Prior to recent studies by Luiselli (2003a) and Luiselli et al. (2000), little information on the biology of Central Africa's two forest-dwelling hingeback tortoises, Kinixys erosa and K. homeana, has been available beyond anecdotal reports appearing in regional accounts (e.g., Schmidt 1919; Laurent 1956; Lawson 1993) and taxonomic and morphological treatments (Loveridge and Williams 1957; Lawson 2001). Summarizing information available at the time, Pritchard (1979) commented that Kinixys erosa and K. homeana “are quite similar in many respects and occupy much of the same range in West and Central Africa. This apparent defiance of the normal ecological rules … is still without explanation and will remain so until the natural history of these poorly-known species has been properly studied.”

On an immediate and practical level, people heavily exploit Kinixys for pets (Hoover 1998) and food (Schmidt 1919; Lawson 2000; Luiselli 2003b; Luiselli et al. 2003). To understand the impact of this exploitation and to begin to address the future of these animals requires a better understanding of their ecology and natural history. To gain a better understanding of these readily recognized, but reclusive and poorly known species, I present information on Kinixys erosa and K. homeana habitat use, home range size, and seasonal behavior based on radiotelemetry and field observations from an area of southwestern Cameroon where the two species are sympatric.

METHODS

I studied tortoises from April 1995 to November 1997 in an approximately 1 km² area just inside the border of the Banyang-Mbo Wildlife Sanctuary (vicinity 5°21′74′′N, 9°30′39′′E), Southwest Province, Cameroon, Africa. Southwestern Cameroon has a pseudo-equatorial climate consisting of a dry season from mid-November through late March, and an extended single rainy season. Monthly mean temperatures range from 21° to 36°C. Annual rainfall ranges from 3.5 to 4.3 m (D.P. Lawson, unpubl. data, 1995–1997). Vegetation of Banyang-Mbo consists of lowland evergreen tropical forest characterized as Atlantic Biafran forest (Letouzéy 1985). Because both tortoises are consumed locally, the study site was intentionally remote from human concentrations.

Descriptive statistics of study specimens and tracking dates are given in Table 1. Study specimens consisted of 3 tortoises (1 Kinixys erosa, 2 K. homeana) obtained from local collectors and relocated to the study area (translocated specimens), and 6 resident tortoises (3 of each species) found within the study site while mapping or following other tortoises. Original localities of the translocated specimens is not known. However, given human activity patterns in the area, it is likely these tortoises originated within 10 km of the study area. I found two resident K. homeana males in June and July 1995, respectively, while they were actively engaged in combat with the translocated male K. homeana. I found one resident K. homeana female and a resident K. erosa female feeding together on the same patch of mushrooms in August.

Table 1. Summary statistics of telemetered Kinixys.^a

View Fullscreen

I attached radiotransmitters (model SB2, AVM Instrument Company, Ltd, Livermore, CA) to the posterior marginal carapace scutes using dental acrylic. Transmitter antennae were passed anteriorly through a hollow plastic tube affixed along the lateral and anterior marginals. To allow unhindered movement of the carapace hinge, the antenna tube was cut, and a gap was left spanning the carapace hinge. Weight of the telemetry package including adhesive ranged from 6% to 13% of body mass (Table 1). The specimen with the heaviest transmitter package relative to body mass also had the largest home range of any tortoise and did not lose weight over the study, supporting the assumption that telemetry units did not negatively impact movement or behavior.

With a receiver (Custom Electronics CE-12 or Telonics TR-2) and hand-held H antenna, I located specimens in the field as often as possible in 1995, approximately twice per month in 1996, and approximately once a month in 1997. Weather, unsafe river crossings, technical difficulties, and illness affected the frequency and specific dates of tortoise tracking. Individual tortoises were followed until transmitter failure as a result of malfunction, battery life, or accident. Each time a tortoise was located, the following information was recorded: current and prior weather conditions; air temperature one meter above the tortoise, at ground, and at ground and 1 m in closest shade; time; descriptions of the canopy, understory, forest floor, nearby habitat features (e.g., tree fall gaps, streams, etc.), microhabitat at the tortoise's location; and comments on the animal's activity, other wildlife or human presence or sign. I used Mann-Whitney U tests to compare frequencies of habitat and microhabitat use between species (Sokal and Rohlf 1981). Once a month, provided the specimen was accessible, I weighed tortoises in the field using hand-held spring scales, and measured carapace height and width at the hinge, and plastron length using calipers.

