Distribution, habitat associations, and conservation status updates for the pilose crayfish Pacifastacus gambelii (Girard, 1852) and Snake River pilose crayfish Pacifastacus connectens (Faxon, 1914) of the western United States

Background

North America is home to the majority of the world’s crayfish diversity, with 414 described species (Richman et al., 2015; Crandall & De Grave, 2017). However, many of these North American crayfishes are highly imperiled and at risk of extinction. Taylor et al. (2007) estimated that 48% of North American crayfishes were at some level of extinction risk, whereas a more recent International Union for the Conservation of Nature (IUCN) assessment placed 32% of North American crayfishes at risk of extinction (Richman et al., 2015). The western United States (US) is more species poor for freshwater crayfishes than the southeastern US, but its endemic genus Pacifastacus is representative of the conservation and management challenge for crayfishes globally. Of the Pacifastacus crayfishes, one is a globally cosmopolitan invasive species (the signal crayfish Pacifastacus leniusculus; Dana, 1852), one species is believed extinct (the sooty crayfish Pacifastacus nigrescens; Stimpson, 1857), another is listed as Endangered under the US Endangered Species Act (the Shasta crayfish Pacifastacus fortis; Faxon, 1914), and two other species, the Snake River pilose crayfish Pacifastacus connectens (Faxon, 1914) and the pilose crayfish Pacifastacus gambelii (Girard, 1852), are effectively unstudied (Larson & Williams, 2015). Currently, P. connectens is listed in the IUCN Red List database as Data Deficient and P. gambelii is listed as Least Concern (Richman et al., 2015), but no distributional or conservation status studies have been conducted for either species (Larson & Olden, 2011; Larson & Williams, 2015). Given that two species of their genus have gone extinct or been listed as Endangered, we sought to evaluate the distribution, habitat associations, and conservation status of the pilose crayfishes P. connectens and P. gambelii.

Pacifastacus connectens and P. gambelii belong to the subgenus Hobbsastacus, which includes the extinct P. nigrescens and P. fortis, relative to the subgenus Pacifastacus, which includes only P. leniusculus and its three recognized subspecies (Larson & Williams, 2015). P. connectens was split from P. gambelii, first as a subspecies by Faxon (1914) and subsequently as its own species by Hobbs (1972). Both crayfishes are morphologically unique relative to other members of their genus owing to the presence of patches of setae or hairs on their chelae, whereas P. connectens is differentiated from P. gambelii by characteristics including an acute (narrow) rather than obtuse (broad) rostrum (Fig. 1). Recent phylogenetic species delimitation analysis has identified some ambiguity within the Hobbsastacus subgenus (Larson et al., 2016); as work on their taxonomic relationship continues, we largely consider both species here combined as the “pilose crayfishes” given their shared taxonomic history and morphological similarity. To date, no studies have investigated the life history or ecology of either pilose crayfish species, although Koslucher & Minshall (1973) included P. gambelii in a study on stream food webs from southern Idaho. Further, historical records for the pilose crayfishes appear to indicate a habitat preference for groundwater-dominated springs with small upstream catchments (Miller, 1960; Hubert, 2010).

Data regarding the distributions of P. connectens and P. gambelii are also limited. Larson & Olden (2011) proposed the pilose crayfishes as endemic to the middle and upper Snake River drainage and adjacent closed or endorheic desert basins (e.g., the Bonneville Basin) of Idaho, Nevada, Oregon, Utah, and Wyoming (Fig. 2). Past guides or keys to North American crayfishes (e.g., Hobbs, 1972) likely over-stated the distribution of these two crayfishes, particularly P. gambelii, per the review of Larson & Williams (2015), although more widespread distributional surveys for these crayfishes throughout western North America would be useful. Within the range proposed by Larson & Olden (2011) and Larson & Williams (2015) for each crayfish, P. connectens generally occurs below Shoshone Falls, a major biogeographic break in the Snake River drainage, and in the neighboring Harney Basin of eastern Oregon. Alternatively, P. gambelii occurs above Shoshone Falls in the Snake River and its tributaries, and in the northern Bonneville Basin, although exceptions in this distributional pattern between the two species have been reported from historical records (Fig. 2). These erratic distributional records for each species may reflect either misidentifications in historical records or a more complex distributional pattern for each species than proposed by past work like Larson & Williams (2015), and further supports our decision to consider the two species combined here rather than separately.

