Turtles are one of the most imperiled groups of vertebrates and about half of all species are threatened, endangered, or critically endangered (Böhm et al. 2013; Rhodin et al. 2018). A first step to prevent more turtle species from becoming extinct is determining the status of their populations in the wild. Turtles are substantially less diverse than other major groups of reptiles, yet the large majority of turtle lineages remain unstudied. With a lack of information on most species, a priority emerges to understand the population status of unstudied species and this is especially important for species that have restricted geographic distributions (Lovich and Ennen 2013; Lovich et al. 2018; Gibbons and Lovich 2019).

Species that are restricted to a very small geographic area (< 50,000 km²), often termed microendemic (Terborgh and Winter 1983; Riddle et al. 2011), are at increased risk of extinction because if something happens within their limited distribution, they can quickly become extinct (International Union for the Conservation of Nature [IUCN] Standards and Petitions Committee 2019). For example, the Viesca mud turtle (Kinosternon hirtipes megacephalum) existed within a limited geographic area in Coahuila, México, and has been extinct since first being described in 1981 (Iverson 1981; Rhodin et al. 2018). A first step to protecting such taxa is to understand basic demographic characteristics such as how many adults, juveniles, and hatchlings are in the population and to document long-term trends in recruitment and density in their populations (Primack 2012).

Given the urgency to conserve turtles, the IUCN Tortoise and Freshwater Turtle Specialist Group has compiled a list of priority species to guide and support research on species that are considered threatened, endangered, or data deficient. Understanding the basic population demographics and long-term trends of populations of microendemic species could contribute to conserving unique evolutionary lineages that are on the list of priority species. However, because many turtles are long-lived organisms (including kinosternids), most of the time long-term population assessments are required, sometimes for decades to make conservation and management decisions (Garcés-Restrepo et al. 2019; Howell et al. 2019). However, when there is a suspicion that species could be highly imperiled and basic life history data are missing, it is important to utilize methods designed to predict long-term projections of a population based on short-term and compiled data from several sources (Brook et al. 2000). These methods may not be optimal, but they still allow elucidation of potential outcomes to make informed conservation decisions (Burgman and Possingham 2000) using the data available.

Population viability analysis (PVA) has been specifically designed to inform future conservation priorities. The required data for a PVA could be pooled from other studies, estimated from related species, or even based on theoretical population parameters (Famelli et al. 2012; McGowan et al. 2017). A PVA requires basic information about a population such as sex ratio, mortality rates, and estimated population size in order to generate a probability distribution for whether the population will go extinct over a predetermined length of time (Morris et al. 1999). In turtles, PVAs have been used in different ways. For example, Enneson and Litzgus (2009) used PVA to assess the risk of extirpation in a population of Clemmys gutatta in Canada, Famelli et al. (2012) explored several survivorship scenarios in Hydromedusa maximiliani in Brazil, and Mazaris et al. (2005) simulated a PVA for Caretta caretta to understand the effect that survivorship at early stages had on long-term population trends. In another study, Spencer et al. (2017) used a PVA to test several different long-term scenarios for populations of Chelodina longicollis in Australia, finding that the most plausible scenarios of future viability require head-starting and eliminating anthropogenic threats. Other studies (Doak et al. 1994; Rivera and Fernández 2004; Canessa et al. 2014; Maulany et al. 2017) support the notion that a PVA can be used to inform conservation and management decisions on turtle populations.

The Mexican rough-footed mud turtle (Kinosternon hirtipes) and its 6 described subspecies have a wide geographic distribution between the 2 largest mountain ridges of México, the Sierra Madre Occidental and the Sierra Madre Oriental. This geographic distribution begins in southern Texas and extends to central México, reaching the Valley of México (also known as Anahuac Valley or México City Valley) and central Michoacán (Iverson 1981, 1985). Of the 6 subspecies that have been described, one is already extinct (K. h. megacephalum) (Turtle Taxonomy Working Group [TTWG] 2017), one has a wide distribution from southern Texas to central Michoacán (K. h. murrayi), and the remaining 4 are microendemic to a specific lake or small endorheic drainage basin. For example, K. h. tarascense is only distributed in Lake Pátzcuaro in Michoacán, K. h. magdalense is restricted to the Magdalena Valley in Michoacán, K. h. chapalense is endemic to Laguna Chapala and Laguna Zapotlán near Guadalajara, and K. h. hirtipes is restricted to the Valley of México near México City (Iverson 1981).

