Contributed Paper Latent Extinction and Invasion Risk of Crayfishes in the Southeastern United States

ERIC R. LARSON∗ AND JULIAN D. OLDEN School of Aquatic and Fishery Sciences, University of Washington Box 355020, Seattle, WA 98195-5020, U.S.A.

Abstract: Crayfishes are both a highly imperiled invertebrate group as well as one that has produced many invasive species, which have negatively affected freshwater ecosystems throughout the world. We performed a trait analysis for 77 crayfishes from the southeastern United States in an attempt to understand which biological and ecological traits make these species prone to imperilment or invasion, and to predict which species may face extinction or become invasive in the future. We evaluated biological and ecological traits with principal coordinate analysis and classification trees. Invasive and imperiled crayfishes occupied dif- ferent positions in multivariate trait space, although crayfishes invasive at different scales (extraregional vs. extralimital) were also distinct. Extraregional crayfishes (large, high fecundity, habitat generalists) were most distinct from imperiled crayfishes (small, low fecundity, habitat specialists), thus supporting the “two sides of the same coin” hypothesis. Correct classification rates for assignment of crayfishes as invasive or imperiled were high (70–80%), even when excluding the highly predictive but potentially confounding trait of range size (75–90%). We identified a number of species that, although not currently listed as imperiled or found outside their native range, possess many of the life-history and ecological traits characteristic of currently invasive or imperiled taxa. Such species exhibit a high latent risk of extinction or invasion and consequently should be the focus of proactive conservation or management strategies. Our results illustrate the utility of trait-based approaches for taxonomic groups such as invertebrates, for which detailed species-specific data are rare and conservation resources are chronically limited.

Keywords: crayfish conservation, ecological traits, invasion ecology, life-history traits Riesgo de Extincion´ e Invasion´ de Langostinos en el Sureste de Estados Unidos Resumen: Los langostinos son un grupo de invertebrados en alto riesgo as´ı como uno que ha producido muchas especies invasoras, que han afectado negativamente a ecosistemas dulceacu´ıcolas en todo el mundo. Realizamos un analisis´ de tendencias para 77 especies de langostino del sureste de Estados Unidos en un intento por entender cuales´ atributos biologicos´ y ecologicos´ que hacen que susceptibles a estas especies al riesgo o la invasion.´ Evaluamos atributos biologicos´ y ecologicos´ con analisis´ de coordenadas principales y arboles´ de clasificacion.´ Los langostinos invasores y en peligro ocuparon diferentes posiciones en el espacio multivariado, aunque los langostinos invasores en diferentes escalas (extraregional vs. extral´ımite) tambi´en fueron distintos. Los langostinos extraregionales (grandes, fecundidad alta, generalistas de habitat)´ fueron muy diferentes de los langostinos en peligro (pequenos,˜ fecundidad baja, especialistas de habitat),´ por lo tanto soportando la hipotesis´ de las “dos caras de la misma moneda.” Las tasas de clasificacion´ correcta para la asignacion´ de los langostinos como invasores o en peligro fueron altas (70–80%), aun cuando se excluyo´ el atributo altamente predictivo pero potencialmente confundido del rango de tamano˜ (75–90%). Identificamos un numero´ de especies que, aunque no enlistados actualmente como en peligro o fuera de su rango de distribucion´ nativo, poseen muchos de atributos de historia de vida o ecologicos´ caracter´ısticos de taxa actualmente invasores o en peligro. Tales especies exhiben un riesgo latente alto de extincion´ o invasion´ y consecuentemente deber´ıan ser el foco de estrategias de conservacion´ o manejo proactivas. Nuestros resultados

∗email [email protected] Paper submitted April 16, 2009; revised manuscript accepted October 26, 2009. 1099 Conservation Biology, Volume 24, No. 4, 1099–1110 C 2010 Society for Conservation Biology DOI: 10.1111/j.1523-1739.2010.01462.x 1100 Crayfish Extinction and Invasion Risk ilustran la utilidad de los m´etodos basados en atributos para grupos taxonomicos´ como los invertebrados, para los cuales son raros los datos detallados y los recursos de conservacion´ estan´ cronicamente´ limitados.

Palabras Clave: atributos de historia de vida, atributos ecologicos,´ conservacion´ de langostinos, ecolog´ıa de invasion´

