 crayfish captured. We captured 9 crayfish at Well 13L012 (Dougherty County), 6 crayfish at Well 08G001 (Miller County), 4 crayfish at Well 10K005 (Calhoun County), and 4 crayfish at Well 08K001 (Early County). The number of crayfish captured during a single trapping event ranged from 0 to 5 (Well 13L012), with a mean of 1.6 ± 1.1 crayfish captured during a single trapping-event at the 9 well sites where crayfish were present (Table 2). Catch per unit effort (CPUE), defined as the number of crayfish captured per trap day, ranged from 0.008 to 0.123 crayfish per trap day at sites where crayfish were present, with an overall mean of 0.023 crayfish per trap day for all well sites. Mean time to first capture (TTFC) of a crayfish at a site was 35.3 ± 24.9 trap days, ranging from 7 (Well 08G001) to 79 days (Well 10H009), which corresponded to 3.0 ± 1.5 sampling events. Of the 32 crayfish captured, 15 were captured alive and 17 were found dead in the traps. Mean well depth and trap depth where crayfish were present were 54.1 ± 16.3 m and 33.8 ± 7.7 m, respectively, versus 73.5 ± 45.5 m and 44.5 ± 29.2 m where we did not detect crayfish. The deepest well where we detected crayfish was Well 09F520 in Decatur County at 76.5 m, which had a trap depth of 38.2 m. The shallowest well where we detected crayfish was Well 06F001 in Seminole County at 30.0 m, which had a trap depth of 29.0 m. The deepest depth at which we trapped crayfish was 40.6 m at Well 10K005 in Calhoun County and the shallowest depth at which we trapped crayfish was 17.9 m at Well 08K001 in Early County. Differences between sites where crayfish were present and where crayfish were absent were not statistically significant for either well depth (t = -1.20, df = 10.01, P = 0.256) or trap depth (t = -0.95, df = 6.65, P = 0.378). We caught 18 crayfish in traps baited with cashews at 5 well sites, and 14 crayfish in traps baited with shrimp at 4 well sites. CPUE was nearly identical between cashews and shrimp treatments, at 0.0252 and 0.0216 crayfish per trap day, Southeastern Naturalist 173 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 respectively. We used cashews and shrimp at 10 well sites each. At 4 of these well sites, we captured crayfish after the change from cashews to shr imp as bait. The cumulative posterior probability of the top 5 BMA models was 0.974, indicating a high probability that the “true” model consists of the following 2 key predictor-variables (posterior probability > 0.75) consistently selected by these models: bait type (shrimp) and the number of crayfish captured in a trap (Table 3). The number of crayfish captured in a trap was positively related with mortality, while the use of shrimp as bait was negatively related to mortality. There was weak evidence (posterior probability = 0.70) that trap duration (length of time a trap was deployed before checking) had a positive relationship with crayfish mortality. Trap depth did not have a statistically significant effect on crayfish mortality. Of the 2 key variables, the number of crayfish captured had the greatest contribution to the total independent explained variance (78.2%), while bait type explained about 23.8% of the total independent variance. In addition to Dougherty Plain Cave Crayfish, we trapped a single stygobitic isopod specimen from a well (11J012) in Mitchell County on 4 March 2015. We identified the isopod as member of the hobbsi species group in the genus Caecidotea based on the presence of elongate, plumose setae along the distal margin of exopod of pleopod 1 and exopod of pleopod 2 (Lewis 1988; Fig. 6). Unfortunately, the specimen was female and could not be identified to species because such identification requires examination of male genitalia. Discussion Distribution and status Dougherty Plain Cave Crayfish is the only stygobitic crayfish known from Georgia (Hobbs 1981, Niemiller et al. 2012). Before this study, Dougherty Plain Cave Crayfish was known from 37 localities in Florida and Georgia, including 8 localities in Georgia: 1 cave, 3 cave springs (Fenolio et al. 2013), and 4 wells in 4 counties (Fig. 2; Purvis and Opsahl 2005). Our study increased the total number of occurrences to 46, and more than doubles the number of occurrences in Georgia, which now stands at 17 sites (Fig. 2). In particular, we documented new county records for Early, Miller, Mitchell, and Seminole counties. Dougherty Plain Cave Table 3. Effects of 4 predictor variables on cave crayfish mortality estimated by Bayesian model averaging. PP is the posterior probability that the coefficient is not equal to zero and that the predictor should be included in the model, AvC is the coefficient averaged across models, SD is the standard deviation of the averaged coefficient. * denotes key predictor variables that have >75% posterior probability that the coefficient (AvC) is not equal to zero. Predictor variable PP AvC ± SD Intercept* 100.0 -3.619 ± 2.508 Trap days 70.0 0.096 ± 0.094 Trap depth 29.7 -0.040 ± 0.092 Number of crayfish captured* 100.0 2.420 ± 1.346 Bait type (shrimp)* 93.6 -5.420 ± 3.527 Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 174 Crayfish is now known from 3 counties in Florida and 8 counties in Georgia. In addition, our samples established a new USGS HUC8 watershed record, the Spring Creek watershed (03130010). The species is now known from this and 5 other HUC8 watersheds: Chipola (03130012), Ichawaynochaway (03130009), Lower Choctawhatchee (03140203), Lower Flint (03130008), and St. Andrew–St. Joseph Bays (03140101). These new records help to fill in existing distribution gaps between the cluster of occurrences in Jackson County, FL, and previously documented occurrences in Georgia (Fig. 2). We estimated 2 geographic-range metrics used in conservation assessments based on locality data from documented occurrences before and after the current study. We calculated extent of occurrence (EOO) and area of occupancy (AOO) using the web-based program GeoCAT (Bachman et al. 2011; available at geocat. kew.org). EOO was calculated as a convex hull, which is the smallest polygon that contains all occurrences and has no internal angles exceeding 180°. We used a grid-cell size of 2 km (4 km2) to estimate AOO. Based on previous occurrences, we estimated EOO at 8463.4 km2 and AOO to 120 km2. With the addition of the 9 new localities, EOO increased slightly to 8600.7 km2 and AOO to 156 km2. Our Figure 6. Sketch of exopod of pleopod 2 from a female stygobitic isopod (Caecidotea sp. hobbsi group) trapped from Well 11J012 in Mitchell County, GA. Southeastern Naturalist 175 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 estimation of EOO is larger than that reported by NatureServe (1000–5000 km2; NatureServe 2015) but substantially smaller than EOO reported in the most recent IUCN Red List assessment (21,000 km2; Cordeiro et al. 2010). We reassessed the NatureServe global conservation rank for Dougherty Plain Cave Crayfish using the NatureServe Rank Calculator v3.186 (Faber-Langendoen et al. 2009). The inclusion of the additional localities documented in this study resulted in the suggested change of the global status rank from “imperiled” (G2; NatureServe 2016) to “vulnerable” (G3). The state status rank for Florida remained the same at “imperiled” (S2), while the state status rank for Georgia changed from “critically imperiled–imperiled” (S1S2) to “imperiled” (S2). Reassessment of the IUCN Red List conservation rank resulted in no change in the listed risk category, “least concern”. Although EOO and AOO meet “vulnerable” thresholds, there is no current evidence or trend data for continuing, observed, inferred, or projected decline in EOO, AOO, quality of habitat, number of occurrences, or subpopulations of Dougherty Plain Cave Crayfish. However, this species warrants continued monitoring because Dougherty Plain Cave Crayfish is known from only 46 localities, and, in Georgia, the Floridan aquifer system is identified as an at-risk aquifer (Nolan et al. 1998). We detected Dougherty Plain Cave Crayfish as deep as 40.6 m below the land surface, which is