Homopus signatus signatus, the Namaqualand speckled padloper (or northern speckled cape tortoise), is the world's smallest tortoise species, occurring exclusively in rocky habitat in northwestern South Africa (Branch 1998; Boycott and Bourquin 2000). The biome in which it occurs is known as Succulent Karoo Shrubland (Branch 1998). Recently, available information on the ecology of H. s. signatus has increased considerably (Loehr 1999, 2002a, 2002b, 2004; Klerks 2002). One aspect studied is diet. In previous literature, the species has been suggested to have a specialized diet (Barzyk 1994), and to feed on succulents (Branch 1998). However, Loehr (1999) reported that captive specimens thrived on a diet that consisted of plantain, dandelion, clover, endive, chicory, and fruits. In a study of fecal material of wild H. s. signatus, the species appeared to feed on a variety of plants, with 4 (Oxalis spp., Leysera tenella, Grielum humifusum, Crassula thunbergiana minutiflora) making up an important proportion of the total (Loehr 2002a).

Loehr (2002a) was unable to identify more than about half of the collected fecal material. This was because much of the material was too digested for identification using available morphological plant features. In addition, 1 plant species that was frequently found in the feces could not be identified because it was not present in the reference collection. Although investigations of dietary preferences (Rall and Fairall 1993; Mason et al. 2000), nutrition (Oftedal and Allen 1996), and digestive efficacy (Barboza 1995; Hailey 1997) will eventually be required to obtain a full understanding of the feeding ecology of H. s. signatus, the current incomplete information on its diet would also benefit from additional studies of the components. In this paper, I analyze focal observations of feeding in wild H. s. signatus accumulated during 5 study periods, and compare these with the previously published results of fecal analysis.

Method

During 5–6 weeks in September and October 2001–2004, a 36,340-m² study site near Springbok, South Africa, was methodically inspected for H. s. signatus. In addition, an average of 15 tortoises were followed by means of thread-trailing (Loehr 2002b), or radiotelemetry. A detailed description of the study site can be found in Loehr (2002b), with the plant species listed in Loehr (2002a). When a tortoise was discovered, its behavior was recorded. For feeding specimens, the food items were identified in the field. Virtually all plant species in the study area had been identified previously by comparison with herbarium material (Loehr 2002a). When a nonfeeding tortoise had remains of plants in its beak, these were included as food items. Each plant species eaten by a tortoise represented 1 observation. The plant parts eaten (leaf, flower, stem, or fruit) were also recorded. To increase sample size, the results of the 2001–2004 study periods were combined with feeding observations made in August–September 2000 at the same site (Loehr 2002a).

Feeding observations were compared with the results of fecal analysis (Loehr 2002a). In the fecal analysis (samples collected in August–September 2000), each plant species present in the total produced fecal material during one tortoise encounter represented one observation. Plant parts were also distinguished and recorded. Plant species on which tortoises had been seen feeding were listed, and their frequencies were compared with the frequencies of these species in the feces by means of Kolmogorov-Smirnov tests (Sokal and Rohlf 1981). Similar analysis was performed for plant part frequencies. The results of the statistical tests are reported as D_max statistic (D_m) and the critical (smallest) value of D_m for p < 0.05 (D_.05).

Results

In total, 1580 observations of 244 tortoises were made, of which only 78 (4.9%) were observations of feeding on one or two plant species. Dietary items were recorded for 37 males, 38 females, and 9 juveniles, and the total number of food plant species identified was 25 (Table 1). Observations of feeding on Oxalis spp., Leysera tenella, and Crassula thunbergiana minutiflora were most frequent (D_m = 0.502 > D_.05 = 0.148) (Table 1). A comparison of the frequencies of the 25 species consumed with their frequencies in the fecal samples revealed no significant difference (D_m = 0.147 < D_.05 = 0.148) (Table 1). However, Grielum humifusum was strikingly absent from the observations made in 2001, 2002, and 2004 (1 and 3 observations in 2000 and 2003, respectively), although it was abundant in the fecal samples. The feeding observations in 2001–2004 revealed 11 new food plant species (five new families) for H. s. signatus: Ballota africana (Lamiaceae), Cleretum papulosum (Mesembryanthemaceae), Erodium cicutarium (Geraniaceae), Hemimeris montana (Scrophulariaceae), Lotononis sp. (Fabaceae), Microloma sagittatum (Asclepiadaceae), Moraea fugax (Iridaceae), Massonia sp. (Lilliaceae), Osteospermum amplectens (Asteraceae), Wahlenbergia annularis, and Wahlenbergia sp. (Campanulaceae).