Introduction approaches may be especially powerful for taxonomic groups such as invertebrates, where many species are Freshwater crayfishes are a globally diverse group of in- poorly understood and resources for conservation initia- vertebrates (Crandall & Buhay 2008) and represent one of tives are scarce (Strayer 2006). the largest aquatic faunal groups in the United States, with Here we provide the first investigation of how biolog- over 360 known species (Taylor et al. 2007). Habitat loss, ical and ecological traits may predispose crayfishes to water-quality impairment, and biotic interactions with in- both imperilment and invasion. Our goal was to priori- vasive crayfish collectively threaten crayfishes through- tize native crayfishes most in need of conservation efforts, out their native ranges (Lodge et al. 2000; Jones et al. identify crayfishes most likely to become invasive in the 2007). As a result, crayfishes are second only to fresh- future, and elucidate the potential mechanisms contribut- water mussels as the most-threatened taxonomic group ing to both the imperilment and invasion processes. In in North America (Wilcove & Master 2005). Recent es- doing so we quantified the degree of latent invasion or timates indicate that 48% of North American crayfish extinction risk (sensu Cardillo et al. 2006) by examining species are at risk of extinction, although at present, the discrepancy between a species’ current status and only four species are listed under the U.S. Endangered its expected status predicted by its suite of biological Species Act (Taylor et al. 2007). If current trends persist, and ecological traits. We also interpreted our findings extinction rates of crayfishes are anticipated to increase in the context of the “two sides of the same coin” de- by more than an order of magnitude, exceeding those of bate about whether traits associated with invasion are freshwater fishes and amphibians (Riccardi & Rasmussen the inverse of those associated with imperilment (Brad- 1999). shaw et al. 2008; Jeschke & Strayer 2008; Blackburn & Crayfishes are important polytrophic consumers in Jeschke 2009). Given that time and resources for con- temperate freshwater environments and play a central ducting detailed species assessments are limited for cray- ecological role by providing a direct link from primary fishes, identifying reliable and easily measured indicators production and detrital-based food webs to fish and ter- of extinction and invasion risk will be extremely valuable restrial predators (Momot 1995; Nystrom¨ et al. 1996). for targeted and proactive conservation strategies. Consequently, the loss (extinction) or gain (invasion) of crayfishes can substantially affect community structure and ecosystem processes (e.g., Rodriguez et al. 2005; Pin- tor & Soluk 2006; Matsuzaki et al. 2009). The disconnect Methods between the ecological importance and high degree of imperilment of crayfishes and the modest conservation Study Organisms and Assignment of Invasive and Imperiled Status efforts dedicated to them highlights the need for proac- tive and cost-effective management strategies to ensure We examined 77 crayfish species and subspecies native their preservation in the face of continued environmental to the states of Missouri and Kentucky (U.S.A.). These change. states vary in topology from mountains to lowland river Faced with such challenges for other taxonomic valleys and are characterized by abundant karst regions groups, conservation biologists have turned to using pre- with cave complexes, providing a representative cross- dictive suites of biological and ecological traits to pro- section of crayfish habitats and corresponding life histo- vide estimates of species extinction risk (Pimm et al. ries, such as stream dwelling, cave dwelling, and terres- 1988; O’Grady et al. 2004). Traits are also useful for trial burrowing (Pflieger 1996; Taylor & Schuster 2004). forecasting invasion success of non-native species (Ko- Consequently, our analysis explores a wide breadth of lar & Lodge 2002). Despite this, trait analyses to date crayfish taxa, traits, and a diversity of imperiled and inva- have been applied predominantly to well-studied taxa sive species in their native ranges. We examined species such as vertebrates (e.g., Purvis et al. 2000; Reynolds in seven genera in the family Cambaridae: Orconectes et al. 2005; Jeschke & Strayer 2008) and plants (e.g., Rej- (n = 41), Cambarus (n = 27), Procambarus (n = 4), manek & Richardson 1996; Reichard & Hamilton 1997), Cambarellus (n = 2), Barbicambarus (n = 1), Fallicam- with far fewer such studies conducted on aquatic in- barus (n = 1), and Faxonella (n = 1). This taxonomic dis- vertebrates, particularly crayfishes (but see Adamowicz tribution of species is consistent with global crayfish di- & Purvis 2005). This is unfortunate because trait-based versity because Orconectes and Cambarus are two of the

Conservation Biology Volume 24, No. 4, 2010 Larson & Olden 1101 three most diverse genera in the world (Crandall & Buhay Table 1. Crayfish trait values reported as mean (SD) for continuous 2008), although the most diverse genus—Procambarus- traits (maximum size, maximum fecundity, egg size) and percent assignment for categorical traits (range, habitat, substrate, annual —is underrepresented in our study region. ∗ reproductive events, chelae size). We confirmed 13 of the crayfishes native to Missouri and Kentucky as invasive, defined as a species occurring Trait Distribution of values outside its native range and encompassing both the intro- Maximum size (mm) 86.3 (25.3) duction and establishment stages of the invasion process. Maximum fecundity (eggs/female) 229 (174.7) These distinct invasion stages are explicitly discussed Egg size (mm) 2.2 (0.4) when species’ traits may relate to the stages separately. Range We did not address the impact stage of the invasion pro- narrowly endemic 26 cess, which has been studied for many but not all of regionally endemic 44 widespread 30 our focal species. We examined the scientific literature Habitat (i.e., Hobbs et al. 1989; Taylor et al. 2007) and used the lotic 58 U.S. Geological Survey’s Nonindigenous Aquatic Species lentic 5 Database (USGS 2009) to determine the invasive status terrestrial 10 of crayfishes. Invasive crayfishes were categorized as ei- cave 8 = generalist 18 ther extraregional (n 8), if they have invaded another Substrate continent or crossed major drainage boundaries within coarse 56 North America (i.e., Procambarus clarkii; Hobbs et al. fine 23 1989), or extralimital (n = 5),iftheyhaveinvadeda generalist 19 drainage or state adjacent to their native range within the Annual reproductive events one 69 southeastern United States (i.e., Orconectes hylas; Riggert multiple 6 et al. 1999). This distinction was made because the eco- Chelae size logical consequences of extraregional versus extralimital small 9 invasions may differ and because scale influences traits average 74 associated with invasion success (Spear & Chown 2008). large 16 Consequently, we anticipated that traits promoting inva- ∗Values are percentage unless otherwise indicated. Due to missing sion success in an adjacent drainage may differ from traits values, not all categorical trait distributions sum to 100%. required to invade more distant locales. To evaluate crayfish imperilment status we used Na- We report maximum size as the total body length (tip tureServe global status rankings (NatureServe 2008) and of the rostrum to end of the telson) and used this mea- a recent assessment of North American crayfishes by Tay- sure rather than mean or median size because maximum lor et al. (2007). We categorized a crayfish species as im- values were most often reported in Pflieger (1996) and periled (n = 21) if its NatureServe status was G1 (highly Taylor and Schuster (2004). Similarly, we used maximum imperiled), G2 (imperiled), or G3 (vulnerable) or if Tay- fecundity (eggs per female) because reliable values for lor et al. (2007) considered it endangered or threatened. mean or median fecundity were lacking. Nevertheless, We categorized a species as secure (n = 43) if its Nature- regression analysis revealed a close linear relationship Serve status was G4 (apparently secure) or G5 (secure) between maximum and mean fecundity when both data or if Taylor et al. (2007) considered it currently stable. types were available (R2 = 0.94, p < 0.001, n = 16). Egg A single crayfish (Orconectes neglectus chaenodactylus) size is a measure of investment in progeny and was most was coded as extralimital, although it is also imperiled often available as a single value in Pflieger (1996) and according to our criteria. Taylor and Schuster (2004). We categorized crayfishes as having either one or multiple annual reproductive events. We categorized chelae size as either large, average, or Crayfish Trait Database small on the basis of qualitative descriptors in Pflieger Through a literature review, we identified key biological (1996) and Taylor and Schuster (2004), and quantitative traits (maximum body size, maximum fecundity, egg size, measurements we made with vernier calipers on plates number of annual reproductive events, and chelae [pin- printed in both sources. The mean ratio of chelae width to cer] size); key ecological traits (habitat preference and carapace length, a predictor of pinching strength in cray- substrate preference); and range size (Table 1) as poten- fishes (Claussen et al. 2008), was significantly different tially relevant to the invasion and imperilment processes (analysis of variance [ANOVA] F2,73 = 100.6, p < 0.001) for crayfishes. We used the guides on crayfish for each among large (mean = 0.44 [SD 0.02]), average (0.30 [SD state by Pflieger (1996) and Taylor and Schuster (2004) as 0.05]), and small (0.14 [SD 0.03]) categories. Qualitative our primary source of trait data and supplemented these categorical descriptors were used rather than continu- texts with papers published in the scientific literature ous measurements because of concerns over reliability (i.e., Riggert et al. 1999; Larson & Magoulick 2008). and consistency of measurements made from only a few

