2,274Grants to


Radiated tortoise (Astrochelys radiata)

Mohamed bin Zayed Species project number 11252584

Mohamed bin Zayed Species Conservation (Project No. 11252584) - Radiated tortoise - Awarded $5,000 on August 01, 2011


Project Kobokara 2011 was established in December 2010. This University of Edinburgh approved project will take place during the summer months of 2011 (June-September) in the village of Kobokara, within Ifotaka North New Protected Area, southern Madagascar (24°45'22.16”S 46°03'25.61”E). We conducted a series of surveys on the Critically Endangered radiated tortoise, Astrochelys radiata, while the rest of the team studied other elements of the local biodiversity and human community. The project, as a whole, is intent on collecting valuable data on the wildlife and culture in this unique ecosystem and contributing to its future integrity. A. radiata is one of four endemic tortoise species in Madagascar. In the past they were abundant and an iconic feature of southern Madagascar. However, they are currently listed as Critically Endangered by the IUCN Red List, and protected under CITES Appendix I. Different approaches to estimate population sizes all conclude that the species has dramatically declined in numbers in recent years. Our research was an observational field-based study, consisting of a systematic survey to assess the status of A. radiata within an area of Usage Rights Forest, within which selective deforestation and livestock grazing is still permitted, thus shedding light on the effect of anthropogenic disturbance on these tortoises. Information relating to the current population status of A.radiata in Kobokara can be used to apply effective conservation measures and predict possible detrimental effects linked to reduced population sizes like fitness reduction due to inbreeding, leading to mutational meltdowns, which could accelerate extinction times. The team was composed of four UK, three Malagasy students, a translator and a local guide. The Malagasy students are from the Centre Ecologique Libanona (CEL), founded in 1995, which now runs a 3 year degree course (Licence Professionelle) in Environmental Management for Development. A secondary objective was to establish protocols for this form of research for future ecological monitoring in Kobokara. To this effect, we trained our guide and students in reptile handling and measurement, to improve their capacity to carry out similar work in future. The data we gathered on this expedition will be given to WWF Madagascar, other conservation agencies, and used in the community-based management of the forest, to inform environmental impact assessments and future ecological monitoring. The completed final report will be disseminated as widely as possible and publications in leading journals will be earnestly investigated. The report will be stored in CEL's library database and will be accessible to other institutions and environmental NGOs working in Madagascar. Four short documentary films are being produced, and will be uploaded to the project website, www.projectkobokara.org.

 The full project report is available from www.projectkobokara.com.



The radiated tortoise, Astrochelys radiata, formerly Geochelone radiata, is one of four chelonian species endemic to the island of Madagascar (Pedrono & Smith, 2003). First described in 1802 in southern Madagascar, A. radiata are predominately found in areas of dry spiny forest that are dominated by Didiereaceae and Euphorbia (Durrell et al., 1989; Pedrono & Smith, 2003), but they have been reported to be present on high inland plateaus and costal sandy dunes (Leuteritz et al., 2005).
Tortoises are susceptible to disturbance, although this threat is secondary to the risk presented by exploitation (Irwin et al., 2010; Leuteritz et al., 2005). It is known that hunting and the trade in bushmeat and pets is a major source of the decline in numbers of A. radiata (Lewis, 1995; Behler, 2002; O'Brien et al., 2003). Additionally, A.radiata is under threat from the changing lifestyles of a growing human population, which has more than doubled in the past 36 years (CIA, 2012). In order to access further fertile lands, people living in rural southern Madagascar have increasingly been practicing hatsake (van Vliet et al., 2012), which causes wholesale destruction of tortoise habitat.
According to the International Union for Conservation of Nature (IUCN), available information indicates that this species has disappeared entirely from approximately 40% of its past range, from a combination of habitat loss and exploitation (Leuteritz & Rioux Paquette, 2008). Remaining populations continue to be exploited, predominantly for domestic consumption (O'Brien, 2002). An overall population reduction of 80% over two past and one future generation is a conservative estimate, thus qualifying the Radiated Tortoise as Critically Endangered under criterion A4d (Leuteritz & Rioux Paquette, 2008; IUCN, 2001). Population modelling indicates collapse and extinction in a period of on average 45 years into the future, thus meeting Critically Endangered under criterion E (Leuteritz & Rioux Paquette, 2008; IUCN, 2001). Habitat loss rates approach or exceed 80% over the three-generation period, so criterion A4c may also be met (Leuteritz & Rioux Paquette, 2008; IUCN, 2001).
The status of A. radiata on the IUCN Red list was last assessed in 2008, resulting in its status being upgraded from Vulnerable which it had held since 1982, probably due to a severe lack of studies in the wild until the late 1990s (Rioux Paquette et al., 2009). A. radiata is also listed in Appendix I of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) since 1975. This makes it illegal to conduct any commercial trade with wild specimens, and very difficult to trade in captive-bred individuals (CITES, 2011).
Although the total population of A.radiata remaining in Madagascar has been estimated at 4.5 million individuals, which is extremely large for a species classified as critically endangered, it has been reported that between 47,000-60,000 tortoises are being illegally harvested for sale in domestic and international markets (Randriamahazo et al., 2007; O'Brien et al., 2003; O'Brien, 2002). Despite IUCN Critically Endangered status, and being placed on Appendix 1 of CITES, there is still a large illegal market threatening the existence of A. radiata. Their international collection has been documented with Asian smugglers collecting tortoises for the pet trade and for their livers (Behler, 2002). Aside from use as food, the Malagasy people often keep the tortoises as pets and in pens with chickens and ducks as a means of warding off poultry diseases (Durrell et al., 1989; Leuteritz et al., 2005).
The establishment of the Ifotaka-North Protected Area is an important step in the conservation of the radiated tortoise at the north-eastern extent of its range. However, its status within the forests of the Mandrare River Valley is largely unknown (Ferguson, 2011a). We investigated the status of A. radiata within a 5.19km2 area of CDU forest within the fokontany of Kobokara. We found a total of 35 tortoises in our short study, and conclude that the population within the forest is very small, likely comprising only approximately 55 individuals. We highlight the need for better understanding of the basic ecology of A. radiata, and make recommendations for future research, monitoring and conservation efforts in the area.


