Data Collection

PNA is composed of 3 parcels that rest on both sides of the Anosyennes Mountains in extreme southeast Madagascar. This chain of mountains constitutes a boundary between eastern tropical forest and semiarid spiny forest that acts as a remarkable rain barrier separating the 2 biomes (Fenn 2003). The largest parcel (parcel 1; Fig. 1) represents eastern humid forests, and the smallest (parcel 3) is in the transitional zone between humid and arid forests. These parcels do not correspond to radiated tortoise natural habitat, and accordingly, sokake does not occur in these areas. On the other hand, parcel 2 (dry spiny forest, 12,420 ha) mostly comprises suitable habitat for the sokake, and ANGAP staff members had confirmed before our expedition that tortoises still occurred in this parcel. We carried out field work in parcel 2, searching for tortoises with a team of 5–7 people for 4 days in the vicinity of Mangatsiaka (Fig. 1) and 6 days in the area west of the village of Ihazofotsy, in March 2007. Searches were carried out daily between 0700 and 1100 hours, and between 1500 and 1830 hours in different portions of the park. Searchers walked in parallel, with ca. 20 m between each person. Habitat was thoroughly inspected, especially features where radiated tortoises usually hide (bushes, fallen trees, depressions in the ground, etc.). Unlike line-transect distance sampling (Buckland et al. 1993), this method does not allow for an estimation of the overall population size but was the only possible choice considering the scarcity of tortoises in PNA. Furthermore, distance sampling greatly overestimates population size if an insufficient number of captures are made (Freilich et al. 2005), which may explain surprisingly high estimates recently obtained for the flat-tailed tortoise Pyxis planicauda in western Madagascar (Young et al. 2008). For every encountered sokake, sex and weight were noted, a blood sample (0.1–0.5 mL) was drawn and stored in blood storage buffer (Longmire et al. 1992), marginal scutes were notched to provide a unique visual mark for each tortoise (Cagle 1939), and a passive integrated transponder (PIT tag; Gibbons and Andrews 2004) was inserted in every tortoise with a body weight > 0.5 kg.

In order to compare indices of genetic diversity and evaluate the extent of population differentiation, we used all available genetic data from other populations that have been analyzed elsewhere (Rioux Paquette et al. 2007; Rioux Paquette et al., in press). Previously sampled localities are all indicated on Fig. 1. DNA extraction, polymerase chain reaction amplification, and microsatellite genotyping at 13 loci were performed following the methodology described in Rioux Paquette et al. (2005, 2007). PNA population size structure data and sex ratio were compared to data from previously sampled exploited and nonexploited tortoise populations (Rioux Paquette and Lapointe 2007) using χ² tests.

Genetic Analyses

Based on microsatellite allele frequencies, previous analyses have found 2 main populations of A. radiata separated by the Menarandra River (Rioux Paquette et al., unpubl. data). For comparison purposes, in addition to these 2 groups (henceforth designated as “West” for all coastal populations west of the Menarandra River and “Central” for all populations comprised between the Menarandra and Manamobovo rivers), we also compared results with other peripheral populations showing significant levels of differentiation: Bezaha-Mahafaly north of the Mahafaly Plateau, Sakotoavo on the plateau, and Ifotaka, just west of the Mandrare River (Fig. 1). Pairwise F_ST values were computed among these 5 groups and the PNA population with the software Arlequin (Schneider et al. 2000) and significance was assessed with 1000 permutations. Furthermore, 3 indices of genetic diversity were computed for PNA and the other populations: heterozygosity (both observed H_O and expected H_E), allelic richness Ar (a measure of the number of alleles corrected for different sample sizes), and the inbreeding coefficient F_IS. These analyses were performed with Fstat (Goudet 1995) and Genepop (Raymond and Rousset 1995), and 95% confidence intervals (CIs) were calculated for each parameter.

The sokake population at PNA was not sampled in previous genetic surveys because it was thought to have served as a release area for confiscated tortoises (Leuteritz et al. 2005). This would have introduced a bias in studies aimed at depicting natural patterns of genetic variation in the species (Rioux Paquette et al. 2007). However, because the objective of the present study was to describe the variation within PNA, we were interested in seeing whether we could detect these confiscated individuals in the mutlilocus dataset. Thus, genetic assignment tests were computed with the program Whichrun (Banks and Eichert 2000), using the 6 groups of populations as putative source populations. The ratio between the most likely allocation and the second most likely solution was used to determine statistical significance of results for each individual. A likelihood ratio > 2, which corresponds to a maximum chance of error of 0.01 (Banks and Eichert 2000), was assumed to reflect statistically significant assignments; whereas, individuals that had a ratio ≤ 2 could not be assigned unambiguously and were simply classified as “unassigned.”

