Introduction and Purpose

The chytridiomycete fungus Batrachochytrium dendrobatidis is an extremely important pathogen of wildlife with respect to biodiversity conservation. The fungus is responsible for amphibian mass mortality, population declines, and possible species extinction at an intercontinental scale (Fisher et al. 2009). Disease emergence across the Neotropics and Australia are the worst on record and have caused catastrophic declines of numerous hosts in both regions (Berger et al. 1998; Lips et al. 2006, 2008; Skerratt et al., 2007; Fisher et al., 2009). In contrast, since its detection in Europe in the late 1990s, infection with B. dendrobatidis has been associated with mortality of a handful of European species in the wild; far fewer than are known to carry infections (Bosch et al. 2001; Bosch and Martínez-Solano 2006; Garner et al. 2006; Bovero et al. 2008; Walker et al. 2008; Garner et al. 2009a; Bielby et al. 2009; Ohst et al. 2011; Sztatecsny and Glaser 2011; Rosa et al. 2012). Published sampling for Europe is strongly spatially biased, includes only a subset of European amphibian species diversity (www.bd-maps.eu/) and many studies are extremely limited in scope and effort (Adams et al. 2008; Federici et al. 2008; Ficetola et al. 2011). This fact suggests that the documented list of European amphibians known to be infected with B. dendrobatidis and experiencing mortality is underestimated. Alternatively, some un- or under-sampled species may truly be resistant to infection (e.g., Bielby et al. 2008; Luquet et al. 2012), but this cannot be distinguished from lack of exposure to B. dendrobatidis without combined spatial surveillance and experimentation.

At the species level, the link between infection and disease is not straightforward (Garner et al. 2011; Luquet et al. 2012). Even when detectable mortality of a highly susceptible host species has been reported for multiple locations (Bosch et al. 2001; Walker et al. 2010; Pasmans et al. 2010; Rosa et al. 2012), the same species may not suffer from the lethal form of the disease at other sites where it is infected at high prevalence (Walker et al. 2010). Conversely, at locations where infection is detected but lethal disease has not been observed, cryptic mortality may occur (Tobler and Schmidt 2010). Studies restricted to a spatial investigation of infection therefore present uncertainties when attempting to assess the risk that infection presents to a host species. Combining spatial studies of prevalence of infection with in situ or ex situ experimental tests of virulence can potentially distinguish between tolerance of infection versus unobserved lethal chytridiomycosis. For obvious reasons, clarifying the relationship between infection with B. dendrobatidis and amphibian mortality has greater conservation merit than simply describing parasite distribution (Garner et al. 2012).

Developing the database for resistance, tolerance, or susceptibility for all of Europe’s amphibians is an enormous research task. The potential conservation threat chytridiomycosis may pose to the European amphibian fauna calls for a more rapid, but still reasoned approach to risk assessment. Previous mass mortalities episodes and population declines recorded in Europe showed that the family Alytidae is a strong candidate for detecting both infection and lethal disease, if they occur. Mortality in one member of the genus Alytes (A. obstetricans) is the flagship case of lethal chytridiomycosis in Europe (Bosch et al. 2001, 2007; Walker et al. 2010; Rosa et al. 2012). Infection and death has also been described for two other members of the family, Alytes muletensis (Walker et al. 2008) and Discoglossus sardus (Bielby et al. 2009). Other members of the genus Alytes exhibit traits (prolonged larval period and restricted species range) that should predispose them to declines due to chytridiomycosis (Martínez-Solano et al. 2004; Gonçalves et al. 2007; Bielby et al. 2008), which strongly argues for targeted studies of previously unsampled Alytes spp.

The Betic midwife toad (Alytes dickhilleni) is a recently described species (Arntzen and García-Paris 1995) that is the sister taxon to A. muletensis and possibly Alytes maurus (Martínez-Solano et al. 2004; Gonçalves et al. 2007). A. dickhilleni is distributed across an extremely restricted species range, occurring in mountains and the nearby plains of six provinces located in the southeast of Spain. It is listed as vulnerable by the IUCN, but no attempt has been made to ascertain if B. dendrobatidis is infecting this species or capable of killing post-metamorphic animals, as is true for both A. obstetricans and A. muletensis (Bosch et al. 2001; Walker et al. 2008, 2010). In this manuscript, we describe a two-phase study of the distribution of infection and lethal disease in A. dickhilleni. In the first phase, we surveyed for the presence of infection in wild populations of A. dickhilleni, a study that covered the entire species range. In the second, after determining where infection was located, we used an ex situ experimental approach (Tobler and Schmidt 2010) to determine if the presence of infection was associated with mortality due to chytridiomycosis in animals near to, or completing, metamorphosis.

