Global representation of threatened amphibians ex situ is bolstered by non-traditional institutions, but gaps remain

Introduction

Amphibians are the most imperilled class of vertebrates, with at least one third of extant species classified as threatened with extinction (Hoffman et al., 2010) and 42% of species having experienced recent population declines (Stuart et al., 2004; Wake & Vredenburg, 2008; Whittaker et al., 2013). These contemporary extinction rates are four orders of magnitude higher than natural background rates for anurans (Alroy, 2015). It is unlikely that this situation will improve without immediate and effective conservation initiatives.

One such initiative was the 2007 Amphibian Conservation Action Plan (ACAP), a guide for implementing global amphibian conservation and research (Gascon et al., 2007). Because of the difficulty in rapidly mitigating particular extinction drivers, namely habitat loss and degradation (even within protected areas; Curran et al., 2004), accelerating effects of climate change (Foden et al., 2013) and the spread of emerging infectious diseases (Olson et al., 2013), 2 of the 11 chapters within the ACAP focus on the importance of ex situ conservation (i.e. captive breeding and reintroduction). The Amphibian Ark (AArk; Amphibian Ark, 2015a) was subsequently initiated to address the captive components of the ACAP and, in particular, to focus on species thought most difficult to safeguard in situ (Zippel et al., 2011). More specifically, AArk helps advise and coordinate regional and global amphibian ex situ efforts while facilitating the prioritization of amphibians through their Conservation Needs Assessments (Amphibian Ark, 2014). AArk maintains its own records of the institutions managing threatened species. These include smaller specialist institutions, often located within developing, high-biodiversity countries within the tropics.

Although exceptions exist (cf. Tapley et al., 2015), amphibians are generally highly suitable for ex situ conservation measures. They are small, relatively inexpensive to keep, often breed quickly and usually cope with captivity, both physiologically and behaviourally, better than do some other taxa (Bloxam & Tonge, 1995; Balmford, Mace & Leader-Williams, 1996; Conde et al., 2015). Ex situ amphibian programmes are also expanding: while Conde et al. (2011) estimated that only 4% of amphibian species were held in captivity worldwide at the turn of the decade (vs. 25% of bird species and 20% of mammal species), Harding, Griffith & Pavajeau (2016) reported a 57% increase in the number of amphibian species involved in captive breeding and reintroduction programmes since the launch of the ACAP in 2007, and Dawson et al. (2016) reported a near doubling of ex situ holdings of amphibians from 1994 to 2014 to a total of 10.9%; this latter figure is more than double the total number of species reported by Conde et al. (2011), just 4 years prior. It is clear that ex situ institutions are playing an increasingly important role in the global conservation strategy for amphibians.

But is the growing number of amphibian species held ex situ representing the species of greatest conservation priority? While raw counts and proportions of International Union for the Conservation of Nature (IUCN)-listed species held in zoos have been reported (see Conde et al., 2011; Dawson et al., 2016), no existing research examines representation with respect to ecological and biogeographical indicators of threat, nor whether the emerging role of non-traditional institutions in ex situ conservation has affected representation of threatened amphibians across the global ex situ network.

Using a phylogenetically controlled matched-pair design similar to a previous study of birds and mammals (Martin, Lurbiecki & Mooers, 2014a), we investigate how variables correlated with extinction risk are related to the likelihood of amphibian species being held in zoos. We contrast ‘in-zoo’ species identified using the Zoological Information Management System (ZIMS; International Species Information System, 2015) and Amphibian Ark (2015b) databases with ‘not-in-zoo’ close relatives across a set of candidate predictors analysed both individually and in multivariate logistic regressions. While the ZIMS dataset is the largest database regarding ex situ species holdings for regionally or nationally accredited zoos [including those accredited by the World Association of Zoos and Aquariums, the (American) Association of Zoos and Aquariums, and the European Association of Zoos and Aquaria], the AArk database includes species holdings from a number of institutions that are not part of a zoo association. In order to both compare the patterns in amphibians with those previously reported for birds and mammals, and to evaluate the specific effects of these non-traditional institutions identified by AArk, we compare two datasets: an ‘All Institutions’ dataset, comprised of species in either or both the ZIMS and AArk databases, and a ZIMS dataset. These comparisons allow us to evaluate the extent to which (1) current ex situ representation of amphibians aligns with species representing the most urgent global conservation priorities; (2) whether the efforts and coordination of AArk have influenced this representation.