Like many freshwater crayfishes, P. connectens and P. gambelii could be impacted by a number of threats and stressors within their native range (Richman et al., 2015; Bland, 2017). These include risk of displacement by invasive crayfishes, including the virile crayfish Faxonius virilis (Hagen, 1870) and the red swamp crayfish Procambarus clarkii (Girard, 1852), which have been reported as introduced in the native range of the pilose crayfishes (Johnson, 1986; Hubert, 1988; Clark & Lester, 2005); reviewed in Larson & Olden (2011). Further, the congeneric crayfish P. leniusculus was not known from the native range of P. connectens or P. gambelii during the earliest historical records for these species (e.g., Miller, 1960), but could represent a “native invader” (e.g., Carey et al., 2012) as it has seemingly spread inland into this region over recent decades from its more coastal native range (Larson et al., 2012). Competitive displacement by P. leniusculus was implicated in both the extinction of P. nigrescens and US Endangered Species Act listing of P. fortis, and P. leniusculus could pose a similar threat to the Hobbsastacus pilose crayfishes (Bouchard, 1977; Light et al., 1995). Invasive populations of P. leniusculus have also been attributed as a cause of declines of native European crayfish of the family Astacidae (Chucholl & Schrimpf, 2016; Maguire et al., 2018). Additionally, freshwaters of the native range of the pilose crayfishes have experienced impacts due to livestock overgrazing, flow regime modification by dams and irrigation development, and water quality impairments from agricultural and urban runoff (Belsky, Matzke & Uselman, 1999; Anderson & Woosley, 2005; Caldwell et al., 2012). In particular, the Snake River Plain has been identified as a region of hydrologic impairment and poor water quality resulting from agricultural land use (Hill et al., 2016; Thornbrugh et al., 2017). Such land use changes and their effects on water quality and freshwater habitats have similarly been attributed as contributing to native crayfish declines elsewhere (Chucholl & Schrimpf, 2016; Chucholl, 2017).

We sought to model the historical distribution and habitat associations of P. connectens and P. gambelii combined in the western US and compare these predictions to their current distribution from field sampling. We first developed a species distribution model (SDM) using historical occurrence data for P. connectens and P. gambelii to predict the distributions and habitat associations for these crayfishes using GIS environmental data layers (Domisch, Amatulli & Jetz, 2015). We then conducted field sampling in the presumed native range of these crayfishes to characterize their current distributions in comparison to both their historical occurrence records and predictions of suitable habitat by our SDM. Finally, where our SDM model predictions diverged from results of our field sampling, we used a single classification tree on factors like the presence of invasive crayfishes and GIS layers on possible stream habitat impairment to explore and explain these misclassifications. Cumulatively, our work should help to better define the historical distribution and habitat associations for the pilose crayfishes P. connectens and P. gambelii, as well as their current conservation status.

Methods

We evaluated the historical and current distributions and habitat associations of the pilose crayfishes P. connectens and P. gambelii using an SDM on GIS environmental data layers, along with contemporary field sampling (Fig. 3). We first used historical occurrence records for the pilose crayfishes to generate an SDM describing their past distribution and habitat associations. Upon developing this SDM, we sampled study sites predicted by our model to be suitable and unsuitable for the pilose crayfishes throughout their native range to characterize their current distribution. In relating contemporary presences or absences of P. connectens and P. gambelii to modeled predictions of habitat suitability, we anticipated that the SDM would misclassify some sampled sites. For example, false positives are places where the SDM predicted pilose crayfish to occur but we failed to find them in our field sampling. We then sought to explain such true and false positives using a subsequent, single classification tree using information like presence of invasive crayfishes at sampled sites and habitat conditions or impairment (Fig. 3).

Results

Our boosted regression tree model classified combined P. connectens and P. gambelii historical occurrences relative to background points with a moderate AUC of 0.824 based on testing data withheld in ten-fold cross-validation (Fig. 4). Our most important environmental variables from our primary model included upstream temperature seasonality (relative importance of 13%), flow accumulation (12%), annual upstream precipitation (11.6%), upstream precipitation seasonality (11%), flow length (10%), and average slope (8%). Based on our SDM, the pilose crayfishes had a mostly negative relationship with the smallest streams in our study region, as measured by environmental variables including annual upstream precipitation, average slope, and flow length (Fig. 5). However, flow accumulation showed a positive association with some very small streams, with lower values of flow accumulation predicting a high likelihood of pilose crayfish occurrence. The pilose crayfishes also had a negative relationship with high annual upstream temperature and precipitation seasonalities.