I surveyed tortoise locations using a compass and hip-chain to measure direction and distance to known points previously geo-referenced with a Garmin hand-held global positioning system (GPS). I mapped locations in Universal Transverse Mercator (UTM) coordinates and converted them to latitude and longitude for home range analysis using the Computer Aided Mapping and Resource Inventory System program (CAMRIS ver. 3.46, Ecological Consulting, 1997).

I calculated minimum convex polygons and unweighted harmonic surface area home range estimates on seasonal, yearly, and total locations for each tortoise using CAMRIS (ver. 3.46, Ecological Consulting, 1997). I calculated harmonic surface areas at probabilities of 0.9, 0.75, and 0.5 (i.e., areas including 90%, 75%, and 50% of individual tortoise locations) on both total and seasonal locations. Harmonic surface areas are influenced not only by the number of separate locations in a given area, but also by the number of times an individual was located at a single location. Therefore, areas containing a number of retreats and/or retreats that were utilized for multiple days influence the resulting probability surface area. One translocated adult male Kinixys erosa established a well-defined home range some distance from the initial release point. For this reason, the release points of all translocated specimens were not included in home range analyses.

RESULTS

Morphometrics

Specimens of both species exhibited seasonal trends of weight gain and loss. Tortoise weights tended to increase at the beginning and end of the rainy season (October–November, March–April) followed by gradual decreases throughout rainy and dry seasons (Fig. 1). Following a dramatic early rainy season weight increase in April 1996, adult female K. erosa 59 had 2 evident weight decreases during the rainy season (April–June and July–August). One or both of these decreases may indicate egg deposition. Male K. homeana 15 showed a gradual increase in weight over 2.5 years.

View Figure

• Download as Powerpoint
• Download Figure

Morphometric measurements did not change perceptibly in any specimen over the course of the study. This is not surprising given that all animals were considered to be adults (Lawson 2001) at the onset of the study.

Home Range

Seasonal, yearly, and total minimum convex polygon and harmonic surface area home range estimates are summarized in Table 2. Kinixys erosa home ranges varied from 5.5 ha to 45.37 ha. The 2 specimens with small total home ranges had larger dry season home range estimates than rainy season estimates. Resident male K. erosa 21 had considerably larger rainy season home range estimates than dry season estimates over both tracking years.

Table 2. Minimum convex polygon and harmonic surface area seasonal and total home range size estimates (ha) for 8 telemetered Kinixys in southwestern Cameroon, 1995–1997.^a

View Fullscreen

Kinixys homeana home range estimates varied from 3.1 ha to 47.8 ha. Rainy season home range estimates were consistently larger than dry season home range estimates for the three specimens tracked over multiple seasons. For both species, the 50% harmonic surface home range areas show a consistent concentration of locations within a relatively small area (i.e., 5%–25%) of each total home range, indicating that Kinixys individuals concentrated their activity in a core area within the home range. With the exception of resident K. homeana males 14 and 15 in 1995, seasonal and total core areas defined by the harmonic surface 0.5 isopleths were small (1.23–1.39 ha) and highly consistent between species and among individuals (Table 2). However, the 1996 seasonal home range estimate for K. homeana 15 may have been influenced by reduced tracking frequency.

Resident males K. erosa 21 and K. homeana 14 had the largest total home range estimates. Some of the area of these total estimates is attributable to shifting seasonal activity centers or core areas. No overlap was found between rainy and dry season 0.5 and 0.75 harmonic surface areas for either individual. There was substantial overlap of subsequent rainy season isopleths for both individuals and dry season isopleths for K. erosa 21, indicating that these tortoises spent their rainy seasons in one discrete area and (with the exception of dry season core areas for K. homeana 14) their dry seasons in another. Resident male K. homeana 15 had the third largest total home range estimate, but with the exception of an ancillary dry season activity center in year one (1995–1996), exhibited considerable overlap of all other seasonal core areas. Individuals of both species with small overall home range estimates showed no obvious trends toward shifting seasonal activity centers.