We generated a map using high-resolution satellite imagery available online from the geological surveys conducted of the region by mining companies. The map was divided into a grid of 100m x 100m cells, which were used to randomly select start points for walking
transects. Transects were walked by groups of two or more people, separated by approximately 10m. Each transect was walked at a slow pace,
taking time to look 5m to either side of each person's transect line, paying special attention to known hiding places of the A. radiata, in bushes, under fallen trees, and at the base of large trees.
GPS coordinates were recorded every 50m so that cells could be identified for analytical purposes. Upon discovery of an individual, the time, date, GPS coordinates, and altitude were noted. Each individual was examined and the sex (using the Anal Fork Ratio [fork width:fork depth], following O'Brien, 2002), age (estimated from the number of rings on the 2nd central scute), mass, surrounding habitat and 39 morphometric measurements were recorded as shown in Figure 13.4 of the full report (Rioux Paquette & Lapointe, 2007). Tortoises smaller than 270mm in length were classified as juveniles (O'Brien, 2002).
After recording all required measurements for an individual, each was marked using a unique code created by filing notches in a combination of scutes on the left and right side (Cagle, 1939). Marked tortoises could then be individually recognised. In the event of recapture, the GPS coordinates, time and date would be noted, and this information would be used for Capture-Mark- Recapture (CMR) analysis.
Collected GPS data for captured individuals were used to estimate the populations of A. radiata in the forest to the east of Kobokara. By calculating the mean number of tortoise found per 100m x 100m grid cell searched, and then extrapolating to include the total area of the whole forest (5.19 km2), we were able to attain an estimate of the number of tortoise which can be found in the forest. In order to generate an accurate estimate however, it was necessary to accommodate the detection rate of the tortoise (Durso et al., 2011).
For CMR analysis, we can use the Schnabel equation (Schnabel, 1938; Equation 1) to calculate the population density for the area searched, which can be taken as being the density of A. radiata found in the forest to the east of Kobokara.
N = (((M + 1)(C + 1))/(R+1))-1

Equation 1 - The Schnabel Equation
Where N = number of individuals; M = number of individuals marked on first survey; C = number of individuals captured on second survey; and R = number of recaptures in C.
A selection of 12 people was interviewed from each of 3 villages in the area including Kobokara, across a diverse selection of the population, both men and women of different ages and professions, to find local opinions on the radiated tortoise.

Results & Analysis

Population Estimate
To estimate the population size, the total area of the forest to the east of Kobokara was calculated to be 5.19km2. Transects searches covered an area of 3.29km2, finding a total of 35 living individuals, and 4 deceased. If we assume a 100% detection rate and equal distribution, we may derive a population density of 10.638 tortoises per km2, and therefore estimate a total population of approximately 55 individual tortoises in the 5.19km2 study forest. However, it is highly unlikely that a 100% detection rate was achieved, so this number likely represents an underestimate. It is noteworthy that we found no individuals smaller than 153 mm in total length. This may indicate a detection bias towards larger tortoises, and may therefore contribute yet further to an underestimate of the tortoise population.
We plotted an accumulation curve to find the completion of our survey. The accumulation curve clearly indicates that we continued to find many new individuals, even towards the end of the study period. As the majority of the population are recorded, fewer and fewer new individuals will be discovered despite an increasing time of study. This results in the curve plateauing once nearly all individuals have been found.
We recorded a total of 11 male and 6 female adult tortoises, and 14 juvenile tortoises, which cannot be sexed reliably (O'Brien, 2002). This ratio may be a result of sampling bias if there is a behavioural dimorphism between the sexes (Rioux Paquette et al., 2010), or an actual demographic difference. Although we cannot rule out the first of these difficulties, we consider the lattermost to be most likely, as our findings are corroborated by those of Rioux Paquette et al. (2009), who recorded a 2:1 male to female ratio.