considerably deeper than previous reports. Purvis and Opsahl (2005) trapped this species at depths ranging from 14 to 26 m below the land surface. These depths are well within the range of depths where other subterranean crustaceans have been detected (reviewed in Fiser et al. 2014). The depth limit for Dougherty Plain Cave Crayfish is unknown, as are the environmental factors that impose such depth constraints on this species. However, levels of dissolved oxygen may not be limiting at depth (Fiser et al. 2014). Dougherty Plain Cave Crayfish can tolerate dissolved oxygen conditions as low as 3.6 mg/L (Caine 1978, Purvis and Opsahl 2005) and it possesses a metabolism lower than that of surface-dwelling relatives (Caine 1978). Thus, Dougherty Plain Cave Crayfish may occur at depths deeper that those reported in this and previous studies. Detection of other taxa In addition to Dougherty Plain Cave Crayfish, we collected a stygobiotic isopod in the genus Caecidotea (Isopoda: Asellidae) from a well in Mitchell County, GA. Stygobiotic isopods identified as Caecidotea sp. have been collected previously from the Floridan aquifer system in southwestern Georgia. A well driller collected a single specimen from the Wildmeade Plantation in Calhoun County (Opsahl and Chanton 2006), but this specimen was not identified to species group or species. Stygobitic Caecidotea sp. also were collected from Radium Springs in Dougherty County by cave divers (Opsahl and Chanton 2006, Opsahl et al. 2005). We identified the isopod collected from a well in Mitchell County as a member of the Caecidotea hobbsi species group, and it may possibly be C. hobbsi (Maloney) (Hobbs Cave Isopod). Caecidotea hobbsi is known from groundwater habitats, such as caves, springs, wells, and crayfish burrows, in several counties of northern Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 176 Florida, including Calhoun, Jackson, and Washington counties (Franz et al. 1994, Maloney 1939, Steeves 1964, Walsh 2001). The documented Florida localities of Hobbs Cave Isopod nearest to the well in Mitchell County, GA, are >100 km to the southwest. Hobbs Cave Isopod co-occurs with Dougherty Plain Cave Crayfish at several cave systems in the Marianna Lowlands of Jackson County, FL. A presumably disjunct population has also been documented 275 km to the north of the well in Mitchell County from a spring on the Emory University campus in DeKalb County, GA (Franz et al. 1994). Additional sampling of groundwater habitats and the collection of males are needed to positively determine whether Hobbs Cave Isopod or possibly an undescribed species occurs in the Floridan aquifer system in southwestern Georgia. This study failed to detect Georgia Blind Salamander, despite its documented co-occurrence with the Dougherty Plain Cave Crayfish at sites within both states of occurrence (e.g., Fenolio et al. 2013). It is possible that the trap design or bait used is unsuitable for sampling Georgia Blind Salamander. However, the trapping methods employed in this study have been successfully used to sample several species of groundwater Eurycea in Texas (Chippindale 2009). Detection probabilities of salamanders in groundwater environments are typically low (Bendik 2010, Krejca and Gluesenkamp 2007). For example, continuous weekly sampling of an open-hole monitoring well, known to be occupied by Eurycea of the subgenus Typhlomolge, resulted in a single collection event (n = 2 individuals) over a 9-month period (~270 trap days) (A.G. Gluesenkamp, unpubl. data). Similarly, McDermid et al. (2015) detected E. sosorum Chippindale et al. (Barton Springs Salamander), at a new site only after checking a bottle trap on 212 occasions, more than twice the number of trapping events of the present study. Our failure to detect salamanders may be because Dougherty Plain Cave Crayfish is a known predator of Georgia Blind Salamander (Fenolio et al. 2013 and citations therein). Any salamanders that entered the traps may have been consumed by crayfish before trap recovery. In addition, Georgia Blind