Methods

To ascertain the distribution and prevalence of infection in A. dickhilleni we first selected 30 sample sites covering the entire species range (Fig. 1). We focussed on sampling overwintered larvae because this life history stage has proven to be the most appropriate life history stage for detecting infection in two congeners (Walker et al. 2008, 2010). This was not possible at several sites, so young of the year tadpoles were sampled as an alternative (Table 1). To collect samples for molecular diagnosis of infection we swabbed the oral disc of each tadpole using a sterile swab (MW100, Medical Wire UK), a method proven to be reliable in previous studies of Alytes spp. tadpoles (Walker et al. 2008, 2010; Tobler and Schmidt 2010). At one location oral disks were excised from euthanized tadpoles and used for analysis. Our intention was to sample a minimum of 20 tadpoles however this could not be achieved comprehensively due to poor tadpole abundance at some locations and an absence of tadpoles at others. At these latter sites we sampled smaller numbers of recently metamorphosed animals or adults by taking toe clips (Table 1). In all cases recently metamorphosed animals were dead at time of sampling, and discovered that way. Swabs were refrigerated and stored dry while toe clips were stored in ethanol until DNA extraction and qPCR amplification as per Boyle et al. (2004). Extractions were diluted 1:10 before real-time PCR amplification, performed in duplicate, with B. dendrobatidis genomic equivalent (GE) standards of 100, 10, 1, and 0.1 GE. When only one replicate from any sample amplified, we ran this sample a third time. If the third amplification did not result in an amplification profile, we considered the sample negative for infection. We used a Potthoff–Whittinghill’s test of overdispersion to determine if risk of disease was homogeneous among sites (Potthoff and Whittinghill 1966). To do this, we condensed the UTM values for each sampled location into 10 × 10 km2 cells. For the null hypothesis, we assumed that the number of infections per cell followed an expected Poisson distribution.

Figure 1
figure 1

Sampling locations, including sites where Batrachochytrium dendrobatidis infection was detected (filled, black circles) and those where it was not (open circles). Small grey points show all known breeding sites for the species (from Bosch and González-Miras 2012).

Full size image
Table 1 Summary of Sampling Effort and qPCR Analysis of Field Samples (Swabs of Oral Disks were Collected from Alive Tadpoles at 25 Sites, Excised Oral Mouth disks from Euthanized Tadpoles were used at One Site and Toe Clips were Collected from Dead Metamorphosed Animals at Four Sites)
Full size table

We assessed the consequences of infection ex situ by broadly following the experimental protocol of Tobler and Schmidt (2010). Overwintered tadpoles were collected on the third week of September of 2009 from three locations, one where we had detected infected tadpoles (La Rahige in Sierra Tejeda), and two where we had not detected any evidence of infection (two sites located in Cazorla, Segura y Las Villas Natural Park). To minimize any possible effects of time in captivity, we restricted the collection of experimental tadpoles to 1 day per location. Consequences of this collection approach included uneven sample sizes among groups in the laboratory observations (Sierra Tejeda, n = 16; 2 sites in Cazorla, Segura y Las Villas Natural Park, n = 19 or 10).

We transported tadpoles to Bioparc Fuengirola where they were housed individually in plastic cups containing 1 l of aged tap water. Tadpoles were fed every two days and water was changed at each feed. Temperature in the experimental unit was kept at a constant 18°C and on a 12:12-h light schedule. Tadpoles were maintained in this manner until 14 days post-metamorphosis or death. Metamorphosis was defined as the day the tail was first observed to be resorbed (dark tail stub present, but stub not protruding beyond the “heels” of the hind limbs). To determine infection status, one hind toe was clipped from each animal either 7 days after metamorphosis or, in the event the animal died before 7 days post-metamorphosis, on the day of death. Toe clips were subjected to the same laboratory procedure, as were field surveillance samples. Animals that survived for 14 days post-metamorphosis were treated with Itraconazole (Itrafungol, Esteve) following Garner et al. (2009b) and returned to their site of collection after two months’ quarantine and no evidence of infection was detected using molecular diagnostic techniques applied after quarantine. We used a Cox proportional hazards (CPH) model to determine whether time to metamorphosis, infection with Bd as a categorical value (infected yes/no) or mean GE were significant predictors of survival of tadpoles collected from the field. Because infection data were unavailable for one of the Sierra Tejeda animals that died on day 42, it was excluded from the CPH model.