Population Density
Unfortunately the length of the study was insufficient to complete a CMR program; only one individual was recorded for recapture, even though a further two were sighted. Miscommunication with other research groups on the project resulted in recapture locations not being recorded, and therefore in the inability to use the data for the study. Population density was therefore extracted by extrapolation, as explained above.

Morphometric Measurements
A table detailing all morphometric measurements collected for individuals can be found in Appendix II of the full report. Despite collecting morphometric measurements for 33 of our 35 individuals, we were unable to identify a genetic population for the area based on Rioux Paquette & Lapointe's work (Rioux Paquette & Lapointe, 2007). The data was processed using a complete algorithm provided alongside their paper, but our results prove to be insignificant. Indeed, even our own results were not consistent with each other, and it is impossible to draw any conclusions from the data. This may become possible by using a larger sample group.


Devaux (2010) summarised the available data on the historical distribution of these tortoises, highlighting the decrease in their range over time. Importantly, his summary suggests that the radiated tortoise's eastern range does not stretch as far north as Ifotaka. Additionally, Devaux conducted a study in 2009 that found radiated tortoises to the west of Amboasary to be at low densities, not exceeding 10 individuals per hectare. The furthest east population above 10 individuals per hectare was reported to be 25 km to the south-west of Tsiombe, approximately 110 km to the south-west of Kobokara (Devaux, 2010, p.69).
Our study revealed a population of very low density, with just 0.139 tortoises per hectare. Although this is not a good density, it nevertheless proves the existence of these tortoises, albeit at low concentration, significantly north of their assessed range. If they are found at similarly low densities over a considerable area, their overall population size may be significantly underestimated.

Although our predictions were that the high levels of disturbance found in the area would result in a low number of tortoises being found, we were surprised to discover that this was not the case. The capture of individuals was spread throughout completely undisturbed, semi-disturbed, and disturbed forest. However it should be noted that fewest tortoises were found in the undisturbed forest.
This may indicate that the tortoises are simply easier to detect in the semi- and disturbed forest, despite living predominantly in undisturbed forest. Indeed, we spent far less time in undisturbed (categories 1 and 2, see Section 15 of the full report) than more disturbed forest, and sampling bias must therefore be considered to be prominent. Additionally, tortoises were very difficult to detect. This means that we need to find an accurate estimate of our detection rate with which to modify the initially calculated results, to account for sampling bias and detection probability. It is difficult to generalise a grid over the whole area of study near Kobokara, when many different types of terrain were present. One would naturally expect some terrain to be a more suitable habitat for tortoises than others. Some indication of this is shown in Figure 13.7 of the full report, which shows high-density areas of tortoises in red and areas of moderate density in orange. It is noteworthy to compare these density polygons with the disturbance distribution shown in Figure 15.1b of the full report. Those areas of highest density do in fact occur in the least disturbed areas of forest.
The observed 1.83:1 male to female sex ratio is concerning. If this ratio is not a result of sampling bias, it may suggest a limited recovery potential for the population at large. It is conceivable that the reason for this ratio may be a result of tortoise harvesting, for which the largest tortoises (females) are typically selected (Rioux Paquette et al., 2009; Leuteritz et al., 2005). Regardless of the cause, this observation is worthy of further investigation, as the pattern of male population dominance in A. radiata has, as yet, not been explained (Rioux Paquette et al., 2009). To address this curious observation, we advocate studies conducted in the middle or at the end of the wet season (January-March); during this period, the eggs of A. radiata hatch (Leuteritz & Ravolanaivo, 2005). Taking genetic samples from a significant hatchling sample, and thereby sexing them, may reveal whether the sex ratio is natural, or a result of negative selection on females, of natural or human cause.
Conducting wet season studies on hatchling demographics will determine whether this ratio is present in post-natal tortoises, but the underlying cause would still be unclear; if these tortoises, like almost all other chelonians (Bull, 1980), utilise environmental sex determination (e.g. Valenzuela & Janzen, 2001; Janzen & Paukstis, 1991; Spotila et al., 1994), the bias could be a direct outcome of environmental changes. Thus, the effect of environmental stimuli on sex ratio in A. radiata is in need of study, before results of hatchling studies may be thoroughly interpreted.
In addition to using transect searches to learn about the population of A. radiata in the Kobokara region, interviews were conducted with a diverse selection of the local communities. It is apparent from these interviews that the popular view is that the local population of radiated tortoises is on the decline. Despite the species having a protected status for many decades, they continue to be hunted to be eaten, their carapace used for jewellery, and more recently exported for the foreign pet trade, with a growing interest from East Asia. This species is in drastic need of some intervention to prevent its extinction; it took 3 weeks of transect searches in the field to find more individuals in the wild than had been seen in kitchens of various restaurants since our arrival in Madagascar. However, as with any nation, developing or more developed, people will only change their way of life if a better alternative is presented to them; it is not practical or even feasible to imprison tens or hundreds of thousands of people who do not respect laws imposed on them by their own or other governments, restricting practices that were previously acceptable and widely practiced. Education is needed, to teach the people why the tortoises are in need of conservation, and how they can help. The future of A. radiata in this area is uncertain. Although the tortoises are protected from local consumption by fady (taboo), which dictates that they may not be eaten, they are still at threat from a variety of sources. The fady does not prevent the tortoises from being touched, so they are often hurled out of fields when they are found, resulting in shell fractures and ultimately death in many individuals (M. Scherz, pers. obs.). Furthermore, this fady is restricted to the Antandroy people, and the tortoises face an increasing threat from migrating individuals from neighbouring ethnic groups who consider tortoise flesh to be a delicacy (O'Brien, 2002).
The imposition of IUCN and CITES listings have only limited bearing in rural Madagascar. This is unsurprising, as the distance from power-bearing authority is significant. We were surprised, however, to learn that individuals (who shall remain anonymous) with ties to the WWF were fond of tortoise flesh, and oblivious to the illegality of its consumption. If rural people see the restrictions on consumption as a power play, restricting these delicacies to the lawmakers, the protection of these critically endangered tortoises will remain ineffective.
The status of the radiated tortoise in Kobokara is somewhat worrying. The Mandrare Valley lies at the north-eastern extent of its range (Rioux Paquette et al., 2010). At peak density, these tortoises are reported to exceed 1500 individuals per km2 (O'Brien, 2002). A mere 10.6 per km2 is, therefore, meagre. This area is under considerable anthropogenic pressure, and the CDU forest in which this research was conducted is rapidly declining in quality (see full report, section 15). Management schemes like those suggested in the Section 15.5 of our full report would doubtless be beneficial to the status of A. radiata in these forests. However, this area of forest represents only a very small portion of the Ifotaka-North Protected Area. Further studies on its status, particularly to the north of the Ikonda River in the ZdC, would be extremely beneficial in order to generate better informed management and conservation schemes.