The conservation status of K. hirtipes according to IUCN is “least concern” (van Dijk et al. 2007). This assessment is largely based on studies focused on the most wide-ranging subspecies, K. h. murrayi, in Chihuahua (Iverson et al. 1991), Texas (Platt et al. 2016a, 2016b; Smith et al. 2018), and Michoacán (Enríquez-Mercado et al. 2018). There is very little information available on the other 4 living subspecies. The only study that attempted to assess the conservation status of some of the subspecies of K. hirtipes found them to be extremely rare in their known distribution. Reyes-Velasco et al. (2013) attempted to assess the conservation status of K. h. chapalense in Laguna Chapala, K. h. magdalense in the Magdalena Valley, and K. h. megacephalum in Viesca, Coahuila. They captured very few K. h. chapalense and K. h. magdalense and confirmed that K. h. megacephalum is in fact extinct. Although the Mexican government gives K. hirtipes special protection status (Macip-Ríos et al. 2015), it is alarming that this species is still considered least concern by the IUCN when one of its unique lineages has gone extinct, and the other subspecies are rare in their native habitats.

In this article we address the conservation status of the 4 subspecies of K. hirtipes that are distributed across central México. To do that, we conducted a PVA on K. h. murrayi (the subspecies with the most data available) as our best (short-term) approach to conduct projections on the future viability of this taxon, and then extrapolated to the other subspecies. We understand the limits of our approach, but due the limited available information for these taxa and the reduced or nonexistent habitat, our study may be useful for future conservation efforts. Recently, Enriquez-Mercado et al. (2018) and Aparicio et al. (2018) reported on the basic population ecology of K. h. murrayi in Michoacán and for the first time reported K. h. tarascense from a Lake Pátzcuaro locality. However, to date there are no studies that address the conservation status of the other K. hirtipes subspecies. We continued previous efforts by Reyes-Velasco et al. (2013) to address the conservation status of the subspecies of K. hirtipes. The results of this research are a step toward understanding the basic conservation and management needs for each of the subspecies of K. hirtipes in central México.

METHODS

Study Site. — We sampled K. hirtipes throughout its range in central México. This includes Patzcuaro Lake in Michoacán for K. h. tarascense, the Magdalena Valley in Michoacán for K. h. magdalense, Chapala Lake in Jalisco for K. h. chapalense, and the Xochimilco wetland area in México City Valley for K. h. hirtipes. To maximize capture probability, each site was visited during the peak activity season, which occurs during the wet season (June–November). These study sites are distributed across the Mexican Transvolcanic Belt (MTB), characterized by many isolated drainage basins created during the last phase of the MTB uplift over the last three million years (Ferrari et al. 2012). Many of these basins are home to other vertebrate microendemic linages, such silverside fishes in Chapala Lake and Patzcuaro Lake (Moncayo-Estrada and Buelna-Osben 2001; Lyons et al. 2019), ambystomid salamanders in Lake Patzcuaro (Duellman 1961), and aquatic snakes in Patzcuaro Lake and Chapala Lake (Conant 1961). Selected field sites were chosen from the historical localities where K. hirtipes subspecies have been recorded and documented by Iverson (1981, 1992) and Reyes-Velasco et al. (2013).

The limnology of the permanent water bodies of the MTB is mainly correlated with the seasonality of central México's rain and temperatures (Sigala et al. 2017). Most of the study sites are above 1500 m above sea level (asl). For example, Laguna Chapala is at 1537 m asl and localities in Michoacán and México City Valley are at 2400 m asl. Climate among the sampled sites is temperate at México City and Pátzcuaro basins, while the Chapala, Zapotlán, and Cuitzeo basins are warmer (García 2004). Most of the study sites have been highly modified since pre-Columbian times, and water bodies are heavily polluted in México City Valley, Chapala Lake, and Pátzcuaro Lake (National Water Commission of Mexico [NWCM] 2010). Aquatic habitats are those typical of eutrophic, lentic, temperate freshwater habitats, with high densities of cattails (Typha spp.) and introduced water hyacinths (Eichhornia spp.).