Total home ranges for all specimens showed considerable overlap both between and within species and sex (Fig. 2). However, plotting only the 0.5 harmonic surface area of each tortoise reveals little overlap of actual core areas (Fig. 3). Resident female K. erosa 59 shared 0.17 ha of her core area with resident male K. homeana 14. Male K. erosa 21 and male K. homeana 14 appear to have considerable overlap of core areas calculated from total locations. However, when seasonal core areas are separated by year, K. erosa 21 shared only 0.08 ha of its 0.75 harmonic surface area with K. homeana 14. Intraspecific core area overlap was seen between resident female K. homeana 18 and resident male K. homeana 15 (0.83 ha), and translocated male K. homeana 6 (0.01 ha). No K. erosa core areas overlapped, and K. homeana core areas showed no intrasexual overlap (Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

For both species, total home range size varied widely among individuals, but at least half of each tortoise's locations occurred within a relatively small core area of 1.23–4.9 ha. Males of both species had larger home ranges than females, and residents for which multiple year and season home range estimates are available (n = 3) showed a decrease in home range size estimates over time (Table 2).

Habitat Use and Activity

General habitat at tortoise locations and observations of active individuals are summarized in Table 3. Considerable overlap occurred between species in habitats used, and observed differences were not statistically significant. Kinixys erosa was found more frequently in closed canopy forest while K. homeana utilized forest gaps relatively more often. Both species were found in proximity to streams and standing water with nearly equal frequency, although considerable individual variation occurred in apparent preference for proximity to aquatic habitats. Tortoises were associated with streams in both rainy and dry seasons. Kinixys erosa 21 and K. erosa 59 did not use forest gaps as frequently as other K. erosa or K. homeana. However, both specimens spent considerable time in or near streams and may have benefited from streamside reduced canopy cover or increased ground cover analogous to a gap. Kinixys homeana 15 showed the greatest site fidelity, using the same treefall gap on at least five separate occasions in mid-rainy season and early dry season (July 1995, June 1996, July 1996, August 1996, November 1996).

Table 3. Summary statistics of telemetered Kinixys habitat use and activity.^a

View Fullscreen

With the exception of translocated male K. homeana 6, tortoises were encountered while active on only 7%–8% of occasions (Table 3), and were found active throughout the day from 0800 hours to 1510 hours. No observations were made at night. Tortoises changed locations in every month indicating at least some activity throughout the year. Kinixys erosa 7 remained inside a hollow log in forest for at least 44 days from 11 December 1995 through 23 January 1996. Similarly, K. homeana 14 remained buried in leaf litter along a tree buttress for at least 24 days from 12 December 1995 through 4 January 1996. Female K. homeana 18 was observed spending 3–8 days at individual sites on 7 occasions during 3.5 dry season months, compared to a single 3-day individual site stay during 3 rainy season months of tracking. Prolonged stays at individual sites support the assumption that repeated observer presence did not meaningfully alter tortoise movement.

Microhabitat at each of 405 tortoise locations is summarized in Table 4. Microhabitat categories are considered mutually exclusive (i.e., a tortoise may have been associated with leaf litter and under a fallen log, but only the log is scored). The seven categories tabulated in Table 4 accounted for 83%–85% of all locations. “Other” categories included no cover for active specimens, seldom-used microhabitats (e.g., buried in dirt), or generalized cover that could not be broken apart (e.g., under logs, vines, and leaf litter near buttressed tree). Kinixys homeana was found more often in leaf litter than K. erosa (U =  15.0, p = 0.017). Half of all K. homeana locations were either in leaf litter with little or no other cover item or under tightly tangled vines (often in overgrown forest gaps). Kinixys erosa was found in association with aerial roots more often than K. homeana (U = 15.0, p =  0.024). Differences in stream use approached significance (U =  12.5, p = 0.051). Streams were used by K. erosa in both rainy and dry seasons. However, streams were often dry outside of the rainy season, and stream use at this or any time may have been incidental to cover choice. Both species were most often found under fallen logs and branches (Table 4). Although not statistically significant, K. erosa showed a greater propensity to shelter under larger logs (> 50 cm diameter) than did K. homeana, and was found in association with tree buttresses about twice as often as K. homeana. At buttress retreats, tortoises were typically found resting against a buttress and partially covered by leaf litter or small fallen branches.