We found the pilose crayfishes at 20 (12%) of the total 163 sites we sampled, with P. connectens and P. gambelii each at 10 (Fig. 6; Table S2). We found the native invader P. leniusculus at 29 sites (18%) and the invasive virile crayfish F. virilis at 22 sites (13%), with only one site where any two crayfish species occurred sympatrically (F. virilis and P. gambelii; Fig. 6). Of the 50 historical sites we sampled, we found the pilose crayfishes at only 12 (24%), but we found F. virilis at 12 (24%) and P. leniusculus at 6 (12%). Our boosted regression tree model predicted presences and absences of pilose crayfishes from field sampling with relatively low success, based on a Cohen’s Kappa (K) of 0.14. Of the 163 sites we sampled, 73 (45%) were classified as suitable pilose crayfish habitat from our boosted regression tree model. Overall, our model correctly predicted 14 out of 20 (70%) presences for these native crayfishes (true positives), but misclassified 6 (30%) presences as unsuitable habitat for these crayfishes (false negatives). Similarly, our model correctly predicted 84 out of 143 (59%) absences (true negatives), but misclassified 59 (41%) absences as suitable habitat for P. connectens and P. gambelii (false positives).

Our single classification tree differentiated false positives from true positives relatively well with a Cohen’s K of 0.64 (Fig. 7). False positives were more likely to occur at sites where invasive crayfish were present; at sites with either very poor or very good stream benthic community conditions; at sites where we used baited trapping rather than timed searches; and at sites with both greater hydrologic regulation as well as lower upstream agricultural land cover (Fig. 7).

Discussion

We modeled suitable habitats for the pilose crayfishes P. connectens and P. gambelii based on their historical occurrence records using boosted regression trees and a series of environmental variables. We found that these crayfishes occurred historically in larger streams and rivers with lower upstream precipitation seasonality, low to intermediate upstream temperature seasonality, and higher annual upstream precipitation. We interpret these results as suggesting that the pilose crayfishes did not generally occur in high elevation, montane streams with extreme seasonality like the Uinta and Teton mountains. When related to contemporary, conventional field sampling, we found that the pilose crayfishes had seemingly experienced large population and range declines. For example, we found the pilose crayfishes at only 24% of the 50 historical occurrence records we sampled, and at only 19% of sites that our SDM predicted as suitable for them. In many cases, these declines appear attributable to displacement by the invasive crayfishes F. virilis and P. leniusculus and degraded stream benthic community condition, but choice of sampling method may also have affected the frequency of false positives we observed for the pilose crayfishes relative to modeled habitat suitability. Regardless, the pilose crayfishes seemingly require increased management and conservation attention, because they may be at risk of the types of population declines or even extinction that have been observed for similar crayfishes of the subgenus Hobbsastacus (Bouchard, 1977; Light et al., 1995).

We found from our SDM that the pilose crayfishes P. connectens and P. gambelii occurred historically in larger streams and rivers in less extreme environments, featuring moderate to low temperature seasonality, low precipitation seasonality, and moderate slopes. Based on this model, the pilose crayfishes did not generally occur in the absolute smallest streams in our study region, as measured by predictors like flow accumulation, annual upstream precipitation, and flow length. Previous studies have found other crayfish species to either favor or disfavor smaller and potentially intermittent streams due to different tolerances to abiotic factors like stream drying and biotic factors like longitudinally structured predator communities (Flinders & Magoulick, 2005; Creed, 2006). However, as an exception to our finding that pilose crayfishes did not historically occur in the smallest streams, we did find some positive association between these crayfishes and the absolute smallest streams in our region as measured by flow accumulation. This likely reflects the known tendency for these crayfishes to occur in some small, groundwater-dominated springs with minimal upstream surface watersheds (Miller, 1960; Hubert, 2010). Our contemporary field sampling similarly supported an association of the pilose crayfishes with some groundwater-dominated spring habitats (Fig. 8), which parallels habitat use of the similar and endangered P. fortis in northern California (Light et al., 1995). These isolated spring systems should perhaps be priorities for pilose crayfish conservation, as they have represented strongholds against displacement by invasive crayfishes for P. fortis (Cowart et al., 2018).