Capture Protocol. — During each sampling event we deployed 18 hoop traps, 4 fyke nets, and a seine (when possible) to sample for the presence of K. hirtipes. On several occasions, turtles were captured by hand. Traps were baited with canned sardines in oil or fresh fish. At each site we deployed the traps for at least 2 nights and checked them twice a day. If turtles were not captured in a particular trap after the first night of trapping, we moved that trap to another location at the same locality to increase the probability of capture by extending the sampling area.

Upon capture, turtles were assigned unique numbers and marked using a triangular file following Cagle's (1939) shell-notching method. Morphological measurements of each turtle were recorded, including maximum carapace length (mm), and body mass (g) using a digital caliper (Swiss Precision Instruments, Inc., Melville, NY) and a spring scale (Pesola Präzisionswaagen AG, Schindellegi, Switzerland). Select females were brought back to the laboratory at Universidad Nacional Autónoma de México in Morelia, Michoacán, where x-rays were taken to determine if turtles were gravid (Gibbons and Greene 1979; Hinton et al. 1997). All turtles were returned to their capture location following data collection. Geographic location and habitat type (categories of habitat types) were also recorded for each turtle.

Population Viability Analysis. — We follow Famelli et al. (2012) on their PVA approach.

Data were pooled from different studies on K. h. murrayi (Iverson et al. 1991; Enriquez-Mercado et al. 2018) (Table 1). Other data, such as reproductive data and hypothetical initial population size were also included using field (unpublished) data (Table 1). Inputs for the PVA analysis also included information from other studies on other kinosternids (Iverson 1991; Macip-Ríos et al. 2011). The PVA model was constructed with VORTEX ver. 10 (Lacy 1993, 2000). Table 1 shows the input data that we used for our modeling. Some entries in VORTEX were left as default values for the software such as genetics (number of neutral loci to be modeled = 5; number of loci to be subject to mutation = 2 with a rate of 0.001), no supplementation (head-starting), mate monopolization (100% breeding males), and inbreeding. We included 1 natural catastrophe in our simulations. Table 1 shows how VORTEX considers the effect of a natural catastrophe on the population in terms aquatic habitat loss (%), and how the natural catastrophe directly affects reproduction and survivorship on the modeled population. We ran 3 simulations (1000 iterations per simulation), each simulation representing a different scenario: pessimistic, intermediate, and optimistic. A pessimistic scenario was described as 100 individuals (50% of a typical Kinosternon population based on earlier research [Forero-Medina et al. 2007; Macip-Ríos et al. 2009, 2011; Vázquez-Gómez et al. 2016; Aparicio et al. 2018; Enríquez-Mercado et al. 2018; among others]) as the initial population size (which we consider as a low population size), a limit of 1500 individuals at carrying capacity, and an adult harvest of 50 turtles (25 females and 25 males) each year; an intermediate scenario, with an initial population size of 150 individuals, a carrying capacity of 2250, and an adult harvest of 26 turtles (13 females and 13 males); and an optimistic scenario, with an initial population size of 200 individuals, a higher carrying capacity of 3000 individuals, and a reduced harvest of 2 adults (1 male and 1 female) per year.

Table 1. Input parameters used for the population viability analysis in VORTEX ver. 10 (Lacy 2000).

View Fullscreen

Statistical Analysis. — Comparisons among the scenarios were performed using a one-way ANOVA. Body size and mass among turtle population and localities were compared with a nonparametric Kruskal-Wallis test. Statistical analyses were conducted in JMP ver. 5.0.1 (SAS Institute 2002). Statistical analyses were performed with an α = 0.05.

RESULTS

During 12 sampling events conducted throughout this study, a total of 216 turtles were captured and processed, including 177 Kinosternon integrum, 62 K. hirtipes, and 2 Trachemys scripta. Table 2 highlights the number of individuals collected by locality, taxon, sex ratio, and main aquatic habitat type where turtles were captured. Carapace length (Kruskal-Wallis, H_7,213 = 38.10, p < 0.0001) and body mass (Kruskal-Wallis, H_7,213 = 43.70, p < 0.0001) varied significantly across the surveyed populations. Wilcoxon pairwise comparison post hoc tests showed that the only statistical differences were between the larger and heavier K. integrum from Jamay, Jalisco, and the Magdalena basin, Michoacán, and K. h. tarascense in both carapace length and body mass (Table 2). All K. integrum had a mean carapace length above 114 mm and a mean body mass above 330 g, while all K. hirtipes populations had a mean carapace length below 110 mm and a mean body mass below 252 g (see Table 2 for details).