Table 4. Summary of microhabitat use by 8 telemetered Kinixys in southwestern Cameroon.^a

View Fullscreen

Temperature measurements at individual tortoise locations ranged from 17° to 33°C, but in general were concentrated between 20° and 22°C. High temperature readings occurred during afternoon observations when tortoises were located in a forest gap.

DISCUSSION

The potential pitfalls of translocating tortoises have been reviewed, predominantly in connection with translocations for conservation programs (e.g., Berry 1986; Burke 1991; Dodd and Seigel 1991). The use of translocated tortoises in home range studies may be problematic because of the animal's propensity to home to their original range, and/or because of competition and active exclusion from resources by resident individuals (refer to observations of K. homeana 6). However, despite the inherent problems, translocation is still used as a conservation tool (see discussions in references above). Translocation was used to initiate this study because resident specimens of both species were so rarely encountered (i.e., 2.4 animals/km²/year). Kinixys populations in the study area have been depressed by human hunting pressure (Lawson 2000; Luiselli et al. 2003), so unoccupied habitat was assumed to be available and introductions were assumed to be minimally disruptive to residents. That translocated specimens did not differ appreciably in home range size and home range characteristics such as size of core area, or appear to differ markedly from residents in habitat and microhabitat use warrants their inclusion in this study. However, the aggression experienced by K. homeana 6 implies that competition exists for some resource(s)—at least for K. homeana—and home range information from translocated animals should be viewed with some caution.

Few studies of tropical forest testudinids are available to compare these results (Moskovits and Kiester 1987; Moskovits 1988). Quantitative information on forest tortoise natural history includes studies of syntopic Geochelone carbonaria and G. denticulata in Brazil (Moskovits and Bjorndal 1990). Forest tortoises and predominantly terrestrial turtles (e.g., Lue and Chen 1999) appear to be consistent in their extreme individual variability in home range size and other ecological parameters (i.e., habitat use, activity levels/patterns).

Moskovits and Kiester (1987) found a 200-fold difference in individual Geochelone home range sizes, although individuals with the smallest ranges were tracked for < 60 days, and in some cases < 30 days. Most Geochelone minimum convex polygon estimates were on the order of home range sizes seen here for the two sympatric Kinixys species. Kinixys and Geochelone home range estimates are considerably larger than estimates for the more extensively studied temperate tortoises (e.g., Gopherus, Smith et al. 1997, and references therein), and turtles (e.g., Kaufmann 1995; Lue and Chen 1999; Dodd 2001, and references therein). Unlike temperate turtles, home ranges and movements of tropical forest tortoises are not constrained by potentially limited or limiting retreats (i.e., burrows) or suitable hibernacula, and individual tropical forest tortoises are presumably free to roam for extended periods if they choose. However, the magnitude of forest tortoise home ranges may be somewhat misleading because most individual Kinixys locations occurred within relatively small core area(s) within the total home range.

As Moskovits and Kiester (1987) predicted, Kinixys mirror the individual variability in home range size and activity seen in Geochelone. They attributed this individual variability to the combination of a “benign” ectotherm environment in the rainforest and reduced vulnerability to predation. Although temperatures at Kinixys retreats were typically lower than ambient, extensive use of forest gaps by both species can plausibly be construed as active habitat choice to counter cooler closed-canopy conditions. That certain individual Kinixys remained quiescent for extended periods during the dry season (see below) is further evidence of physiological stresses within the rainforest environment. In addition to significant human harvest pressure (Lawson 2000; Luiselli 2003b; Luiselli et al. 2003), Kinixys face a host of other predators (e.g., snakes, Akani et al. 2003) and potential predators (e.g., chimps, drill, leopards, civets, etc.). While the magnitude of Kinixys (or Geochelone) home range size is likely influenced by an individual's ability to roam extensively with the expectation of finding suitable cover, food, water, mates, or nest sites, sufficient evidence of physiological and predatory stress exists in the rainforest environment to undermine the benign environment argument (Moskovits and Kiester 1987). Variation in individual spatial ecology seen in forest tortoises is likely dependent on individual behavior coupled with the distribution of resources (patchiness), predators, and conspecifics.