Conservation Biology Volume 24, No. 4, 2010 1102 Crayfish Extinction and Invasion Risk plates. We included chelae size because this trait has sig- nally, we tested the null hypothesis of no trait differences nificance in crayfish dominance interactions and compet- among species status (i.e., secure, imperiled, extralimital, itive ability (Rutherford et al. 1995), which may relate to extraregional) with a permutational multivariate analysis imperilment or invasion success (Garvey & Stein 1993). of variance (perMANOVA) (McArdle & Anderson 2001) Crayfishes were categorized by habitat preference as and used a test of multivariate homogeneity of group dis- either lotic (i.e., streams or rivers), lentic (i.e., swamps, persions to assess differences in within-group trait vari- marshes, ponds, or lakes), terrestrial, cave, or generalist ation (Anderson 2006). We used Gower’s resemblance (reported from two or more habitats within the species’ coefficient in both analyses. These statistical methods native range). For substrate preference, we categorized allowed us to test directly the “two sides of the same crayfishes as preferring coarse (i.e., gravel, pebble, cob- coin” hypothesis. Statistical analyses were performed in ble, or boulder) or fine substrates (i.e., clay, silt, or fine R (R Development Core Team 2008), DISTLM (Anderson detritus) or as generalists (i.e., all substrate types). We cat- 2004a), and PERMDISP (Anderson 2004b). egorized crayfishes by range size with the distributional We used classification trees to identify and isolate trait maps in Pflieger (1996) and Taylor and Schuster (2004) as correlates of crayfish species status. Classification trees narrowly endemic (occurring within a single drainage or are a machine-learning technique that can operate on small area in a single state), regionally endemic (occurring mixed variable types, are invariant to monotonic transfor- within a single state or adjacent states), or widespread mations of data, and are well suited for modeling the non- (occurring throughout the region or continent). linear and nonadditive relationships typical of species’ Initial trait values were reviewed by two experts famil- trait data (De’ath & Fabricius 2000; Olden et al. 2008). iar with the crayfishes of the study region. On the basis Classification trees use a recursive partitioning algorithm of comments from both reviewers we adjusted trait val- to split data into a nested series of mutually exclusive ues for a small number of crayfishes. Despite our use of groups with the goal of maximizing homogeneity of the Pflieger (1996), Taylor and Schuster (2004), the scientific response variable. In our case, we constructed a classifi- literature, and two expert reviews, cases still existed in cation tree in which species status was modeled as a func- which values were not available for some traits for some tion of the eight traits and two indices of phylogenetic crayfishes. Thus, we used statistical analyses that could relatedness. The relatedness indices were derived from account for or were robust to missing trait data. Trait a PCoA on the phylogenetic distance matrix (described values for all crayfish species we evaluated are available above) that projected species in reduced multivariate (see Supporting Information). space according to their phylogeny. The two principal coordinates included in the classification tree accounted for 69.2% of original variation (47.1 and 22.1%, respec- Data Analyses tively) and were significant on the basis of the broken- Selection of statistical analyses was guided by a desire to stick model (Peres-Neto et al. 2003). We also constructed account for the innate complexities of species’ trait data, a classification tree that excluded range size as a predictor including the use of continuous and categorical variables to account for the fact that this is the most common jus- and the presence of some missing values. Consequently, tification for recognizing crayfishes as imperiled (Taylor we broadly characterized relationships among traits and et al. 2007). Optimal variable splits in classification trees crayfish imperilment and invasion status by applying a were determined with the Gini impurity criterion (mini- multivariate ordination technique, principal coordinate mum parent node size: n = 5; minimal terminal node size: analysis (PCoA), using Gower’s resemblance coefficient n = 2); final tree size was based on 10-fold cross valida- (Pavoine et al. 2009). We then performed a phylogenetic tion and selected on the basis of minimum cost tree; and correction to account for trait similarities among species Cohen’s κ coefficient of agreement was used to assess on the basis of shared ancestry (Fisher & Owens 2004) by classification tree performance. We report top compet- applying the Diniz-Filho et al. (1998) eigenvector method ing surrogate splits for each tree and variable importance to partition variance in our trait matrix into phylogenetic calculated from performance of variables across all sur- and specific components (e.g., Olden et al. 2006). First, rogate splits relative to the best performing variable. All we used the most current crayfish phylogeny (Tree of classification trees were constructed with CART 6.0 (Sal- Life 2008) to assemble a phylogenetic distance matrix ford Systems, San Diego, California). on the basis of the total number of nodes separating all species in our data set (following Webb et al. 2002). Sec- ond, we performed a Mantel test, in which we regressed the phylogenetic distance matrix against the trait distance Results matrix to produce a residual matrix representing trait sim- Multivariate Patterns of Trait Similarities ilarities among species after controlling for phylogeny. The residual matrix was subjected to a PCoA to ordinate The PCoA performed on the species trait matrix (after ac- dominant patterns of trait variation among crayfishes. Fi- counting for shared phylogeny) explained 20.3% of total