Table 2. Turtles captured during the study in different localities and habitat type. n = total captured individuals; CL = carapace length. For Wilcoxon pairwise comparison, different letters mean statistical differences (p < 0.05).

View Fullscreen

PVA showed that under the optimistic scenario, the modeled population displayed a slightly positive value for the intrinsic rate of population increase (average r = 0.001 ± 0.084 SD), a 0.001 extinction probability, and a final population size of 338 individuals after a simulation of 100 yrs. Under the intermediate scenario, the intrinsic rate of population increase was negative (average r = –0.012 ± 0.089 SD), a 0.016 probability of extinction, and a final population size of 82 individuals after a simulation of 100 yrs. Finally, the pessimistic scenario showed a negative intrinsic rate of population increase (average r = –0.028 ± 0.104 SD) and high probability of extinction of 0.306, with a final population size of 11 turtles after a simulation of 100 yrs (Table 3).

Table 3. Comparison between scenarios among the modeled population of Kinosternon hirtipes murrayi. r = intrinsic rate of population growth ± SD, calculated by VORTEX; PE = probability of extinction.

View Fullscreen

When comparing the scenarios on the modeled population we found statistical differences between scenarios (F_2,302 = 166.43, p < 0.0001). The optimistic scenario showed a larger and distinctive growth compared with the intermediate and pessimistic scenarios. The intermediate scenario was also significantly larger than the pessimistic scenario. Figure 1 shows the modeled growth over 100 yrs for each scenario.

View Figure

• Download as Powerpoint
• Download Figure

The optimistic scenario (which started at 200 individuals) reached the maximum population size after 45 yrs, with a slow decline beginning at about year 55. Under this scenario the final population size was 338 individuals (138 more than the initial population size). Carrying capacity (set at 3000) was the main factor limiting population growth under the optimistic scenario. For the pessimistic scenario, the maximum population size was reached at year 35 of the projection, and the population declined severely after year 40. Under the pessimistic scenario, the modeled population went extinct near the 95th year of projection (Fig. 1), with only 11 individuals remaining at the end of the simulation. Under this scenario, harvest was the main factor affecting population growth and stability. For the intermediate scenario, the modeled population reached its maximum population size at year 40 of the projection. Population declined after year 45 with only 82 remaining turtles in the population at the end of the simulation. Harvest and carrying capacity were the main factors affecting the intermediate scenario.

DISCUSSION

Our results indicate that the subspecies of K. hirtipes are facing an alarming conservation threat. Other than the widely distributed K. h. murrayi, the remaining subspecies are imperiled and should be considered endangered taxa. Reyes-Velasco et al. (2013) also reported very low capture rates for K. h. chapalense and K. h. magdalense. Other kinosternid studies (Macip-Ríos et al. 2009, 2011, 2018; Reyes-Grajales et al., unpubl. data, 2020) demonstrated that the sampling techniques we used (baited hoop traps and fyke nets) are highly effective (Mali et al. 2012; Enríquez-Mercado et al. 2018; Goulett et al. 2019). Therefore, it seems that the reason why “catchability” or detectability of these turtles is low is a consequence of actual rarity.

Historically, or at least from some chronicles from naturalists at the beginning of 20th century (Gadow 2012), K. hirtipes was once abundant in Chapala Lake and in the Xochimilco wetlands (Valley of México). Today, these subspecies have not been collected at their type localities for at least 20 yrs (Reyes-Velasco et al. 2013). Our PVA of K. h. murrayi showed that only under the optimistic scenario could a population be sustained during at least a 100-yr time period with very low extinction probability. Intermediate and pessimistic scenarios showed negative intrinsic population growth, although the pessimistic scenario was the only one with a high extinction probability. We recognize the limited interpretation of our PVA results due the use of data from other species, but the differences between our scenarios have value. According to our results, turtles could prevail with a minimum of 200 individuals in population size and reduction in harvest, a scenario that might be applicable to each of the microendemic subspecies. Other studies reported typical kinosternid population sizes of around 200 individuals (Forero-Medina et al. 2007; Macip-Ríos et al. 2009, 2011; Vázquez-Gómez et al. 2016; Aparicio et al. 2018; Enríquez-Mercado et al. 2018; among others). Adult survivorship is fundamental for turtle population persistence (Heppell 1998; Eskew et al. 2010; Macip-Ríos et al. 2011) and limiting harvest of K. hirtipes (most of the time untargeted as a bycatch of fishing using gillnets) is necessary for improving their population persistence over time. However, the PVA's intermediate and pessimistic scenarios indicated that for a small population size (arbitrarily set at 100 and 150, respectively) and even with low harvest (intermediate scenario), the turtle population modeled did not show a significant increase in a hypothetical 100-yr period. For that reason, we conclude that assisted population management strategies are needed.