Genetic Diversity and Distinctiveness of PNA Tortoises

Contrary to our expectations, the PNA population does not exhibit particularly reduced genetic diversity and does not appear to be threatened by inbreeding or the loss of genetic variability, for at least the next few generations. Similarly, it has been noted that even highly depleted sokake populations did not display the typical genetic signature of demographic bottlenecks, which is most likely a consequence of the extremely long generation time of these organisms (Rioux Paquette et al., unpubl. data). Although long generation times allow chelonian populations to maintain high levels of genetic diversity for longer periods after bottlenecks, this may mask the evidence of population size reduction and henceforth cause the necessity of conservation actions to be overlooked (Kuo and Janzen 2004). Indeed, our genetic results do not suggest that the PNA population may be close to extinction.

If confiscated individuals from other populations had been released at PNA, this could have been a source of added diversity. Assignment tests ruled out this possibility for individuals from other sampled populations, but tortoises from unsampled, remnant populations east of the Mandrare River could have been involved (which is almost certain, according to ANGAP staff). Assignments in PNA approached 100% (with one unassigned individual and one individual putatively originating from the nearby Ifotaka population), so it appears that genotypes from PNA tortoises can be identified unambiguously. Genetic differentiation of the PNA population is the highest among all populations, and the F_ST value across the Mandrare River (between PNA and Ifotaka) is the largest reported across any river in the range of the sokake (Rioux Paquette et al., unpubl. data). Thus, according to criteria usually applied for conservation unit recognition (Moritz 1994; Rioux Paquette et al. 2007), the PNA population qualifies as a distinct management unit. This is yet additional support for the conservation value of the PNA sokake population, in addition to its distinction as perhaps the last remaining population in southeast Madagascar. Even though recent sokake sightings have been confirmed from at least 4 different sites located in the Mandrare River Valley, west of the river, during field surveys conducted as part of an initiative to create a landscape-scale reserve in the area (Ferguson 2008), further work is required to determine whether these sightings correspond to remnant viable populations. Some of these observations have been made in newly established protected areas such as the Nord-Ifotaka reserve (22,000 ha, thus larger than parcel 2 of PNA), situated just across the river from the Ifotaka population that we have sampled in previous work (Fig. 1). In view of the relatively high genetic differentiation between Ifotaka and PNA, and bearing in mind that the Mandrare Valley represents one of the largest remaining areas of connected forest in southern Madagascar, that region may provide an interesting opportunity to examine the role of rivers in the establishment of genetically distinct populations of A. radiata.

Where Have All the Large Tortoises Gone?

The observed sex ratio and size structure suggest that tortoises are being harvested from PNA (Leuteritz 2002; O'Brien 2002; Seui 2006). A male-biased sex ratio could be a result of the poachers' preference for females because of the texture of their meat, which contains more fat (Pedrono et al. 2000); although, based on our experience, it seems unlikely that poachers would not collect a tortoise if it was a male. However, the vast majority of tortoises were also relatively small, and the size structure of the population is very similar to that of exploited populations (Fig. 2). Recruitment in the PNA tortoise population is probably very low, especially in light of studies suggesting that large females contribute the most to recruitment in tortoise populations (O'Brien et al. 2005).

After several days of fieldwork, we asked inhabitants of an Antandroy village about the absence of large tortoises in PNA. They led us to the answer. After a 45-minute walk in the spiny forest, we reached a small cave where we found piles of broken shells, bones, and scutes, apparently from adult radiated tortoises. Hidden next to the cave was an enclosure where our guides explained that collectors from nearby Antanosy villages house live tortoises. Several times a year, often in preparation for a holiday like Christmas or Easter, they slaughter the tortoises, mostly making dried meat, which is easily transported and concealed from police checks. The guides estimated that the piles of broken shells accounted for approximately 50 large tortoises and informed us that at least one other cave like this was located nearby. Interestingly, after seeing us inject PIT tags in captured tortoises, our guides told us that if Antanosy people became aware of these tags, they might be reluctant to eat tortoises from the park in the future.