Conservation Biology Volume 24, No. 4, 2010 Larson & Olden 1103

Table 2. The F statistics for permutational multivariate analysis of variancea and test of homogeneity of multivariate dispersionb between crayfish status groups.c

Secure Imperiled Extralimital Extraregional

Secure 10.100∗∗∗ 4.129∗ 8.289∗∗∗ Imperiled 0.801 4.664∗ 15.734∗∗∗ Extralimital 0.105 0.299 14.892∗∗ Extraregional 1.825 2.300∗ 1.033 aUpper diagonal. bLower diagonal. cSignificance: ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

sive crayfishes occupied different positions in multivari- ate trait space (Figs. 1a & c). Trait differences between se- cure, imperiled, extralimital, and extraregional crayfishes were statistically significant (perMANOVA F3,73 = 9.100, p < 0.001). Trait compositions were most distinct be- tween extraregional and imperiled species, whereas ex- tralimital and secure species exhibited the smallest trait differences (Table 2). Within-group trait dispersions were similar between most crayfish status groups, but signif- icantly different between extraregional and imperiled crayfishes (Table 2). The first PCoA axis identified a trait gradient that parsed widespread lentic, terrestrial, and habitat generalist cray- fishes characterized with relatively high fecundity and maximum size (negative scores on PC1) from narrowly and regionally endemic lotic crayfishes with lower fecun- dity and maximum size (positive scores on PC1) (Fig. 1b). Egg size correlated negatively with the second axis and separated narrowly endemic and cave-dwelling crayfishes (positive scores on PC2) from all other species (nega- tive scores on PC2). Imperiled crayfishes were gener- ally associated with habitat specialization (lotic and cave- dwelling), small body sizes, low fecundity, and small range sizes (Figs. 1b & c). Extraregional crayfishes oc- cupied the lower-left quadrant in ordination space, with no overlap with imperiled crayfishes, and were instead characterized as habitat generalists, lentic, or terrestrial with large body sizes, high fecundity, large range sizes, Figure 1. Ordination plot from the principal and multiple annual reproductive events. Interestingly, coordinate analysis (PCoA) of 77 crayfish species on extralimital crayfishes were distinct from extraregional the basis of eight biological and ecological traits: (a) crayfishes because they were characterized by lotic habi- species, (b) trait loadings, and (c) species status. tat preferences and endemic ranges (Figs. 1b & c). Crayfish species abbreviations are provided in Table 3. Some of the data points were jittered to Trait Correlates of Crayfish Imperilment and Invasion improve clarity. The classification tree that included range size correctly classified species status for 56 of 77 crayfishes (73%), with a Cohen’s κ of 0.565 (p < 0.001) indicating a classifica- trait variation in the first two axes; both principal co- tion performance that exceeded random expectations. ordinate axes were statistically significant based on the The correct classification rate was 70% for imperiled broken-stick model. Subsequent axes did not affect result species, 72% for secure species, 75% for extrare- interpretation, although several explained significant (al- gional species, and 80% for extralimital species. Extrare- beit small) amounts of variation. Results of this PCoA con- gional crayfishes were generally characterized as having firmed our initial hypothesis because imperiled and inva- widespread native ranges and relatively higher fecundity

Conservation Biology Volume 24, No. 4, 2010 1104 Crayfish Extinction and Invasion Risk

Figure 2. Classification tree (a) with traits, phylogenetic relatedness, and range size to discriminate among secure, imperiled, extralimital, and extraregional crayfishes. For all bar graphs in (a), the y-axis represents the number of species, which is reported over each individual bar. Terminal node assignments are based on a comparison of the percentage of species in each status category relative to their prevalence in the entire data set (root node). (b) Relative importance of predictor traits across all surrogate splits. Top competing surrogate splits (a) include (1) habitat (

Conservation Biology Volume 24, No. 4, 2010 Larson & Olden 1105

Figure 3. Classification tree (a), excluding range size, with traits and phylogenetic relatedness to discriminate among secure, imperiled, extralimital, and extraregional crayfishes. Configuration of bar plots and assignment of terminal nodes are as described in Fig. 2. (b) Relative importance of predictor traits across all surrogate splits. Top competing surrogate splits in (a) include (1) habitat (< Generalist), (2) egg size (< 1.85 mm), and (3) body size (< 102.3 mm).

Discussion Consistent with studies of other taxonomic groups, we found that small range sizes were associated with crayfish By considering the biological and ecological traits of a imperilment (Purvis et al. 2000; Reynolds et al. 2005) and crayfish-rich region of the world, our results provide range size was conversely correlated between invasive new insight into why certain crayfish species appear fa- and imperiled species (Bradshaw et al. 2008; Jeschke & vored in a human-modified environment, whereas others Strayer 2008). Nevertheless, because small range sizes are rapidly decline or disappear. We found that invasion and a common justification for recognizing crayfish species imperilment in crayfishes were highly predictive by a as imperiled (Taylor et al. 2007), we performed analyses small number of traits and that particular species have a with and without range size to avoid or identify circu- latent risk for invasion or extinction in the future. This lar correlations among predictor and response variables. finding demonstrates that management of crayfishes can Withholding range size from our classification tree anal- be more proactive through application of a trait-based ap- ysis showed that fecundity and egg size were the most proach that relates patterns of imperilment and invasion important predictors of imperiled status. Although im- to underlying biological and ecological mechanisms. De- periled crayfishes had low fecundities relative to invasive veloping, testing, and applying such approaches will be crayfishes, low fecundity was not the sole discriminat- critical for prioritizing conservation needs of the world’s ing factor. Instead, low reproductive investment (small many diverse, understudied taxa (Strayer 2006). egg size) was associated with the majority of imperiled

Conservation Biology Volume 24, No. 4, 2010 1106 Crayfish Extinction and Invasion Risk