The pilose crayfishes also showed a negative relationship to streams with high upstream temperature and precipitation seasonalities, which reflect those streams and rivers draining high elevation mountain ranges in our study region, where winters are extremely wet and cold relative to warm and dry summers. Such locations likely have high spring and summer stream discharge owing to snowmelt-dominated flow regimes, as well as lower stream temperatures relative to valley bottom streams (Reidy Liermann et al., 2012). Invasive P. leniusculus experienced declines in abundance following high flow years in the similar Sierra Nevada mountains of California (Light, 2003), and the congeneric pilose crayfishes may be similarly intolerant of higher stream flows or discharge associated with ultra-snowmelt systems. Despite this, we did find a positive association between the pilose crayfishes and moderate upstream slope, and these crayfishes may do better in slightly higher gradient streams that maintain the type of cobble rock substrate that many crayfish species prefer as habitat (Garvey et al., 2003; Nyström et al., 2006). Our SDM revealed a number of potential habitat associations for the pilose crayfishes based on historical occurrence records at a relatively coarse 1 km² spatial grain, but much more work needs to be done in order to understand the habitat preferences of these crayfishes from micro-habitat (e.g., 1 m²) to reach scales (e.g., 100 m; Flinders & Magoulick, 2007; Wooster, Snyder & Madsen, 2012). Such finer-grain habitat work may, in part, clarify the frequency of false positives we observed for these crayfishes when comparing SDM predictions to contemporary field sampling.

Overall, our SDM on historical occurrence records for P. connectens and P. gambelii predicted contemporary distributions for these crayfishes with relatively low success in comparison to our field sampling, with many false positives but comparatively few false negatives. Because false positives may represent range declines for the pilose crayfishes, whereas false negatives were seemingly locations where GIS data simply did not reflect in-stream conditions well (e.g., groundwater springs; Fig. 8), we sought to explain false positives relative to true positives. We did this by using a single classification tree on a series of predictors either related to factors potentially causing range declines for the pilose crayfishes based on past studies in other crayfish species (Twardochleb, Olden & Larson, 2013; Richman et al., 2015), or predictors related to possible differences in detection probabilities between our field sampling methods (Larson & Olden, 2016). We found that the best explanation for false positives for the pilose crayfish was presence of an invasive crayfish species at the site. This is consistent with many past studies which have found displacement by invasive crayfishes to be a leading driver of native crayfish population declines (Lodge et al., 2000; Pintor, Sih & Bauer, 2008), and is consistent with causes of imperilment or extinction for other Pacifastacus crayfishes (Bouchard, 1977; Light et al., 1995), as well as native crayfishes of the family Astacidae in Europe (Chucholl & Schrimpf, 2016; Maguire et al., 2018). The second best explanation for false positives for the pilose crayfishes was highly degraded stream benthic community condition; these crayfishes have seemingly experienced range declines at locations where stream communities are the most impaired, including many lower elevation valley bottoms which have experienced high agricultural and urban development in this region (Hill et al., 2016). Water quality impairment from agricultural and urban land use has similarly been attributed as a cause for native crayfish declines in regions like Europe (Chucholl & Schrimpf, 2016; Chucholl, 2017). We also found that false positives for the pilose crayfishes were more likely to occur at locations where we sampled by baited trapping, rather than locations where we conducted timed searches. Different sampling methods can have different detection probabilities for crayfishes across habitat types (Larson & Olden, 2016), and in this case, we routinely only collected one to two crayfish with four to six baited traps effort overnight, whereas timed searches routinely collected higher numbers of crayfish over an hour of effort (Fig. S3). As such, we recommend that future studies focused on the pilose crayfishes use timed searches where possible, and if requiring the use of baited trapping, increase trap effort (number of traps) to improve detection probabilities with this method. Finally, false positives were associated with some habitat variables that we cannot necessarily explain as being associated with likely range or population declines for the pilose crayfishes. Specifically, false positives were associated with some sites of very high stream benthic community condition (unimpaired) and were also associated with sites with low agricultural land cover. Again, we propose that better understanding of micro- to reach-scale habitat associations for the pilose crayfishes might improve our understanding of some of these false positives where we failed to find these species at places predicted suitable for them (Flinders & Magoulick, 2007; Wooster, Snyder & Madsen, 2012).