Protected Areas and the Durban Vision

Antandroy villagers around PNA believe that the sokake should not be hunted because of their cultural taboo. When asked if they ever tried to intervene when they saw Antanosy people hunting tortoises in the park, they indicated that they did not want to cause trouble and feared being injured or even killed if they opposed poachers, who carry weapons. Discussions with former rangers revealed that corruption at the local level often hampered conservation efforts around PNA, citing as an example small-scale illegal logging occurring in PNA. They also emphasized, in light of the rising violence of altercations with poachers and loggers, that park staff lacked the resources and support necessary to assure their own security if quarrels occurred.

At the World Parks Congress in Durban, South Africa, in 2003, Madagascar's President Ravalomanana committed to tripling the country's protected area network within 5 years (a commitment accordingly dubbed the “Durban Vision”), a measure intended to preserve remaining habitats and biodiversity in Madagascar. This initiative was applauded worldwide by conservationists, and the President has so far kept his promise: over a million hectares of remaining forest cover have either already been added to the network of national parks and reserves, or are in the midst of being included (Mittermeier et al. 2006). However, it has also raised skepticism among experts on Malagasy socioeconomic, cultural, and political questions. To cite Horning (2008), “[the notion] that resources are better protected when they are placed under the formal protection of the state is, at best, questionable (p. 114).” The situation of the PNA tortoise population clearly shows that creating protected areas does not necessarily confer protection for species within them.

This is not an exclusive problem to PNA; for instance, during our first trip to Beza-Mahafaly Special Reserve in 2004 (a protected area with similar status to ANGAP parks and reserves, but managed in collaboration with the School of Agronomic Science of the University of Antananarivo), we were able to locate 18 sokake. This was a relatively impressive effort considering that only 16 tortoises had been marked by reserve rangers over the previous 2 years. Three years later, with a larger field team, we did not encounter a single tortoise after several days of intensive work. Although the human population around the reserve is mainly composed of Mahafaly and Antandroy people (Ratsirarson 2003), there are a few Antanosy settlements in the area. Reserve staff believes that the most devastating blow to the tortoise population took place prior to our first visit, between 2001 and 2003, when the reserve was temporarily closed because of a lack of funding, during which time Antanosy people massively collected tortoises. Furthermore, according to reserve employees, Mahafaly people of the region sometimes sell tortoises to Antanosy despite the taboo, and poaching has continued. Our experience shows that tortoises at Beza-Mahafaly went from being uncommon to almost extinct at best within a few years. The PNA population may be facing the same fate in the near future.

A few years ago, there were 4 protected areas within the extant range of the sokake (PNA, Beza-Mahafaly Special Reserve, Cap Sainte-Marie Special Reserve, and Tsimanampetsotsa National Park), but PNA and Beza-Mahafaly are located in suboptimal tortoise habitat. In order to increase survival probabilities of the sokake, Nussbaum and Raxworthy (2000) recommended establishing additional natural reserves on the Mahafaly and Karimbola plateaus, at the core of the sokake's range in the southwest. This recommendation will be followed as part of the Durban Vision, as Tsimanampetsotsa will be expanded from 43,200 ha to over 200,000 ha, covering most of the Mahafaly Plateau north of the Linta River. In addition, as mentioned above, new protected areas have been recently created in the Mandrare Valley as part of a larger project that would protect, if implemented in its entirety, over 300,000 ha of connected forest in that region, and there are several other planned protected areas within the range of the sokake.

Regrettably, this expanded network may not provide much protection for tortoises under current policies; the Durban Vision will not translate into efficient biodiversity conservation unless essential conditions within and around protected areas are met. These have been discussed elsewhere (e.g., Goodman and Benstead 2003) and undeniably involve the development of sustainable resource management strategies (including for the sokake) in tight partnership with local communities. There are examples among the network of national parks where benefits of the protected area may exceed those that would come from unsound resource management and failing to maintain a protected core area. For instance, the famous Ranomafana National Park in tropical eastern Madagascar now generates substantial income for local residents and the park through tourism and research activities, and there is a sense of great pride toward the conservation of biodiversity in the surrounding community (Wright and Andriamihaja 2002). The situation in Ranomafana and PNA is, of course, very different: the number of visitors in Ranomafana has reached over 10,000 annually in the late 1990s (Wright and Andriamihaja 2003); whereas in PNA, less than 2000 people have visited the park in total between 1997 and 2003 (Fenn 2003). And even Ranomafana has a history punctuated with friction and conflicts (e.g., Peters 1998).