Materials and methods

Species pair construction

Our basic statistical approach is outlined in Fig. 1: the method was first suggested by Felsenstein (1985) and is also the one used in an earlier paper that considered birds and mammals (Martin et al., 2014a; Martin, Lurbiecki & Mooers, 2014b). The goal is to identify, on a phylogenetic tree, independent pairs of species that differ in a character of interest (here, contrasting in zoo vs. not in zoo), and then to ask how members of each pair differ in other characters (e.g. IUCN status or range size). Because each pair (or ‘contrast’) is phylogenetically independent of others, we can perform tests of association (e.g. sign tests) on the full set of contrasts. We present the phylogenetic tree we used (adapted from Pyron & Wiens, 2011) and outline in detail how the final 219 contrasts were constructed in Supporting Information (Appendix S1).

Results

Z tests demonstrated that our sample of in-zoo species were representative of all species in their respective orders for all variables in all analyses with one exception – body size for Caudata in the All Institutions analysis and the ZIMS analysis (P < 0.05 for both datasets). This was due to the presence of the two in-zoo giant salamanders (genus Andrias), which were dropped from the main analyses because they could not be paired with not-in-zoo close relatives. Given only two atypical outliers, we included body size for the other Caudata in further analyses. We found that correlations between our predictor variables were weak or moderate (all r < 0.7), indicating that our interpretation of contrast results should be straightforward.

All our contrast results are presented in Table 1. When all institutions are considered, we found no significant differences in threat status, high-altitude endemism, island endemism or tropical endemism between species held in zoos and their close relatives not held in zoos (all P > 0.05), while stream obligates tended not to be found in zoos (P < 0.07). In contrast, when we considered the ZIMS subset, species held in ZIMS institutions are less likely to be considered threatened than their close relatives not held in zoos (P = 0.05). All other categorical variables showed no difference for species held in ZIMS-associated institutions.

For both the All Institutions and the ZIMS tests, in-zoo species were significantly larger (P < 0.001 for both datasets), had significantly larger geographic range sizes (P < 0.001 for both datasets) and broader habitat breadths (P < 0.001 for both) than their close not-in-zoo relatives. Considering all institutions, in-zoo species are on average 13.5% larger, occupy a geographic range three and a half times the size and occur in 27% more discrete habitats than their not not-in-zoo close relatives. For ZIMS species, the average differences were even greater: 13.9% larger body size, over four times larger geographic range size and 35% broader habitat breadths.

Multi-model inference across the All Institutions dataset and for species in the ZIMS database indicated similar sets of the most parsimonious models. For both, top models suggested that a larger habitat breadth (P = 0.062, All; P = 0.017, ZIMS), larger geographic range (P < 0.001 for both) and higher threat (P = 0.001; P = 0.025), all increased the probability of being held ex situ (Table S3.1a in Supporting Information Appendix S3). However, differences emerged in the ranked importance of variables (averaged across models) predicting the probability of being held ex situ across these two datasets. While geographic range size and IUCN threat status were the two most highly weighted predictors for the All Institutions dataset, habitat breadth was more important than threat status for the ZIMS dataset (Table 1).

Discussion

Consistent with patterns for birds and mammals (Martin et al., 2014a), amphibian species held in zoos are significantly larger bodied, possess larger geographic ranges and are more generalist in their habitats than their not-in zoo counterparts. Importantly, however, and in contrast to patterns for birds and mammals (Martin et al., 2014a), amphibians currently held in zoos are equally as threatened as their close out-of-zoo relatives. This result is driven by the relatively small number of amphibian captive breeding programmes in ‘non-traditional’ zoos, which are not recorded in the ZIMS database; when species found only in these institutions are removed, amphibians in zoos are less threatened than their out-of-zoo close relatives (Table 1).