Resident males K. homeana 14 and K. erosa 21 had minimum convex polygon home range estimates 2–9 times larger than conspecifics. Some of the size of these particular ranges is attributable to shifting seasonal and yearly activity centers not seen in other individuals. Shifting activity centers and the magnitude of differences seen in home range sizes may indicate that 1) certain individuals were in the process of establishing more discrete home range(s); 2) large-range individuals inhabited areas deficient in some key resource relative to small-range animals, and/or; 3) home range discrepancies represent alternative individual habitat or other resource use strategies. Both males with large home ranges were smaller (PL) than conspecific residents or males with small home ranges, but there is no indication that they were naive to the area, or that suitable small ranges were not available (e.g., translocated male K. erosa 7 was immediately able to establish a small range). No readily apparent differences in habitat or microhabitat use were observed between these individuals with large home ranges and conspecifics. However, the distribution of potentially critical habitats within their ranges (e.g., gaps) or resources (e.g., mates) could not be examined. Although the home ranges of K. homeana 14 and K. erosa 21 encompassed those of tortoises with small home ranges, they may have been excluded from quality patches.

Male–male combat was observed in K. homeana on two occasions. Both instances occurred between translocated K. homeana 6 and resident conspecific males. Aggressive encounters were not observed between resident males of either species, and none was seen between translocated male K. erosa and resident conspecific males. The relatively large range of translocated K. homeana 6 over a short period (14.9 ha/2 mo) increased its probability of encountering conspecific males. Non-overlap of the 50% harmonic surface core areas and the absence of observed aggression among resident males, despite > 2 years of tracking for some specimens, implies that resident animals experience fewer aggressive encounters than would be expected based on the short-term observations of K. homeana 6 and the extensive overlap of male total home ranges (Fig. 2). A system of conspecific awareness and/or reduced aggression akin to the “dear enemy” recognition of Fisher (1954) and Jaeger (1981) may be operating among conspecific males in these populations. The observation of a female K. homeana and a female K. erosa feeding together on mushrooms, and the slight overlap of harmonic surface area isopleths of a male K. homeana and female K. erosa implies that aggression is confined to conspecific males. Potential seasonality or association of aggression with particular resources (i.e., gaps, females) is an enticing explanation but cannot be demonstrated at this time.

Luiselli (2003a, 2003b) and Luiselli et al. (2003) reported that Nigerian Kinixys exhibit strong seasonal differences in activity levels, with animals being active in the rainy season and largely quiescent during the dry months. In the present study, Kinixys showed similar seasonal trends. However, seasonal home range sizes did not differ consistently in a particular direction (Table 2). Kinixys erosa 7 remained inside a hollow log for at least 44 dry season days. Following resumption of activity, K. erosa 7 obtained a larger home range estimate in the last two months of the dry season than was obtained throughout the previous 7-month rainy season, indicating increased activity relative to rainy season movements. Kinixys homeana 14 also exhibited a period of dry season quiescence, remaining buried in leaf litter next to a tree buttress for at least 24 days. However, K. homeana 14 had large overall home range estimates and consistently smaller dry season estimates than rainy season estimates (Table 3). Female K. homeana 18 had more prolonged retreats during the dry season than during the rainy season. Other individuals spent several days to a week at a single location—often but not exclusively during the dry season—but did not exhibit the prolonged sedentary behavior of specimens discussed above. Individual sedentary behavior may represent period(s) of aestivation (Luiselli et al. 2000) relying on late rainy season weight gains to avoid a portion of the presumably stressful dry season. Monthly weight and morphometric measurements may have disturbed individuals that otherwise would have moved less frequently during the dry season (although K. erosa 7 was not disturbed while in the hollow log, but moved on its own after approximately 44 days).

