Animals that live in dry environments, or in regions with intense dry seasons, are often challenged with obtaining enough water to survive. However, animals have evolved morphological, physiological, and behavioral adaptations to meet that challenge (Ward 2009). For example, some desert lizards channel water droplets from their heads into their mouths using complex morphological structures (Sherbrooke 1990). Likewise, a number of tortoises assume postures that enable raindrops to collect at the front of their shells for drinking (Auffenberg 1963). These types of natural history observations are often made serendipitously while conducting fieldwork for other purposes. However, chance discoveries like these can be important precursors to quantitative studies designed to determine the value and complexity of such adaptations.

While conducting a monitoring program for radiated tortoises (Astrochelys radiata) in arid southwest Madagascar, we observed a large aggregation of individuals that appeared to have gathered in anticipation of rainfall. Our observations were made on 5 November 2009 along a dry creek bed approximately 10 km from the village of Itampolo. The survey location was an area of protected spiny forest that was dominated by Alluaudea comosa, Comiphora spp., Euphorbia spp., and Gatropha mahafaliensis (Leuteritz 2002). The weather conditions included early morning drizzle followed by moderate rain in the afternoon. The air temperature was 22° to 24°C and a light to moderate breeze was present. Seven surveyors spaced approximately 10 m apart walked along a 150-m transect through the forest. The survey and associated data collection were conducted between 0730 and 1130 hours. At 0750 hours we discovered the first tortoises: 10 individuals within a 20-m stretch of the dry creek bed. Given the positioning of these tortoises along the creek bottom, the number of individuals present, and the prevailing weather conditions, we hypothesized that they were awaiting heavier rainfall to drink. The tortoises were not feeding, and many were resting in areas with no vegetation, compared with an abundance of understory in adjacent areas.

Over the next 3 hours we observed, measured, and marked a total of 99 tortoises along an 80-m stretch of the dry creek bed. Most tortoises were alert and fully exposed; however, there were several individuals that were stationary beneath vegetation within 3 m of the creek. The width of the creek bed ranged from 4 to 20 m wide, was < 1 m deep, and was mostly gently sloping. Some tortoises were less than 1 m from each other and 22 individuals were recorded in a single cluster. The amount of rainfall increased at 0900 hours, and the tortoises immediately began to drink. They imbibed water through their nostrils from shallow pools in crevices on flat rocks, a behavior that had been previously described by Leuteritz (2003) for this species.

The density of tortoises at the dry creek site (1031.2 tortoises/ha) was considerably higher than that estimated from a “creek-less” transect located 400 m to the east (6.8 tortoises/ha). In addition, one individual captured at the second transect had been marked the previous day at the dry creek site, which indicates that some tortoises traveled to the creek from considerable distances.

The wet season in southwest Madagascar follows a long dry season that spans nearly 7 months (April to October). Thus, the tortoises that we observed had little or no access to freestanding water in the weeks prior to when our observations were made. However, since tortoises were able to drink after a relatively small amount of rainfall on this occasion, they may have had the opportunity to drink after similar intermittent rains during the dry season. Desert tortoises (Gopherus agassizii) drink free-standing water after infrequent rainfall events (Medica et al. 1980). This behavior often results in weight gain, an increase in foraging, and expulsion of stored urinary wastes, and several authors have contended that this drinking behavior may be critical to their survival during dry periods (Medica et al. 1980; Nagy and Medica 1986; Peterson 1996).

Our observations suggest that radiated tortoises may occur at high densities at sites that provide drinking opportunities and that individuals possess knowledge of these sites and travel to them. These observations are consistent with those reported for desert tortoises that travel directly to drinking sites before and during rainfall and congregate at river catchments (Medica et al. 1980; Peterson 1996). Moreover, it has been suggested that desert tortoises have the ability to sense oncoming rain and have a familiarity of important sites within their home ranges (Medica et al. 1980). Aggregation behavior exhibited by the radiated tortoise at discrete sites within this landscape may increase its survival during the lengthy dry season. However, it may also make this species more susceptible to human harvest for food or trade because it facilitates easy collection. These factors should be considered when designating protected areas for this species' conservation.