Table 3. Crayfish species evaluated in the study, current status, and predicted status from classification trees.∗

Predicted status Species Abbreviation Status w/range w/out range

Cambarellus puer puer secure secure imperiled Cambarellus shufeldtii shuf secure secure imperiled Barbicambarus cornutus corn secure secure secure Cambarus bartonii cavatus bart secure secure secure Cambarus batchi batc imperiled imperiled imperiled Cambarus bouchardi bouc imperiled imperiled imperiled Cambarus buntingi bunt secure secure imperiled Cambarus cumberlandensis cumb extralimital extralimital extralimital Cambarus deweesae dewe secure secure imperiled Cambarus diogenes diog secure secure secure Cambarus distans dist secure secure secure Cambarus dubius dubi secure secure secure Cambarus friaufi fria secure secure imperiled Cambarus graysoni gray secure secure secure Cambarus hubbsi hubb secure secure secure Cambarus hubrichti hubr secure secure imperiled Cambarus ludovicianus ludo secure secure imperiled Cambarus maculatus macu secure imperiled secure Cambarus ortmanni ortm secure secure extraregional Cambarus parvoculus parv secure secure imperiled Cambarus robustus robu extraregional secure secure Cambarus rusticiformis Crus extralimital extralimital extralimital Cambarus sciotensis scio secure extralimital extralimital Cambarus setosus seto imperiled secure secure Cambarus sphenoides sphe secure secure extraregional Cambarus striatus stria secure secure extraregional Cambarus tenebrosus tene secure secure secure Cambarus thomai thom secure extraregional extraregional Cambarus veteranus vete imperiled secure imperiled Cambarus polychromatus poly secure secure secure Fallicambarus fodiens fodi secure secure imperiled Faxonella clypeata clyp secure secure imperiled Orconectes australis packardi aust imperiled imperiled imperiled Orconectes barrensis barr secure extralimital extralimital Orconectes bisectus bise imperiled imperiled imperiled Orconectes burri burr imperiled imperiled imperiled Orconectes compressus comp secure extralimital extralimital Orconectes cristavarius cris secure secure imperiled Orconectes durelli dure secure secure secure Orconectes eupunctus eupu imperiled imperiled imperiled Orconectes harrisoni harr imperiled imperiled imperiled Orconectes hylas hyla extralimital extralimital extralimital Orconectes immunis immu extraregional extraregional extraregional Orconectes inermis inermis iner secure secure secure Orconectes jeffersoni jeff imperiled imperiled imperiled Orconectes juvenilis juve extraregional secure imperiled Orconectes kentuckiensis kent imperiled secure imperiled Orconectes lancifer lanc secure extraregional extraregional Orconectes longidigitus long secure secure secure Orconectes luteus lute secure secure imperiled Orconectes macrus macr secure extralimital extralimital Orconectes marchandi marc imperiled extralimital extralimital Orconectes margorectus marg imperiled imperiled imperiled Orconectes medius medi secure extralimital extralimital Orconectes meeki meeki meek secure secure imperiled Orconectes neglectus neglectus negn extraregional extraregional extraregional Orconectes neglectus chaenodactylus negc extralimital extralimital extralimital Orconectes ozarkae ozar secure extralimital extralimital Orconectes palmeri palmeri palm secure secure imperiled continued

Conservation Biology Volume 24, No. 4, 2010 Larson & Olden 1107

Table 3. (continued).

Predicted status Species Abbreviation Status w/range w/out range

Orconectes pellucidus pell imperiled secure imperiled Orconectes peruncus peru imperiled imperiled imperiled Orconectes placidus plac extralimital secure imperiled Orconectes punctimanus punct secure secure secure Orconectes putnami putn secure secure imperiled Orconectes quadruncus quad imperiled imperiled imperiled Orconectes rafinesquei rafi imperiled imperiled imperiled Orconectes ronaldi rona imperiled imperiled imperiled Orconectes rusticus Orus extraregional extraregional extraregional Orconectes sanbornii sanb secure secure secure Orconectes stygocaneyi styg imperiled imperiled imperiled Orconectes tricuspis tric secure imperiled imperiled Orconectes virilis viri extraregional extraregional extraregional Orconectes williamsi will imperiled secure imperiled Procambarus acutus acut extraregional extraregional extraregional Procambarus clarkii clar extraregional extraregional extraregional Procambarus gracilis grac secure extraregional imperiled Procambarus viaeveridus Viae secure extraregional imperiled ∗Predictions are from classification trees including (w/range) and excluding (w/out range) range size as a variable. crayfishes, and correct classification rates for imperiled These species included many of the world’s most harm- crayfishes actually improved when range size was not ful invasive crayfishes, such as the red swamp crayfish included in the model. In contrast, correct classification (Procambarus clarkii) and rusty crayfish (Orconectes rates for secure crayfishes declined in this analysis be- rusticus) (Hobbs et al. 1989; Lodge et al. 2000). Large cause many of these species were misclassified as imper- range sizes and high fecundities were particularly impor- iled on the basis of their suite of traits. tant in classifying extraregional crayfishes. This is con- Our findings conflict with the traditionally perceived sistent with studies that show range size is positively dichotomy between r life-history strategies characteriz- associated with invasion (Bradshaw et al. 2008; Jeschke ing invasive species and K life-history strategies com- & Strayer 2008), although this pattern may be caused by mon among imperiled species that underlies the “two relationships between range size and either likelihood sides of the same coin hypothesis” (Rejmanek & Richard- of introduction or successful establishment because of son 1996; Jeschke & Strayer 2008), as high investment broader environmental tolerances associated with larger in progeny is a K attribute that appears to be absent ranges (Blackburn & Jeschke 2009). Extraregional cray- in imperiled crayfishes. This indicates potential for a fishes were also characterized by high fecundities, a life- more complicated life-history classification system for history trait that may be important in establishment be- crayfishes, similar to Winemiller and Rose’s (1992) life- cause it allows small founder populations to overcome history pyramid that has been useful in trait analysis for Allee effects and environmental stochasticity (Sakai et al. fishes (Olden et al. 2006). Trade-offs among crayfish size, 2001). Classification trees can be useful in partitioning fecundity, and investment in progeny should be investi- the contribution of life-history and ecological traits to gated in field studies and literature syntheses, with results distinct stages of the invasion process (Kolar & Lodge interpreted in relation to crayfish responses to natural and 2002), but accomplishing this for crayfishes will require anthropogenic factors. data on known introductions where populations failed to Our findings also elaborate on those of Adamowicz establish. and Purvis (2005), who report that extinction risk in Accounting for scale of invasion was important in our crayfishes is associated with small body size and habi- analysis because traits of extraregional crayfishes were tat specialization. Although the same pattern is evident not consistent with those of extralimital crayfishes. Ex- in our PCoA ordination, we found that crayfish imperil- tralimital crayfishes had smaller range sizes, lower fecun- ment was related to the synergistic combination of low dity, and more-specific habitat requirements (lotic) than fecundity and small egg size. This result provides better extraregional crayfishes. This finding confirms our pre- insight into crayfish vulnerability because species with diction that traits related to invasion in adjacent drainages both these traits may exhibit lower resistance to and re- differ from traits required for invasion of distant regions covery from environmental and biological change. and continents. Regional processes related to human- Extraregional crayfishes were distinctly different in assisted transport (introduction) and environmental com- their trait composition from those that are imperiled. patibility (establishment) may be most important in