Results

We sampled a mean of 14.4 individuals (min. 3, max. 20) at 30 breeding sites distributed across the six provinces where A. dickhilleni is endemic (Fig. 1; Table 1). The risk of infection was not homogeneously distributed across sites (Potthoff–Whittinghill’s test of overdispersion; T = 11004.33, P = 0.001) and infection was detected at three sites, clustered in two 10 × 10 grid squares (Fig. 1). The grid square containing two of the sites where infection was detected includes the recreational area of La Rahige located in the Sierra Tejeda, Málaga. One of these two sites is a completely artificial breeding site (cattle tank, La Rábita Fountain, site Málaga 1, Table 1), while the other is a man-made pool (La Rahige, site Málaga 2, Table 1) located in small, slow-moving, naturally occurring and permanent stream within the recreational area. Stream flow downstream from La Rahige has been partially dammed to increase water depth in the pool for recreational purposes. The third positive location is an undisturbed stream (Guadahornillos, site Jaén 2, Table 1) located near to the Roblehondo Biological Station CSIC in Cazorla, Segura y Las Villas Natural Park, located in another grid cell several cells away from the cell containing La Rábita Fountain and La Rahige.

Prevalence at the two Sierra Tejeda sites varied: tadpoles at La Rábita Fountain were 100% infected, while at La Rahige only 36% of tadpoles were infected. In comparison, tadpoles at Guadahornillos were more similar to La Rábita Fountain in terms of prevalence (95%). Burden of infection was, on average, high at all three locations and mean GE was similar amongst sites (Table 1). All recently metamorphosed juveniles and the two adults that were tissue sampled also tested negative for infection.

Survival of post-metamorphic A. dickhilleni in captivity was not consistent across treatments. Experimental tadpoles from the Sierra Tejeda were the only animals that tested positive for infection at time of death or seven days post-metamorphosis and experienced substantial mortality. Tadpoles collected from the sites where infections were not detected during the field survey experienced significantly less mortality when compared to the Sierra Tejeda group (Fig. 2). After excluding the one animal for which we lacked infection data, animals that died in the Sierra Tejeda group (12/15) were almost all infected (10/12) and the majority died within 60 days after the start of the laboratory observations (n = 10, minimum number of days to death = 6, mean number of days to death ± SD = 24.6 ± 15.4, mean GE ± SD = 12,741.4 ± 10,734.8, range GE = 0–29,367.6). The two other tadpoles from Sierra Tejeda that died did so 143 and 261 days after the start of the laboratory observations and exhibited infectious burdens of, respectively, 0.1 and 3.9 mean GE. Cox proportional hazards analysis revealed that infection status (yes/no, P = 0.002) but not mean GE (P = 0.470) nor time to metamorphosis (P = 0.329) significantly increased probability of death. Being infected increased the relative instantaneous mortality hazard by a factor of 15.7 times.

Figure 2
figure 2

Survival curves for tadpoles sampled at three locations, two within Cazorla Segura y Las Villas Natural Park (CR, R) and at Sierra Tejeda (LR).