Salamander may exhibit predator-avoidance behavior in response to chemical cues, as documented in the congeneric E. nana (Bishop) (San Marcos Salamander) and E. sosorum by Epp and Gabor (2008) and DeSantis et al. (2013), respectively. Therefore, capture success of salamanders may be improved by reducing the duration of trapping events and/or checking the traps with greater frequency and removing trapped crayfish. Considerations for future trapping efforts Several studies have demonstrated the efficacy of various methods of sampling invertebrates in groundwater wells (Boutin and Boulanouar 1983, Collins and Holsinger 1981, Hahn 2005, Hutchins and Orndorff 2009, Hutchins et al. 2010, McDermid et al. 2015, Purvis and Opsahl 2005, Sorenson et al. 2013). Even though groundwater wells have proven to be important sampling points for aquifer fauna (Longley 1978; Longley and Karnei 1978a, 1978b; Stejneger 1896), sampling of these wells is not widely practiced in North America. Our study demonstrated that deployment of baited bottle traps in groundwater wells can be an effective approach Southeastern Naturalist 177 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 to survey stygobiotic taxa, such as Dougherty Plain Cave Crayfish, especially in areas where surface access to groundwater is limited. The potential exists to further expand the known range of Dougherty Plain Cave Crayfish. All records to date are associated with the Floridan aquifer system, which extends an additional ~60 km to the northeast in Georgia and ~160 km to the west in a narrow band in southern Alabama (Figs. 2 and 3). In particular, the species may eventually be detected from groundwater wells developed in the Floridan aquifer system in the southeastern-most part of Alabama. Increased trapping-effort and experimentation with different types of bait may yield new occurrences of other groundwater species that may exist at low densities, such as Caecidotea isopods and the Georgia Blind Salamander (Fenolio et al. 2013). Future trapping efforts for Caecidotea should consider utilizing an increased number of small-sized drain holes (~1 mm) in the bottle traps to allow retention of smaller-bodied stygobiotic fauna. However, smaller holes may decrease water flow which could influence mortality risk of Cambarus (discussed below). Additional research is needed to understand the influence of trap-hole size on trapping efficiency and safety for various taxa occupying groundwater environs. Bait choice and amount of bait used may affect trapping results for Dougherty Plain Cave Crayfish, along with potential seasonal or trap-avoidance/attraction effects on capture and mortality rates. We used unsalted cashews and shrimp as bait, whereas similar sampling efforts in Texas used salted pistachio nuts and successfully captured both salamanders and invertebrates (A.G. Gluesenkamp, unpubl. data). Too much bait may create unfavorable conditions (e.g., reduced oxygen due to microbial activity during bait decomposition) within the confines of the trap. Indeed, a concern of this approach was the elevated proportion of crayfish trapped that were dead (53.1%) during this study. The most important factor positively related to crayfish mortality was the number of cave crayfish captured in the trap, which perhaps could be reduced with more frequent sampling, although the results of our BMA analysis suggested that trap duration was not a key indicator of cave-crayfish mortality. Although the exact cause of death is unknown, it is possible that decay of bait or dead individuals lowered dissolved oxygen levels below critical thresholds. In addition to feeding on bait, crayfish may feed on each other if too many individuals are trapped together or bait is completely consumed. Cannibalism has been documented in the laboratory for Georgia Blind Salamander (Fenolio et al. 2014). Regrettably, logistics and funding did not allow for more frequent and regular checking of traps when they were deployed. Regardless, we strongly encourage that future studies employing this or a similar well trapping method increase the frequency in which traps are checked. In addition, studies are needed to estimate detection probabilities and to examine factors causing detection heterogeneity using this trapping method. Moreover, comparative studies of bait efficacy in groundwater organism capture (e.g., salted/unsalted, amount of bait) could improve trapping success and decrease the likelihood of mortality of captured specimens. Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 178 Acknowledgments We thank Gary Holloway, Michael Peck, and John McCranie for assisting with field research. This research was authorized under Georgia Department of Natural Resources scientific collection permit no. CN 27743. Funding for this project was provided by a Joint Funding Agreement between the US Geological Survey and the Georgia Department of Natural Resources (Joint Funding Agreement #14ESGAMF0000063) and the San Antonio Zoo. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government. Literature Cited Bachman, S., J. Moat, A.W. Hill, J. de Torre, and B. Scott. 2011. Supporting Red List threat assessments with GeoCAT: Geospatial conservation assessment tool. Zookeys 150:117–126. Bendik, N.F. 2010. Jollyville Salamander status report. City of Austin Watershed Protection Department SR-11-10, Austin, TX. Boutin, C., and M. Boulanouar. 1983. Méthodes de capture de la faune stygobie: Expérimentation de différents types de pièges appâtés dans les puits de Marrakech. Bullettin de la Faculté des Sciences de Marrakech (section Sc. de la vie) 2:5–21. Caine, E.A. 1978. Comparative ecology of epigean and hypogean crayfish (Crustacea: Cambaridae) from northwestern Florida. American Midland Naturalist 99:315–329. Chippindale, P.T. 2009 Population genetic analysis of the Texas Blind Salamander, Eurycea rathbuni. Final Section 6 report submitted to Texas Parks and Wildlife Department. Texas Parks and Wildlife Department, Austin, TX. Collins, T.L., and J.R. Holsinger. 1981. Population ecology of the troglobitic isopod crustacean Antrolana lira Bowman (Cirolanidae). Proceedings of the 8th International Congress of Speleology 1:129–132. Cordeiro, J., K.A. Crandall, T. Jones, C. Skelton, and R.F. Thoma. 2010. Cambarus cryptodytes. The IUCN Red List of Threatened Species 2010: e.T3691A10023413. Available online at http://dx.doi.org/10.2305/IUCN.UK.2010-3.RLTS.T3691A10023413.en. Accessed 2 December 2015. DeSantis, D.L., D.R. Davis, and C.R. Gabor. 2013. Chemically mediated predator-avoidance in the Barton Springs Salamander (Eurycea sosorum). Herpetologica 69:291–297. Epp, K.J., and C.R. Gabor. 2008. Innate and learned predator-recognition mediated by chemical signals in Eurycea nana. Ethology 114:607–615. Faber-Langendoen, D., L. Master, J. Nichols, K. Snow, A. Tomaino, R. Bittman, G. Hammerson, B. Heidel, L. Ramsay, and B. Young. 2009. NatureServe conservation-status assessments: Methodology for assigning ranks. NatureServe, Arlington, VA. Fenolio, D.B., M.L. Niemiller, M. Levy, and B. Martinez. 2013. Conservation status of the Georgia Blind Salamander (Eurycea wallacei) from the Floridan aquifer system of Florida and Georgia. Reptiles and Amphibians Conservation and Natural History 20:97–111. Fenolio, D.B., M.L. Niemiller, and B. Martinez. 2014. Observations of reproduction in captivity by the Dougherty Plain Cave Crayfish, Cambarus cryptodytes (Decapoda: Astacoidea: Cambaridae). Speleobiology Notes 6:14–26. Fiser, C., T. Pipan, and D.C. Culver. 2014. The vertical extent of groundwater metazoans: An ecological and evolutionary perspective. Bioscience 64:971–979. Franz, R., J. Bauer, and T. Morris. 1994. Review of biologically significant caves and their faunas in Florida and South Georgia. Brimleyana 20:1–109. Southeastern Naturalist 179 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 Gluesenkamp, A.G., and J.K. Krejca. 2007. Assessing the status of Eurycea salamanders and aquifer invertebrates in northern Hays County, Texas. Report prepared for Texas Parks and Wildlife Department, Austin, TX. 15 pp. Gruver, B.J., and C. Coffey. 