This contrast has two main implications. First, as with larger bodied taxa, the ‘traditional’ zoo network is keeping amphibian species for reasons additional to threat status (Bowkett, 2014). These additional reasons may relate to the other variables examined in this study: Table 1 indicates that biases towards keeping larger bodied, more widely distributed, and less habitat-specific species in zoos all become more pronounced when only ZIMS institutions are considered. This may relate to zoos finding generalist species easier and cheaper to hold in captivity than closely related specialists; if such pairs of species are otherwise equally appealing to zoo visitors, it would be logical for zoos to select the species with fewer husbandry requirements (Martin et al., 2014a). Indeed, zoos may also actively choose to keep species of low conservation concern in order to learn husbandry techniques that can be applied to holding threatened relatives in the future (K. Johnson, pers. comm.). A future study tailored specifically to contrast species in conservation breeding programmes with their close relatives might reveal the traits associated with amenability to captive breeding. This and other potential drivers of ex situ selection for amphibians (e.g. coloration and activity cycles) might be interesting avenues for further comparative research.

Although high-altitude, island and tropical endemism are all considered to be important factors for predicting future threat status, species held in zoos are not more likely to have these traits. In contrast, species that rely on streams for breeding habitat are marginally less likely to be in ex situ programmes. This may be noteworthy, given that many stream-associated amphibians are purported to be at a higher risk of extinction (Lips et al., 2003; Stuart et al., 2004). However, given that closely related species tend to share many of these traits, leading to few contrasts and so low power using our approach, other analytical methods may be needed to explore these issues further.

The second implication of these patterns is that it is a relatively small number of institutions peripheral to the main zoos network, but highlighted by the AArk database, that are bolstering ex situ-threatened amphibian representation. While ZIMS institutions are mostly (albeit not 100% restricted to) ‘traditional’ zoos and aquaria, these ‘non-traditional’ institutions include specialist breeding centres, university departments and botanical gardens (and even a nunnery). Many of these institutions are also located within high-biodiversity countries in the tropics, which allows better integration of ex situ and in situ conservation strategies, reduces the risk of the transfer of novel pathogens from other species from outside the range distribution of the species, reduces acclimatization issues for captive species and increases the ability to obtain species for ex situ breeding without having to navigate difficult international administrative and veterinary barriers (Conde et al., 2011; Martin et al., 2014b). The important positive effect of AArk includes the support and coordination of these specialized institutions, allowing them to be integrated into and make meaningful contributions to the global ex situ community. AArk serves to highlight and recommend priority species for ex situ rescue or research; indeed, species which have been recommended for ex situ intervention by AArk have a higher likelihood of being held in zoo collections (see full analysis in Supporting Information Appendix S4). Of course, such recommendations are only one step: zoos are fettered by multiple goals and so multiple selection criteria (Fa et al., 2014). Despite obvious barriers (e.g. the costs associated with breeding large mammals ethically and sustainably ex situ), the establishment of other taxa-specific organizations modelled on AArk to help coordinate ex situ management of threatened species in less well-known and less centralized institutions should be considered. We propose the creation of a working group to produce an Avian Ark initiative.

We conclude by highlighting that even with the inclusion of institutions outside the ZIMS database, ex situ programmes as a whole are still not targeting those amphibian species predicted as being most at risk both imminently and in the near future, namely range-restricted habitat specialists. We therefore encourage all zoos to continue to increase their conservation-focused amphibian species holdings to help meet the ambitious ACAP targets, and to do so using strategic planning efforts that include multiple facets of conservation need. Given the previously discussed benefits of establishing ex situ programmes within the home range of target species, we also encourage North American and European zoos (where the majority of breeding programmes still occur) to establish more collaborative projects with institutions within the tropics, as has recently been achieved for several threatened species in Honduras (Honduras Amphibian Rescue and Conservation Centre, 2016). We acknowledge that zoos play other important roles in species conservation besides keeping threatened species (Bowkett 2014; Moss, Jensen & Gusset, 2015), and that simply holding endangered species in captivity is neither in itself a mark of conservation success (Harding et al., 2016) nor a guarantee of a successful breeding programme for all species held (Tapley et al., 2015). However, it is often a vital first step.