Full size image

Discussion

The spatial pattern and number of infected populations we detected strongly suggests that B. dendrobatidis is a recently emerged parasite of A. dickhilleni. There are few robust spatial studies of B. dendrobatidis distribution available for European hosts, but of those that are published the majority involve congeners of our study species, and report a significantly greater proportion of sites with infections than we found (Walker et al. 2008, 20% sites with infected animals; Walker et al. 2010, 25% sites with infected animals; Tobler et al., 2012, 61.5% sites with infected animals; this study, 10% of sites with infected animals). Walker et al. (2008, 2010) identified other Iberian locations where recent introductions of B. dendrobatidis into Alytes spp. populations had occurred, including human-mediated introduction into populations of A. muletensis in the 1990s with little or no evidence of post-introduction dispersal (Walker et al. 2008). The pattern described by our study fits this latter case: two highly distinct foci of infection where prevalence is at or near to saturation in overwintered tadpoles (La Rábita Fountain and Guadahornillos), indicative of two recent introductions, and limited evidence of fungal dispersal to nearby sites (La Rahige: located near to a saturated population, but with significantly lower prevalence). Both of our proposed sites of introduction are locations where human activity that could promote pathogen introduction is common. Guadahornillos is located just 1.5 km away from an important location for biological research and Sierra Tejeda Natural Park is a multi-use recreational park frequently visited by amateur herpetologists looking for the species at its terra typica.

Our spatial study revealed no observable evidence of disease-linked mortality in recently metamorphosed juveniles. Mass mortality at metamorphosis is the signature of lethal chytridiomycosis in A. obstetricans (Bosch et al. 2001, 2007; Walker et al. 2010) and some field mortality has been observed in A. muletensis (Walker et al. 2008; JB personal observations), but we never encountered large numbers of dead animals of any life history stage at any location. The few recently metamorphosed and dead animals that we did encounter in nature did not test positive for infection (Table 1). The results of laboratory observations of field-collected animals, however, showed that it is highly probable that A. dickhilleni metamorphs at Sierra Tejeda are experiencing substantial, cryptic mortality as nearly 70% of overwintered tadpoles collected from this area died in captivity. Mortality was significantly associated with infection status, typically occurred soon after the onset of the laboratory observations and was commonly associated with heavy burdens of infection.

These results also provide the first experimental evidence that infected tadpoles are capable of maintaining infections over long periods of time without external forcing of infection (Briggs et al. 2010) and dying as a result. This finding is important because increased exposure to B. dendrobatidis through forcing of infection can be strongly influenced by transmission from other infected hosts, and has been linked to increasing risk of mortality through the accumulation of fungal load (Briggs et al. 2010; Vredenburg et al. 2010) but the relationship between host density and transmission is unclear (Rachowicz and Briggs 2007). Because our experimental animals were housed individually we can exclude transmission among hosts as an influence, removing the effect of persistent external forcing from the individual disease dynamic. Most of the Sierra Tejeda tadpoles spent more than 20 days without being exposed to any transmission vector, two greater than 140 days, and still died after metamorphosis. Infections acquired earlier during development were therefore sufficient to elicit death long after the initial transmission event, as was the case for A. obstetricans tadpoles (Tobler and Schmidt 2010). This may be an indication that costs accrued by tadpoles during this time relate to post-metamorphic death, which has been seen in species with relatively short-larval periods (Garner et al. 2009a; Luquet et al. 2012). It seems sensible, therefore, that a more prolonged larval period with an associated prolonged period of sustained infection should be costly to the infected host.

Studies of chytridiomycosis in Alytes spp. do not always provide a definitive link between mortality caused by chytridiomycosis and species decline. In Peñalara Natural Park, A. obstetricans was driven locally extinct due to disease emergence (Bosch et al. 2001; Martínez-Solano et al. 2003; Bosch and Rincón 2008). A recent publication by Tobler et al. (2012) outlines the alternative and describes a lack of a clear relationship between the presence of infection and host population dynamics although the authors do postulate that their system may have suffered historical declines due to disease. We lack quantitative data on population responses at locations where A. dickhilleni is infected. Irrespective of the inconsistencies, evidence for introduction of B. dendrobatidis into populations of a high risk species coupled with evidence for substantial mortality due to disease is cause for conservation intervention. Bioparc Fuengirola and its Foundation, in association with the Amphibian Ark, have established a biosecure facility for the maintenance of a disease-free captive assurance colony of A. dickhilleni. This is coupled with more general conservation activities including habitat restoration and improvement, local education programs, and other efforts to raise awareness regarding amphibian decline, conservation, and chytridiomycosis. Disease monitoring and the development of methods to mitigate infection in the wild are underway, as part of the in situ, “Betic Midwife Toad Conservation Project” lead by Bioparc Fuengirola and in collaboration with the National Museum of Natural Sciences of Madrid (CSIC).