Changes to Methodology
After our two-week pilot study yielded very few results, it was decided that in addition to scientific research into the population of the A. radiata in the region, a social study would also be adopted. This informal method of assessing the trend of tortoise populations was beneficial, and should be adopted by a larger number of studies, as it is unlikely that foreign teams can become as familiar with tortoise populations over a short study period as local people become over their whole lives, despite their obvious fondness for skewing reality.
Initial plans for the tortoise study suggested working as a team with the herpetology research team, so that we could assist in laying pit traps and we could have an increased manpower for tortoise surveys. However it soon became apparent that we needed to cover very large areas of the forest for our surveys and this was impractical given the amount of time required to be around the pitfall traps. This loss of manpower caused a significant decrease in our detection rate, which was unfortunately not consistent throughout the course of the study. Each day when a large number of tortoise were found, e.g. day 11 and day 16 when more than 5 new individuals were found, we had a work force of at least 4 people.


Recommendations for future studies
The lack of available manpower had a detrimental effect on this project, as such a small area was searched, and the area that was searched had a low rate of detection. Although in theory a 10m wide transect seemed very achievable, in practice it was far too much, especially in areas of little disturbance, but also in semi-disturbed areas. This could be solved in the future by employing local people to help search transects.
As previously mentioned, we advocate further studies, particularly to the north of the Ikonda River in the ZdC of the Kobokara section of the Ifotaka-North Protected Area. This large area of forest has not yet been subject to any tortoise surveys, but several individuals informed us that it was a good area in which to search for tortoises. This forest is evidently also under increasing anthropogenic pressure, judging by circles of forest clearance that have appeared in the last five years on Google Earth®. However, its more protected status should mean that it might remain a refuge for these tortoises, even after the almost inevitable disappearance of the CDU forest, which was the subject of the present research.
Finally, the sex ratio conundrum remains unaddressed, despite the observations of Rioux Paquette et al. (2009). It is of particular interest however, and demands further study. It is surprising that this topic has, as yet, gone unstudied in a critically endangered species, for which several breeding initiatives, (e.g. Behler & Iaderosa, 1991), are in place. These studies would shed light on an important topic in the conservation biology of A. radiata. 


Project 11252584 location - Madagascar, Africa