Conservation Biology Volume 24, No. 4, 2010 1108 Crayfish Extinction and Invasion Risk extraregional invasions, whereas local processes related some relatively common or widespread species, such as to competition or other biotic interactions may dictate Fallicambarus fodiens and Orconectes luteus, were clas- success of extralimital invasions (Ricklefs 1987; Spear & sified as imperiled. The biological and ecological sim- Chown 2008). Consistent with this expectation, large ilarity between many secure and imperiled crayfishes chelae size was associated with extralimital crayfishes. demonstrates the potential for common and widespread Chelae size in crayfishes is related to behavioral dom- crayfishes to possess traits that may predispose them to inance and competitive ability (Garvey & Stein 1993; population declines, extirpation, or even eventual ex- Rutherford et al. 1995), and thus this morphological trait tinction (e.g., Guiasu 2007). Although narrowly endemic may be important in determining which crayfishes intro- crayfishes deserve conservation attention because envi- duced at the extralimital scale can overcome local biotic ronmental perturbations can rapidly result in extinctions, resistance from native crayfish communities to establish population declines of more widespread crayfishes may populations. Extralimital invasions have a greater homog- similarly involve loss of unique phylogenetic diversity enizing effect on communities than extraregional inva- (Crandall 1998) or affect members of the aquatic com- sions (Spear & Chown 2008) and should be of particular munity dependent on crayfishes (Pintor & Soluk 2006). conservation concern in species-rich regions where ex- We advocate that trait-based susceptibility to anthro- tralimital crayfishes may threaten imperiled species with pogenic impacts, ecological role, and unique taxonomic displacement (Riggert et al. 1999; Larson & Magoulick or phylogenetic diversity be considered in combina- 2008). tion with range size in prioritizing crayfish conservation Our results support the validity of the “two sides of needs. the same coin” hypothesis for crayfishes because inva- Trait-based predictions of crayfish as extraregional or sive and imperiled species occupied statistically distinct extralimital represent an opportunity to quantify latent multivariate trait space. This finding is consistent with invasion risk and proactively allocate management ef- Bradshaw et al. (2008) but contrasts with Jeschke and forts to prevent future invasions. Cambarus thomai, Or- Strayer (2008) and Blackburn and Jeschke (2009), who conectes lancifer, and several others were misclassified found that few individual traits are conversely related to as extraregional because of their large range sizes and invasive and imperiled species of birds and fishes. Our high fecundities. This finding emphasizes that these cray- findings may differ from these studies because we only fishes and others like them are potential candidates to be- evaluated a single taxa, accounted for scale of invasion, come widely invasive in the future. Similarly, Cambarus and simultaneously considered multiple traits rather than sciotensis, Orconectes ozarkae, and others were misclas- individually. Failure to account for trait interactions or sified as extralimital. Although the smaller range sizes and synergisms may provide only a limited perspective on lower fecundities of these crayfishes may make them less the extinction and invasion processes (Kolar & Lodge likely to invade habitats beyond the southeastern United 2002; Davies et al. 2004; Olden et al. 2008). Nevertheless, States, large chelae size identifies them as prospective our findings did agree with those of Jeschke and Strayer candidates for extralimital invasions within this region. (2008) and Blackburn and Jeschke (2009) on the role We recommend monitoring areas where species exhibit- of range size in differentiating invasive from imperiled ing high latent invasion risk are in close proximity to species, and we also found evidence that the contribution endemic communities or species whose traits may pre- of life history to the invasion and imperilment processes dispose them to become imperiled. One limitation of our is more complicated than the classic r to K dichotomy. approach is that we lacked the information necessary to Our results support the intuitive hypothesis that species relate crayfish traits to the distinct introduction, establish- that respond to anthropogenic change with range expan- ment, and impact stages of the invasion process. Models sions differ from those that respond to anthropogenic derived from documented instances of crayfish introduc- change with range contractions or extinction. tions that failed to establish, or instances of established Examining model misclassifications from our analysis populations that failed to produce impacts, would pro- allowed us to identify latent extinction risk for “secure” vide a more mechanistic modeling framework and permit crayfishes that possessed traits characteristic of imper- more accurate predictions of latent invasion risk (Kolar iled species. When we included range size as a pre- & Lodge 2002). dictor, Cambarus maculatus and Orconectes tricuspis Our study should serve as a jumping-off point for fur- were both misclassified as imperiled species, which sug- ther investigations of trait relationships with imperilment gests these species may represent future conservation and invasion for understudied invertebrate taxa. We be- challenges. Crandall (1998) also recommends upgrading lieve our models may be transferable or testable in other the conservation status for C. maculatus from secure to geographic regions because of the relative simplicity of imperiled because of this species’ small range size and the trait data we considered, although our analysis also its unique phylogenetic diversity. When we excluded highlights that much is still unknown about the basic range size from our analysis, the number of crayfishes ecology of many crayfish species. Although we encour- misclassified as imperiled increased dramatically, and age researchers to continue pursuing basic ecological