Table 1. Frequencies of plant species on which wild Homopus signatus signatus were observed feeding (present study), and frequencies of these species in tortoise fecal samples (from Loehr, 2002a).^a

View Fullscreen

Feeding on leaves (44%) and flowers (35%) was more often observed than feeding on stems (20%) and fruits (2%) (D_m = 0.289 > D_.05 = 0.118). Although fecal samples contained more flowers than leaves, no significant difference with the feeding observations could be distinguished (D_m = 0.109 < D_.05 = 0.118). In September 2002 and 2004, observations were made of 2 males eating 8–10 flowers of L. tenella without eating any other readily available part of this plant species. A third observation of selective feeding was made in September 2002, when a female managed to eat a flower of M. fugax that was 15 cm above the surface (without showing any intention to eat other plant parts or species that were closer), by gradually pushing the flower stalk to the ground.

The unidentified food item that had been found in 41% of the fecal samples in 2000 (Tables 1 and 2 in Loehr 2002a) was found to be the flower of Diospyros ramulosa. These small flowers were present in relatively large quantities under D. ramulosa shrubs but had previously been overlooked.

Discussion

Although Loehr (2002a) was able to identify only 50%–60% of the fecal material of H. s. signatus, the current analysis of focal feeding observations confirms that Oxalis spp., L. tenella, and C. t. minutiflora are among the most important food sources for this species in its studied habitat. Two plant species, G. humifusum and D. ramulosa, that occurred frequently in feces, were rarely or not at all recorded in feeding observations. The first species was common during most of the 2000 study period; in contrast, it was much scarcer in 2001–2004 and this may have been reflected in the feeding observations. Diospyros ramulosa flowers were eaten by tortoises when they had fallen off the shrubs and were scattered in dead foliage under them. Such locations were very hard to search for tortoises, and especially to make feeding observations. Tortoises usually stopped feeding immediately and moved away when they noticed an observer.

The new food plant species recorded in 2001–2004 add to the reported large variation in the diet of H. s. signatus (Loehr 2002a), and these species may have been part of the unidentifiable fecal material in 2000. The newly recorded species accounted for as much as 24% of the total number of species the tortoises were found to consume. Conversely, 37% of the plant species identified in the fecal matter were also noted as food items in this study. The research protocol did not allow temporal variation in dietary components to be determined (Nagy and Medica 1986; Moskovits and Bjorndal 1990; Rall and Fairall 1993), and both methods may underestimate the total number of food plants utilized.

Food items frequently eaten may have been preferred or may simply reflect abundance of the food item. The larger proportion of flowers in the feces, compared to feeding observations, may have been the result of two relatively dry study periods, 2003 and 2004, with fewer flowers (V.J.T. Loeh, pers. obs., 2005). On the other hand, the observation of three tortoises specifically feeding on flowers of L. tenella and M. fugax suggests that preferences may exist. Captive H. s. signatus have also been reported to have a preference for feeding on flowers (Loehr 1999). Furthermore, since dietary preferences have been documented in other tortoises (Moskovits and Bjorndal 1990; Rall and Fairall 1993; Hailey et al. 1998; Mason et al. 1999, 2000; Jennings 2002; Oftedal et al. 2002), this aspect requires further investigation in H. s. signatus.