The Absence of Financial Alternatives

There are virtually no economic activities in the extreme south of Madagascar, including very little tourism. Discussing this topic in detail is beyond the scope of this article, but it is crucial to recognize that economic development and conservation often go hand in hand. The view that Malagasy villagers are as intrinsic to the solution as they are to the environmental problems in Madagascar has greatly progressed in the past decade (Kaufmann 2006), and ensuring that local populations can meet their most basic needs would remove tremendous pressure from biological resources and habitats. As an example, the illegal fishing of marine turtles along the southwestern coast provides strong support for this position. In spite of official attempts to protect green turtles (Chelonia mydas) in that region dating back to 1923, conservation efforts have been deemed a failure (Lilette 2006). One of the main reasons for this failure is that there are no financial incentives to limit the exploitation of marine turtles, and fishermen have expressed that if there were alternatives to earn more money while protecting turtles, they would certainly do it (Lilette 2006). In northwestern Madagascar (e.g., the island of Nosy Iranja), marine turtles have been integrated into ecotourism tours, and poaching of these animals has subsequently declined. Unfortunately, as noted above, there is much less tourism in the south.

One reason for the lack of tourism in southern Madagascar is the deplorable condition of the roads; improvement of the roads would certainly favor economic development and tourism in the region. But roads are a double-edged sword with respect to sokake conservation. In addition to the direct tortoise mortality that they induce (Goodman et al. 1994), large numbers of tortoises are collected on roads (Nussbaum and Raxworthy 2000), and increased traffic would correspond to more opportunities for collecting tortoises; additional roads would also provide easier access to remote regions where sokake are still abundant and facilitate transportation of collected animals. However, one would have to be oblivious to the needs of the people of southern Madagascar not to recognize the importance of improving roads in the region. If road networks are improved, extra measures such as enhanced road checks and improved anticorruption strategies will have to be taken to minimize impacts on tortoises.

The Taboo (Fady)

Another aspect of sokake conservation is the understanding of the taboo, or fady in Malagasy, in Mahafaly and Antandroy communities. Blaming the breakdown of the fady on the increasing immigration of people from different ethnic groups, experts often claim that reinforcing the fady in local communities is important to promote tortoise conservation (Lingard et al. 2003). Traditional taboos often act as efficient de facto protection means for certain species or forests (Jones et al. 2007; Tengo et al. 2007), but it may be naive to believe that reinforcing them either with legal instruments or awareness campaigns would be effective. One has to consider past experiences of Antandroy and Mahafaly communities with the outside world, and the fact that any given rural community from southern Madagascar will have experienced several, if not all, of the following: violence and pillaging from the colonial French; racist attitudes by a central authority or administration since French colonial rule; destruction of the prickly pear cactus, or raketa, by colonial administration despite it being an essential food source in times of drought, and resulting subsequent famines (Middleton 1999); theft of livestock by bandits armed by Malagasy military (Rasamoelina 2007); oppression exerted by corrupt officials at multiple levels; repeated promises by the government and nongovernmental organizations concerning improvements to water supply problems or other projects that do not meet their targets; and researchers investigating the most private and sensitive issues in their lives. Thus, external modifications or reinforcements of taboos that have usually remained untainted by the outside world may be seen as threats. Moreover, meddling with traditional institutions may cause their cultural erosion, and the self-enforcement that they once procured may simply break down (Jones et al. 2007).

In addition, the manifestation of the fady varies among ethnic groups and even among villages within the same group. In 2004, we spent a week in a small village north of the Linta River, where a sacred forest harbors a rather large, undisturbed sokake population. The fady, in that community, was motivated by respect for the spirits of the ancients, which are believed to dwell inside tortoises that inhabit the sacred forest. On the other hand, in several Mahafaly and Antandroy communities, the fady regarding sokake is mainly imputable to loathing rather than respect. There is a widespread legend about an ancestor who tried to boil a live tortoise in a clay kettle, causing the tortoise to kick at the kettle and shatter it (Leuteritz 2002). Because of that legend, several communities believe that attempting to kill or eat tortoises leads to unfortunate events, and even see the sokake as a ‘dirty' animal. Their motivation for the taboo is thus of a very different nature than that described previously. For instance, while we were in coastal villages south of the Linta River, we quickly realized that Mahafaly peasants considered sokake as pests because of the damage they did to their crops. Touching tortoises is not prohibited by the fady there, so they did not hesitate to throw tortoises out of their fields. There is little doubt that poachers proposing to rid fields of crop-eating tortoises would be welcomed. Although there has been a tendency to blame the dramatic decline of the sokake on the loss of cultural traditions of Mahafaly and Antandroy people, in many cases this assessment reveals a misunderstanding of the fady. Although several communities respect their own fady, they simultaneously allow others the right to follow their own norms, including Antanosy and Vezo poachers from outside the region. In fact, these poachers are often pleasing local communities by removing ‘dirty sokake' from the area.