Although sample sizes are small, K. erosa and K. homeana did not differ appreciably in the habitat parameters measured. Overall, K. erosa used more canopied forest retreats while K. homeana used gaps relatively more frequently. However, habitat use varied considerably among individuals, and forest gaps were an important habitat for both species (Table 3). General forest canopy in the study area was relatively open, and the two individuals that sheltered least in gaps were often located near other open habitats (i.e., streams; Table 3). Gaps are important for a variety of forest-dwelling reptiles (e.g., Mabuya spp., Agama agama; Lawson 1993), and male K. homeana 15 repeatedly used a single gap that was occupied on at least one occasion by a 4-m African rock python, Python sebae. Kinixys belliana shell fragments have been found in P. sebae scats in the savannas of northern Cameroon (G. Nicolet, pers. comm., 1997), indicating that this habitat association may not be entirely benign for Kinixys. Moskovits and Bjorndal (1990) mentioned that Geochelone frequent treefall gaps, but did not quantify use of this habitat type.

Anecdotal evidence exists that K. erosa prefers moister habitats than K. homeana (Ernst and Barbour 1989). To some extent this is supported by the propensity of two of three K. erosa specimens to occupy streams, or occur in association with aerial roots of hydrophilic trees. However, both species occurred in the general vicinity of streams with nearly equal frequency (Table 3), and had extensive overlap of total home ranges within a relatively homogeneous environment (Fig. 2).

Luiselli (2003a, 2003b) examined stomach contents and fecal samples from K. erosa and K. homeana from bushmeat markets in Nigeria. He found both species to be omnivorous with K. homeana consuming more invertebrate material than K. erosa. He considered K. homeana to be more of a generalist than K. erosa (Luiselli, 2003a, 2003b). Direct observation, fecal samples, and anecdotal accounts from local residents in Cameroon indicate that both species feed heavily on mushrooms during the rainy season, as Luiselli (2003a, 2003b) also reported for Kinixys in Nigeria. In this study, K. homeana also was observed defecating copious amounts of the fruits/seeds of Pachypodanthium sp. and Vitex sp. Seeds of these plant species recovered from fecal samples were successfully germinated in a nursery, supporting the contention that forest tortoises may be important seed dispersers (Moskovits and Bjorndal 1990).

Assuming that all resident adult Kinixys in the study area may have been encountered over the 2.5 years of the study (n = 6), estimated densities of K. erosa and K. homeana in the Banyang-Mbo Sanctuary could be as low as 0.03/ha for each species. Supporting Janzen's (1974) thesis of depressed reptile biomass in Africa relative to the Neotropics, Kinixys estimates are approximately 37% less than Moskovits' (1988) approximations for Geochelone carbonaria and G. denticulata in northwestern Brazil. Human predation of tortoises in Banyang-Mbo has been estimated at 17%–33% of the adult K. erosa population and 10%–20% of the adult K. homeana population (Lawson 2000). Increased tortoise offtake seen during the rainy season (Lawson 2000; Luiselli 2003b), may indicate increased movement/vulnerability of large home range individuals and/or increased human presence coincident with elevated mammal hunting pressure at this time.

In conclusion, K. erosa and K. homeana exhibited few consistent or definitive ecological trends. Individuals varied greatly in home range size, activity, and habitat use, similar to the individual variability seen in other forest-dwelling testudinids (Moskovits and Kiester 1987; Moskovits 1988; Moskovits and Bjorndal 1990). Not enough is known about tortoise resource requirements throughout life, or the distribution of potential resources (cover, mates, food, etc.) in any rainforest environment to interpret the evolutionary implications of the individual variability seen in tortoise home range and activity levels. In this study, considerable overlap of resource use existed between the two species. Although habitat and microhabitat use reveal subtle differences in spatial ecology, and there was little overlap of core areas, the two species appear to be largely microsympatric. Differences in sexual size dimorphism between the species (Lawson 2001) suggest different selection pressures in their respective evolutionary pasts not yet evidenced in their ecology. Unfortunately, given current levels of exploitation, opportunities are dwindling to refine our understanding of these species and the larger theoretical issues they may encompass.