During the course of this study, we located at least 2 natural populations of K. h. tarascense, a neglected turtle subspecies described based on specimens from the local market in Pátzcuaro (Iverson 1981). One of the populations was located in the northern part of Pátzcuaro Lake, very close to the town of San Jeronimo Purenchecuaro, and the other population is associated with a natural spring in the eastern part of the Pátzcuaro basin. Both populations were composed of very few individuals, but at least we captured yearlings, gravid females, and healthy adults. Like other large bodies of water of central México, Pátzcuaro Lake has been heavily transformed since pre-Columbian times. People from the Pátzcuaro basin used to exploit the endemic silverside fish (Chirostoma spp.) and small goodeid fishes using traditional fishing methods (butterfly nets for surface catch). However, the use of gillnets started in the 1950s, with the introduction of carp (Cyprinus carpio) and cichlids (Orbe-Mendoza et al. 2002). The change in fishing techniques probably diminished adult numbers of K. h. tarascense through bycatch. Informal interviews with native people (older than 60 yrs) from the lake revealed that consumption of turtle meat was a very common practice in the past for traditional medicine (to cure respiratory illness) or food. Local people also mentioned a severe decline in turtle abundance. They reported that Kinosternon were very abundant and conspicuous 30 to 50 yrs ago but are now very rare. A similar pattern of population decline has been documented for the Pátzcuaro's endemic and obligated paedomorphic salamander Ambystoma dumerillii (Huacuz 2002; Parra-Olea et al. 2012).

Kinosternon hirtipes magdalense was even more difficult to find and capture at the localities we surveyed. Nonetheless, we were able to find a new, previously unreported, locality for this subspecies in Laguna El Moral. The Magdalena basin once had a series of springs along a small valley. This spring system (similar to those described for the habitat of the extinct K. h. megacephalum) has changed greatly due to water management and the construction of at least 3 dams: San Juanico (the largest), El Moral (Laguna El Moral), and the Guadalupe spring (now a shallow lake with a small dam). The construction of these dams severely modified the habitat and the connectivity among the former springs. Iverson (1981) mentioned that hydraulic infrastructure at the lower (eastern) part of the Magdalena basin allowed the interchange of aquatic fauna from the formerly isolated basin with that in the Balsas River basin. This exchange included the colonization of K. integrum from the Balsas River basin into the Magdalena basin (Iverson 1981).

We did not find any K. h. chapalense in our surveys at Chapala Lake. We surveyed the southern part of the lake with no success. Nevertheless, due to the size (1100 km²) of Chapala Lake, it is highly likely that populations of K. h. chapalense persist in some other areas. We set traps in the main lake, in the associated wetlands, and in a very complex system of irrigation canals around the lake (always with permanent water), all without success. Similarly, Reyes-Velasco et al. (2013) did not capture any K. h. chapalense during several trapping occasions along the eastern and northern part of the lake. The lake has been heavily modified for aquaculture, irrigation, and tourism, and also fed by the Lerma River which collects waste from the Toluca Valley through Michoacán. Numerous factors, including introduced species, non-sustainable practices, and a high level of habitat degradation render Chapala Lake as unsuitable for several fish species (Moncayo-Estrada and Buelna-Obsen 2001; Moncayo-Estrada et al. 2012; Lyons et al. 2019). Reyes-Velasco et al. (2013) did not capture any K. h. chapalense during several trapping occasions along the eastern and northern parts of the body of water.

We located 4 individuals of K. h. chapalense at the same site in Lake Zapotlán where Reyes-Velasco et al. (2013) also reported individuals. We sampled turtles in other locations around the lake with no success. Zapotlán Lake is much smaller than Chapala Lake and Pátzcuaro Lake, but shares the same habitat degradation problems (Greenberg et al. 2008). Fisheries have been reduced in the last years at Zapotlán Lake (Michel-Parra et al. 2014). Nevertheless, organized fishing cooperatives could further reduce turtle populations.