2015. Florida’s endangered and threatened species list. Florida Fish and Wildlife Conservation Commission, Tallahassee, FL. 11 pp. Hahn, H.J. 2005. Unbaited phreatic traps: A new method of sampling stygofauna. Limnologica 35:248–261. Hobbs, H.H., Jr. 1941. Three new Florida crayfishes of the subgenus Cambarus (Decapoda: Astacidae). American Midland Naturalist 26:110–121. Hobbs, H.H., Jr. 1981. The Crayfishes of Georgia. Smithsonian Contributions to Zoology 318:1–549. Hobbs, H.H., Jr. 1989. An illustrated checklist of the American crayfishes (Decapoda: Astacidae, Cambaridae, and Parastacidae). Smithsonian Contributions to Zoology 480:1–236. Hobbs, H.H., Jr., H.H. Hobbs, III, and M.A. Daniel. 1977. A review of the troglobitic decapod crustaceans of the Americas. Smithsonian Contributions to Zoology 244:1–183. Hoeting, J.A., D. Madigan, A.E. Raftery, and C.T. Volinsky. 1999. Bayesian model averaging: A tutorial. Statistical Science 14:382–427. Hutchins, B., and W. Orndorff. 2009. Effectiveness and adequacy of well sampling using baited traps for monitoring the distribution and abundance of an aquatic subterranean isopod. Journal of Cave and Karst Studies 71:193–203. Hutchins, B., D.W. Fong, and D.B. Carlini. 2010. Genetic population structure of the Madison Cave Isopod, Antrolanalira (Cymothoida: Cirolanidae) in the Shenandoah Valley of the eastern United States. Journal of Crustacean Biology 30:312 –322. Krejca, J.K., and A. Gluesenkamp. 2007. Mark–recapture study of Eurycea rathbuni at two sites in San Marcos, Texas. Section 6 report prepared for Texas Parks and Wildlife Department. 31 March 2007. Austin, TX. 17 pp. Lewis, J.J. 1988. The systematics, zoogeography, and life history of the troglobitic isopods of the Interior Plateaus of the eastern United States. Ph.D. Dissertation. University of Louisville, Louisville, KY. 281 pp. Longley, G. 1978. Status of Typhlomolge (= Eurycea) rathbuni, the Texas Blind Salamander. Endangered Species Report (2). US Fish and Wildlife Service Region 2, Albuquerque, NM. 78 pp. Longley, G., and H. Karnei. 1978a. Status of Satan eurystomus Hubbs and Bailey, the Widemouth Blindcat. Contract No. 14-16-0002-035. US Fish and Wildlife Service, Albuquerque, NM. 48 pp. Longley, G., and H. Karnei. 1978b. Status of Trogloglanis pattersoni Eigenmann, the Toothless Blindcat. Contract No. 14-16-0002-77-035. US Fish and Wildlife Service, Albuquerque, NM. 54 pp. Mac Nally, R. 2002. Multiple regression and inference in ecology and conservation biology: Further comments on identifying important predictor variables. Biodiversity and Conservation 11:1397–1401. Maloney, J.O. 1939. A new cave isopod from Florida. Proceedings of the United States National Museum 86:457–459. McDermid, K., P. Sprouse, and J. Krejca. 2015. Eurycea sosorum (Barton Springs Salamander). Herpetological Review 46:556–557. Means, D.B. 1992. Georgia Blind Salamander, Haideotriton wallacei Carr. Pp. 49–53, In P. Moler (Ed.). Rare and Endangered Biota of Florida, Vol. III. Amphibians and reptiles. University of Florida Press, Gainesville, FL. 291 pp. Southeastern Naturalist D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 180 Morris, T.L. 2006. A biological inventory of aquifer caves in Florida with special emphasis on troglobitic crustaceans and salamanders. Final report for research grant agreement S0057 (Springs Program). Florida Department of Environmental Regulation, Tallahassee, FL. Nally, R.M., L.D. Vries, and J.R. Thompson. 2008. Are replanted floodplain forests in southeastern Australia providing bird-biodiversity benefits? Restoration Ecol ogy 18:85–94. NatureServe. 2016. NatureServe Explorer: An online encyclopedia of life (web application). Version 7.1. NatureServe, Arlington, VA. Available online at http://explorer.natureserve. org. Accessed 9 February 2016. Niemiller, M.L., D.B. Fenolio, and K.S. Zigler. 2012. The obligate cave fauna of Georgia. Bulletin of the Georgia Speleological Survey 2012:6–12. Nolan, B.T., B.C. Ruddy, K.J. Hitt, and D.R. Helsel. 