Conservation Biology Volume 24, No. 4, 2010 Larson & Olden 1109 studies, particularly for understudied taxa and regions of Claussen, D. L., G. W. Gerald, J. E. Kotcher, and C. A. Miskell. 2008. the world, an advantage of a broad trait-based approach Pinching forces in crayfish and fiddler crabs, and comparisons is that it allows conservation biologists to generalize pat- with the closing forces of other animals. Journal of Comparative Physiology B: Biochemical, Systemic, and Environmental Physiol- terns when species-specific data are incomplete. Success- ogy 178:333–342. ful conservation of diverse, understudied taxa will require Crandall, K. A. 1998. Conservation phylogenetics of Ozark crayfishes: acknowledging the limitations of single-species manage- assigning priorities for aquatic habitat protection. Biological Con- ment (Strayer 2006) and compensating for resource lim- servation 84:107–117. itations that prevent intensive research on each of the Crandall, K. A., and J. E. Buhay. 2008. Global diversity of crayfish (Astaci- dae, Cambaridae, and Parastacidae-Decapoda) in freshwater. Hydro- world’s more than 640 crayfish species (Crandall & Buhay biologia 595:295–301. 2008). Research tools need to be developed to guide and Davies, J. F., C. R. Margules, and J. F. Lawrence. 2004. A synergistic prioritize conservation decisions for these species, and effect puts rare, specialized species at greater risk of extinction. we advocate trait analysis as a prospective means to this Ecology 85:265–271. end. De’ath, G., and K. E. Fabricius. 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology 81:3178–3192. Diniz-Filho, J. A. F., C. E. R. De Sant’ana, and K. M. Bini. 1998. An Acknowledgments eigenvector method for estimating phylogenetic inertia. Evolution 52:1247–1262. We are grateful to R.J. DiStefano and C.A. Taylor for Fisher, D. O., and I. P. F. Owens. 2004. The comparative method in reviewing our crayfish trait data, to K.A. Crandall for conservation biology. Trends in Ecology & Evolution 19:391–398. discussions on crayfish phylogeny, and to W.L. Pflieger, Garvey, J. E., and R. A. Stein. 1993. Evaluating how chelae size influ- ences the invasion potential of an introduced crayfish (Orconectes G.A. Schuster, and all other researchers who have con- rusticus). American Midland Naturalist 129:172–181. tributed work on crayfish ecology. Two anonymous re- Guiasu, R. C. 2007. Conservation and diversity of the crayfishes of the viewers contributed comments that improved the draft genus Fallicambarus Hobbs, 1969 (Decapoda, Cambaridae), with an manuscript. E.R.L. was supported during analysis of the emphasis on the status of Fallicambarus fodiens (Cottle, 1863) in data and writing of the manuscript at the University of Canada. Crustaceana 80:207–223. Hobbs, H. H., J. P. Jass, and J. V. Huner. 1989. A review of global cray- Washington by a Victor and Tamara Loosanoof Fellow- fish introductions with particular emphasis on two North American ship and Achievement Rewards for College Scientists. species (Decapoda, Cambaridae). Crustaceana 56:299–316. Jeschke, J. M., and D. L. Strayer. 2008. Are threat status and invasion success two sides of the same coin? Ecography 31:124–130. Jones, J. G., F. B. Andriahajaina, N. J. Hockley, K. A. Crandall, and O. Supporting Information R. Ravoahangimalala. 2007. The ecology and conservation status of Madagascar’s endemic freshwater crayfish (Parastacidae; Asta- The crayfish trait data are available as part of the online coides). Freshwater Biology 52:1820–1833. Kolar, C. S., and D. M. Lodge. 2002. Ecological predictions and risk article (Appendix S1). The authors are responsible for assessment for alien fishes in North America. Science 8:1233–1236. the content and functionality of these materials. Queries Larson, E. R., and D. D. Magoulick. 2008. Comparative life history of (other than absence of the material) should be directed native (Orconectes eupunctus) and introduced (Orconectes neglec- to the corresponding author. tus) crayfishes in the Spring River drainage of Arkansas and Missouri. American Midland Naturalist 160:323–341. Lodge, D. M., C. A. Taylor, D. M. Holdich, and J. Skurdal. 2000. Non- Literature Cited indigenous crayfishes threaten North American freshwater biodiver- sity: lessons from Europe. Fisheries 25:7–20. Adamowicz, S. J., and A. Purvis. 2005. Macroevolution and extinction Matsuzaki, S. S., N. Usio, N. Takamura, and I. Washitani. 2009. Con- risk patterns in freshwater crayfish. Freshwater Crayfish 15:1–23. trasting impacts of invasive engineers on freshwater ecosystems: an Anderson, M. J. 2004a. DISTLM v.5: a FORTRAN computer program to experiment and meta-analysis. Oecologia 158:673–685. calculate a distance-based multivariate analysis for a linear model. McArdle, B. H., and M. J. Anderson. 2001. Fitting multivariate models to Department of Statistics, University of Auckland, New Zealand. community data: a comment on distance-based redundancy analysis. Anderson, M. J. 2004b. PERMDISP: a FORTRAN computer program for Ecology 82:290–297. permutational analysis of multivariate dispersions using permutation Momot, W. T. 1995. Redefining the role of crayfish in aquatic ecosys- tests. Department of Statistics, University of Auckland, New Zealand. tems. Reviews in Fisheries Science 3:33–64. Anderson, M. J. 2006. Distance-based tests for homogeneity of multi- NatureServe. 2008. NatureServe explorer. NatureServe, Arlington, Vir- variate dispersions. Biometrics 62:245–253. ginia. Available from http://www.natureserve.org/explorer/ (ac- Blackburn, T. M., and J. M. Jeschke. 2009. Invasion success and threat cessed September 2008). status: two sides of a different coin? Ecography 32:83–88. Nystrom,¨ P., C. Bronmark,¨ and W. Graneli.´ 1996. Patterns in benthic Bradshaw, J. A., X. Giam, H. T. W. Tan, B. W. Brook, and N. S. Sodhi. food webs: a role for omnivorous crayfish? Freshwater Biology 2008. Threat or invasive status in legumes is related to opposite 36:631–646. extremes of the same ecological and life-history attributes. Journal O’Grady, J. J., D. H. Reed, B. W. Brook, and R. Frankham. 2004. What of Ecology 96:869–883. are the best correlates of predicted extinction risk. Biological Con- Cardillo, M., G. M. Mace, J. L. Gittleman, and A. Purvis. 2006. Latent servation 118:513–520. extinction risk and the future battlegrounds of mammal conserva- Olden, J. D., N. L. Poff, and K. R. Bestgen. 2006. Life-history strategies tion. Proceedings of the National Academy of Sciences of the United predict fish invasions and extirpations in the Colorado River basin. States of America 103:4157–4161. Ecological Monographs 76:25–40.