Importantly, our finding of potentially large range declines for P. connectens and P. gambelii is dependent not only on comparison to modeled suitable habitat from an SDM, but also direct comparison to historical occurrence sites that we resampled. Our SDM estimated an 80% range decline for the pilose crayfishes, whereas comparison to the 50 historical sites we re-sampled found a similar 76% range decline (63% for P. connectens and 85% for P. gambelii). We found the pilose crayfishes at only 24% of the historical sites we resampled, and in another parallel to our SDM and single classification tree results, invasive crayfishes again appeared to be a major driver of this range decline. Thirty-six percent of the 50 historical sites that we resampled were instead occupied by invasive crayfishes, with only one site where a native crayfish species (P. gambelii) occurred in sympatry with an invasive crayfish species (F. virilis). Per IUCN extinction risk assessments, range declines of ≥70% over 10 years or three generations qualify for Endangered status, whereas range declines of ≥50% over the same time periods qualify for Vulnerable status (IUCN Species Survival Commission, 2012). We do not necessarily know the rate at which pilose crayfishes have experienced population declines or range retractions, but propose that neither of the pilose crayfishes are necessarily secure from some extinction risk due to impacts of invasive crayfishes or other stressors associated with habitat loss or degradation. We recommend that state, federal, and international agencies consider elevated conservation status categories for both pilose crayfishes.

Our finding of large apparent declines in the distribution of the pilose crayfishes suggests urgent need for management, conservation, and research of these crayfishes. The presence of invasive crayfishes in particular seems strongly related to declines or local extirpations of P. connectens and P. gambelii. Accordingly, efforts to prevent the further introduction and spread of invasive crayfishes like F. virilis, or the “native invader” P. leniusculus, should be immediately implemented, and may include educational outreach or regulatory change and enforcement to prohibit these organisms from the live animal trade (Larson & Olden, 2011; Lodge et al., 2016). In areas where F. virilis or P. leniusculus are already present, management and maintenance of existing dispersal barriers such as dams and waterfalls may keep these invaders from spreading further and help to conserve existing pilose crayfish populations (Kerby et al., 2005; Fausch et al., 2009). In addition, where local conditions allow (e.g., small groundwater springs; Fig. 8), construction and maintenance of new dispersal barriers might be considered to protect extant P. connectens and P. gambelii populations (Cowart et al., 2018). Range declines of the pilose crayfishes were also seemingly associated with degraded stream benthic community condition (Hill et al., 2016). Management and regulation of point and nonpoint sources of water pollution or sedimentation may help to prevent current pilose crayfish habitat from also becoming highly degraded (Allan, 2004; Novotny & Smith, 2004; Strayer, 2006). Our SDM suggests that the pilose crayfishes most typically occur in the types of larger, low elevation, valley bottom streams that are at most risk of degradation from land use in our study region (Larson & Olden, 2011; Larson & Williams, 2015), and as such, persistence of these crayfishes is likely dependent on good management practices for water quality and in-stream habitat (Bilotta & Brazier, 2008; Strayer & Dudgeon, 2010).

We conclude by emphasizing that our study is the first dedicated to the ecology and distribution of the pilose crayfishes, but further basic distributional and ecological information is urgently needed to support the conservation of these species. We are relatively confident that we have sampled within the true historical range for both crayfishes, but aberrant occurrence records for each species across the larger western US merits investigation (Larson & Williams, 2015). Pacifastacus connectens and P. gambelii would certainly benefit from additional biological and ecological information, including life history studies (Moore, DiStefano & Larson, 2013), investigations of ecological interactions with other organisms, particularly invasive crayfishes (Usio, Konishi & Nakano, 2001; Pintor, Sih & Bauer, 2008), and habitat selection and use at finer grains than we could consider here (Flinders & Magoulick, 2007). We hope that our study will provide a baseline and motivation for future inquiry and conservation intervention for these interesting but minimally studied crayfishes.

Supplemental Information