1998. A national look at nitrate contamination of ground water. Water Conditioning and Purification 39:76–79. Opsahl, S.P., and J.P. Chanton. 2006. Isotopic evidence for methane-based chemosynthesis in the Upper Floridan aquifer-system food web. Oecologia 150:89–96. Opsahl, S.P., S.E. Chapal, and C.K. Wheeler. 2005. Using stream-gauge data to quantify surface water/groundwater exchanges between the Upper Floridan aquifer system and the Lower Flint River, Georgia, USA, 1989–2003. Pp. 764–768, In K.J. Hatcher. (Ed.). Proceedings of the 2005 Georgia Water Resources Conference. Institute of Ecology, University of Georgia, Athens, GA. Purvis, K.M., and S.P. Opsahl. 2005. A novel technique for invertebrate trapping in groundwater well identifies new populations of the troglobitic crayfish, Cambarus cryptodytes, in southwest Georgia, USA. Journal of Freshwater Ecology 20:361–365. R Core Team. 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at https://www.R-project. org. Accessed 7 March 2017. Raftery, A.E., D. Madigan, and J.A. Hoeting. 1997. Bayesian model averaging for linear regression models. Journal of the American Statistical Association 92:179–191. Raftery, A.E. J. Hoeting, C. Volinsky, I. Painter, and K. Yeung. 2009. BMA: Bayesian model averaging. R package version 3.12. Available online at http://CRAN.R-project. org/package=BMA. Accessed 7 March 2017. Selwood, K., R.M. Nally, and J.R. Thomson. 2009. Native-bird breeding in a chronosequence of revegetated sites. Oecologia 159:435–446. Skelton, C. 2008. Dougherty Plain Cave Crayfish. Available online at http://www.georgiawildlife. com/sites/default/files/uploads/wildlife/nongame/pdf/accounts/invertebrates/ cambarus_cryptodytes.pdf. Accessed 2 December 2015. Sorensen, J.P.R., L. Maurice, F.K. Edwards, D.J. Lapworth, D.S. Read, D. Allen, A.S. Butcher, L.K. Newbold, B.R. Townsend, and P.J. Williams. 2013. Using boreholes as windows into groundwater ecosystems. PLoS One 8:e70264. Steeves, H.R. 1964. The troglobiotic asellids of the United States: The hobbsi group. American Midland Naturalist 71:445–451. Stejneger, L. 1896. Description of a new genus and species of blind tailed-batrachians from the subterranean waters of Texas. Proceedings of the United States National Museum 18:619–621. Sutton, B., and K. Relyea. 1971. Notes on the Georgia Blind Cave Salamander, Haideotriton wallacei (Amphibia: Plethodontidae). (Abstract) ASB Bulletin 18:58. Taylor, C.A., G.A. Schuster, J.E. Cooper, R.J. DiStefano, A.G. Eversole, H.H. Hobbs III, H.W. Robison, C.E. Skelton, and R.F. Thoma. 2007. A reassessment of the conservation status of crayfishes of the United States and Canada after 10+ years of increased awareness. Fisheries 32:372–389. Southeastern Naturalist 181 D.B. Fenolio, M.L. Niemiller, A.G. Gluesenkamp, A.M. McKee, and S.J. Taylor 2017 Vol. 16, No. 2 Thomson, J.R., R.M. Nally, E. Fleischman, and G. Horrocks. 2007. Predicting bird-species distributions in reconstructed landscapes. Conservation Biology 21:752–766. US Geological Survey. 2015. Digital surfaces and thicknesses of selected hydrogeologic units of the Floridan aquifer system in Florida and parts of Georgia, Alabama, and South Carolina. US Geological Survey Data Series 926. Available online at http://pubs.usgs. gov/ds/0926/. Accessed 7 March 2017. Viallefont, V., A.E. Raftery, and S. Richardson. 2001. Variable selection and Bayesian model averaging in case-control studies. Statistical Medicine 20:3215–3230. Walsh, C., and R. Mac Nally. 2015. Hier.part. R package version 1.0-4. Available online at http://CRAN.R-project.org/package=hier.part. Accessed 7 March 2017. Walsh, S.J. 2001. Freshwater macrofauna of Florida karst habitats. Pp. 78–88, In E. Kuniansky. (Ed.). US Geological Survey Karst Interest Group Proceedings, St. Petersburg, FL. USGS Water-Resources Investigations Report 01-4011. US Geological Survey, Denver, CO. Wintle, B.A., M.A. McCarthy, C.T. Volinsky, and R.P. Kavanagh. 2003. The use of Bayesian model averaging to better represent uncertainty in ecological models. Conservation Biology 17:1579–1590.