Conservation Biology Volume 24, No. 4, 2010 1110 Crayfish Extinction and Invasion Risk

Olden, J. D., N. L. Poff, and K. R. Bestgen. 2008. Trait synergisms and cus (Creaser) and Orconectes quadruncus (Creaser) in Missouri. the rarity, extirpation, and extinction risk of desert fishes. Ecology American Midland Naturalist 142:348–362. 89:847–856. Rodriguez, C. F., E. Becares, M. Fernandez-Alaez, and C. Fernandez- Pavoine, S., J. Vallet, A. Dufour, S. Gachet, and H. Daniel. 2009. Alaez. 2005. Loss of diversity and degradation of wetlands as a result On the challenge of treating various types of variables: applica- of introducing exotic crayfish. Biological Invasions 7:75–85. tion for improving the measurement of functional diversity. Oikos Rutherford, P. L., D. W. Dunham, and V. Allison. 1995. Winning ag- 118:391–402. onistic encounters by male crayfish Orconectes rusticus (Girard) Peres-Neto, P. R., D. A. Jackson, and K. M. Somers. 2003. Giving mean- (Decapoda, Cambaridae): chelae size matters but chelae symmetry ingful interpretation to ordination axes: assessing loading signifi- does not. Crustaceana 68:526–529. cance in principal component analysis. Ecology 84:2347–2363. Sakai, A. K., et al. 2001. The population biology of invasive species. Pflieger, W. L. 1996. The crayfishes of Missouri. Missouri Department Annual Review of Ecology and Systematics 32:305–332. of Conservation, Jefferson City. Spear, D., and S. L. Chown. 2008. Taxonomic homogenization in ungu- Pimm, S. L., H. L. Jones, and J. Diamond. 1988. On the risk of extinction. lates: patterns and mechanisms at local and global scales. Journal of The American Naturalist 132:757–785. Biogeography 35:1962–1975. Pintor, L. M., and D. A. Soluk. 2006. Evaluating the non-consumptive, Strayer, D. L. 2006. Challenges for freshwater invertebrate conserva- positive effects of a predator in the persistence of an endangered tion. Journal of the North American Benthological Society 25:271– species. Biological Conservation 130:584–591. 287. Purvis, A., J. L. Gittleman, G. Cowlishaw, and G. M. Mace. 2000. Pre- Taylor, C. A., and G. A. Schuster. 2004. The crayfishes of Kentucky. dicting extinction risk in declining species. Proceedings of the Royal Illinois Natural History Survey, Champaign. Society of London B 267:1947–1952. Taylor, C. A., G. A. Schuster, J. E. Cooper, R. J. DiStefano, A. G. Ever- R Development Core Team. 2008. R: a language and environment sole, P. Hamr, H. H. Hobbs III, H. W. Robison, C. E. Skelton, and R. for statistical computing. R Foundation for Statistical Computing, F. Thoma. 2007. A reassessment of the conservation status of cray- Vienna, Austria. fishes of the United States and Canada after 10+ years of increased Reichard, S. H., and C. W. Hamilton. 1997. Predicting invasions of awareness. Fisheries 32:372–389. woody plants introduced into North America. Conservation Biology Tree of Life. 2008. Tree of Life web project. Available from http:// 11:193–203. www.tolweb.org/Astacidea (accessed September 2008). Rejmanek, M., and D. M. Richardson. 1996. What attributes make some U.S. Geological Survey. 2009. Nonindigenous aquatic species database. plant species more invasive? Ecology 77:1655–1661. http://nas.er.usgs.gov/ (accessed September 2008). Reynolds, J. D., N. K. Dulvy, N. B Goodwin, and J. A. Hutchings. 2005. Webb, C. O., D. D. Ackerly, M. A. McPeek, and M. J. Donoghue. 2002. Biology of extinction risk in marine fishes. Proceedings of the Royal Phylogenies and community ecology. Annual Review in Ecology and Society of London B 272:2337–2344. Systematics 33:475–505. Ricciardi, A., and J. B. Rasmussen. 1999. Extinction rates of Wilcove, D. S., and L. L. Master. 2005. How many endangered species North American freshwater fauna. Conservation Biology 13:1220– are there in the United States? Frontiers in Ecology and the Environ- 1222. ment 8:414–420. Ricklefs, R. E. 1987. Community diversity: relative roles of local and Winemiller, K. O., and K. A. Rose. 1992. Patterns of life-history di- regional processes. Science 235:167–171. versification in North American fishes: implications for popula- Riggert, C. M., R. J. DiStefano, and D. B. Noltie. 1999. Distributions tion regulation. Canadian Journal of Fisheries and Aquatic Sciences and selected ecological aspects of the crayfishes Orconectes perun- 49:2196–2218.

Conservation Biology Volume 24, No. 4, 2010