We did not find any individual of K. h. hirtipes. The alarming degradation of the Valley of México wetland during the last 50 yrs also produced dire consequences to the local fish fauna (Griffiths et al. 2004; Vázquez-Silva et al. 2017), salamanders (Ambystoma mexicanum) (Zambrano et al. 2010), and other amphibians and reptiles. There are no records of K. h. hirtipes from the Xochimilco wetland since the last museum specimens were collected in the late 1970s (Sistema Nacional de Información sobre Biodiversidad [SNIB], Comisión Nacional para el Conocimento y Uso de la Biodiversidad [CONABIO] 2015). However, during our surveys in México City (16 August 2019), staff from the Centro de Investigaciones Biológicas y Acuícolas de Cuemnaco–Universidad Autónoma Metropolitana (CIBAC-UAM) showed us a male and a female of K. h. murrayi captured in Xochimilco canals, an unexpected taxon in the Valley of México, since the native subspecies is K. h. hirtipes.

The few K. hirtipes that were captured at each site relative to K. integrum is alarming because K. integrum is native to only 1 locality (Chapala Lake). In Chapala Lake, K. integrum was common in irrigation canals, and in 2 trapping nights we were able to capture 90 individuals in a single locality. Kinosternon integrum was rare in Zapotlán Lake, where it is invasive (Iverson, 1981). Apparently, the sites we surveyed in Zapotlán Lake had low turtle densities of all turtles. In the Magdalena basin, invasive K. integrum is dominant in most water bodies, outnumbering captures of K. h. magdalense 10 to 1 at the San Juanico dam and are conspicuous in the Cotija River and several cattle ponds in the basin. Kinosternon integrum is the most conspicuous turtle in the Valley of México (apparently introduced), along with exotic Trachemys scripta elegans. There are also records (A. Vergara, CIBAC-UAM, pers. comm., September 2019) of Kinosternon leucostomum, map turtles (Graptemys geographica), alligator snapping turtles (Macroclemmys temiminckii), cooters (Pseudemys spp.), and softshells (Apalone spp.) in the same Xochimico canals.

There is no proof that the interaction of K. integrum with K. hirtipes could lead to a decline of the latter. However, there is some evidence suggesting that competition for basking sites (Cadi and Joly 2003) and other negative effects of introduced red-eared sliders (T. s. elegans) in native turtle populations (Thomson et al. 2010). It is also important to determine how K. integrum colonizes new basins and if this colonization is natural or is promoted by human activities. Vigilant efforts should also be made to keep K. integrum from colonizing the Patzcuaro basin, since there is an established population 1 km from the lake shore (Aparicio et al. 2018). Studies comparing environmental, dietary, microhabitat, and thermal niches between K. integrum and K. hirtipes should be conducted to look for potential competition between both species.

In summary, our results suggest that the restricted range of K. hirtipes subspecies faces serious conservation challenges. The Mexican government should consider elevating the protection status of these taxa, and even consider these microendemic lineages in its conservation priority program. In México, aggressive conservation efforts for turtles have only been conducted for tortoises (Turtle Conservancy 2019) and some ex situ programs for Dermatemys mawii in farms (Rangel-Mendoza et al. 2014). One subspecies of K. hirtipes is already extinct (Iverson 1981; Reyes-Velasco et al. 2013) and because the other subspecies (excluding K. h. murrayi) have such restricted distributions, they are highly vulnerable to local extirpation or even extinction in cases such as K. h. tarascense and K. h. magdalense, which are restricted to a very small distributions. These microendemic subspecies also occur in highly modified habitats close to México's largest cities. It appears that aggressive conservation programs need to be developed for these taxa, including establishment of assurance colonies or ex situ colonies in zoos for future reintroduction programs (one is already established at the San Antonio Zoo for K. h. murrayi; San Antonio Zoo Conservation Efforts 2019). Efforts should also be made to identify remaining populations that occupy minimally disturbed habits (especially streams) that could be rigorously protected. Finally, phylogenetic work is sorely needed to test the hypothesis that these allopatric microendemic subspecies might indeed be distinctive species.