Convention on International Trade in Endangered Species (CITES) Listing

The legitimacy and benefits of listing A. radiata on Appendix I of CITES in 1975 have often been questioned (Lewis 1995; Nussbaum and Raxworthy 2000). At first glance, considering the current state of knowledge on the status and dramatic decline of the sokake (Randriamahazo et al. 2007), this listing only appears to be appropriate. However, one has to bear in mind that legislation should favor the sustainable use of resources and ensure they are valued at fair rates (Allen and Edwards 1995). In the case of A. radiata, it has arguably lead to completely opposite results: in the absence of effective law enforcement, total trade prohibition stimulated the establishment of an illicit, unmanageable, and inequitable market. First, the discrepancy between the amount paid to collectors by intermediaries and the final price charged by exporters is staggering; it is estimated that collectors receive 0.08% of the final price of a tortoise (Bidaud and Randria 2008). By comparison, chameleon collectors received 6.5% when export controls were liberalized in the early 1990s (Carpenter et al. 2005). Furthermore, no compensation at all is provided to Mahafaly and Antandroy communities. Finally, in Western countries, it is possible to legally acquire captive-bred radiated tortoises (US$1500 for a juvenile, $4500 for an adult), but this trade does not provide any income to Madagascar (Bidaud and Randria 2008). This apparently contradicts clause 15 of the Convention on Biological Diversity, which advocates sharing profits related to the exploitation of genetic resources. Thus, arguments for downgrading A. radiata to CITES Appendix II deserve serious consideration; although, in order to be beneficial, it would need to be accompanied by careful exportation controls, which have been deficient for other legally traded Malagasy species in the past (e.g., Pyxis arachnoides between 1980 and 2004; Walker et al. 2004). Finally, it is important to emphasize that even though the international smuggling of sokake is worrisome (a recent survey of freshwater turtle and tortoise pet markets in Bangkok revealed that A. radiata was the most common species sold there; Shepherd and Nijman 2008), its impact on tortoise populations is by far exceeded by that of poaching for local consumption of sokake meat.

Captive Breeding?

Successful conservation schemes for the radiated tortoise will inevitably have to incorporate financial incentives. Captive breeding centers, managed by local communities, could be a solution. Through regulated trade, income generated this way could be shared among communities. The idea of commercial breeding centers garnered great interest among Malagasy parties at a recent International Union for Conservation of Nature/Species Survival Commission Tortoise and Freshwater Turtle Specialist Group meeting in Antananarivo, Madagascar (Mittermeier et al. 2008); whereas, it had been dismissed when proposed in the past, as early reports of the ongoing overexploitation crisis were emerging (Lewis 1995). However, commercial breeding can only be successfully implemented in combination with genuine efforts to ensure that authorities charged with enforcement are themselves enforced. Over the years, we have heard too many stories and anecdotes indicating that law enforcement personnel are often at the forefront of the flagrant disregard for the laws concerned with the poaching and smuggling of radiated tortoises. Unless this situation can be improved, commercial farming would only legitimize the current illegal operations. In addition, measures to ensure the conservation of wild tortoise populations will need to be established in parallel. Part of the income generated through farming could serve as funding for monitoring programs. Rewarding communities that maintain abundant wild tortoise populations has been proposed, but payment or reward systems are difficult to enforce and may be impossible to implement considering the complex property rights system of southern Madagascar. Thus, several aspects would have to be thoroughly pondered, and local communities themselves would probably be instrumental in the formulation of guidelines for projects such as this. Fortunately, there may still be a few million sokake remaining in the wild (realistic estimates range between 1.6 and 4 million; Randriamahazo et al. 2007), so it may not be too late to launch audacious and innovative projects.