Habitat occurrence and feeding ecology of Crocidura montis and Lophuromys flavopunctatus on Mt. Elgon, Uganda
Abstract
enThe distribution, habitat occurrence and feeding ecology of Crocidura montis and Lophuromys flavopunctatus coexisting in the afro-alpine zone on Mt. Elgon, Uganda, are described. Crocidura montis was the only shrew species occupying the afro-alpine zone, but was not found in forested habitats below 3200 m. While there was considerable overlap in habitat occurrence between the two species, C. montis was most abundant in the montane grassland and in drier sites while L. flavopunctatus occupied a wider range of habitats and wetter sites. C. montis ate a diversity of invertebrates, the most important being Coleoptera, Araneae and Lumbricidae. Invertebrates, mostly Diptera larvae, Lumbricidae and Coleoptera, formed a major component of the L. flavopunctatus diets, plus plant material (particularly Hepaticae and monocots). The arthropod components of the diets reflected their abundance in pitfall samples.
Résumé
frOn décrit la distribution, l'habitat et les habitudes alimentaires de Crocidura montis et de Lophuromys flavopunctatus qui coexistent dans la zone afro-alpine du mont Elgon, en Ouganda. Crocidura montis est la seule espèce de musaraigne qui occupe la zone afro-alpine mais on ne la trouve pas dans les habitats forestiers en dessous de 3200 m. Bien qu'il y eût un important recouvrement de l'habitat des deux espèces, C. montisétait surtout abondant dans les prairies de montagne et dans les sites plus secs tandis que L. flavopunctatus occupait une plus large gamme d'habitats et des sites plus humides. C.montis mange toute une variété d'invertébrés, les plus importants étant les coléoptères, les arachnides et les Lombricidae. D'autre part, des invertébrés, et surtout des larves de diptères, des Lombricidae et des coléoptères, étaient parmi les composantes principales du régime de L. flavopunctatus, auxquelles s'ajoutaient des matières végétales (particulièrement des hépatiques et des monocotylédones). Les arthropodes entrant dans le régime alimentaire reflétaient leur abondance relevée dans les piégeages échantillons.
Introduction
Mt. Elgon is an isolated massif of volcanic origin (Fig. 1) rising to 4321 m above sea level on the Kenya–Uganda border (0°54′−1°25′N and 34°14′−34°45′E). It covers an area of some 80 km by 50 km, and rises from a plateau lying at 1050–2000 m above sea level. Being of Miocene age it is one of the oldest volcanoes in the East African Rift Valley system. It was deglaciated about 12,000 years BP.
Location of the study area (Mt. Elgon, bold and underlined) within the montane region of central Africa and regions with close affinities. Mountain ranges where Crocidura montis occurs are shown in bold
The vegetation on Mt. Elgon is typically montane afro-alpine which supports a thriving community of small mammals. Above the tree-line this community is dominated by herbivorous rodents, the most numerous being Otomys typus and O. barbouri (Clausnitzer, in press (a); Clausnitzer & Kityo, 2001). There are also two species of insectivorous small mammals which are sympatric on Mt. Elgon: the montane white-toothed shrew (Crocidura montis Thomas 1906) and the harsh-furred rat (Lophuromys flavopunctatus Thomas 1888). Crocidura montis has a disjunct distribution in eastern Africa. It has been recorded from the Ruwenzori Mts. and Mt. Kokanjira in Uganda, the Imatong Mts. in Sudan, the Cherangani Hills, Mt. Kenya, southern face of Mt. Elgon and a few other localities in Kenya, and Kilimanjaro, Mt. Meru, and Momela in Tanzania (Demeter & Hutterer, 1986; Dippenaar & Meester, 1989). Here we record it from the afro-alpine zone of Mt. Elgon, Uganda, where it was found to be the only shrew species. Little is known of the ecology of this shrew.
Lophuromys flavopunctatus , in contrast, has been comparatively well studied (e.g. Hanney, 1964 ; Dieterlen, 1976, 1987 ; Bekele, 1994 ; Lavrenchenko, Verheyen & Hulselmans, 1998 ). It is one of the most widespread and numerous rodents in the moister areas of East Africa, inhabiting a range of different habitats with a preference for montane ones. In the Ruwenzori Mts. it occurs from lowland forests at about 500 m above sea level to the afro-alpine zone reaching well above 4200 m and extending into ericaceous habitats and montane moorlands ( Misonne, 1963, 1969 ).
Unlike most rodents, L. flavopunctatus is highly insectivorous. Many rodents consume small numbers of invertebrates, often on a seasonal basis (Field, 1975; Taylor & Green, 1976; Perrin, 1980), but members of the genus Lophuromys eat large quantities of these prey and are obligate insectivores (Verschuren, Straeten & Verheyen, 1983). Hanney (1964) reported that L. flavopunctatus could not survive in captivity unless fed on earthworms, insects and even frogs. Captive individuals fed on a meat-free diet of nuts and seeds suffered large weight loss and did not reproduce (Dieterlen, 1976). It is therefore a potential competitor with C. montis on Mt. Elgon.
This paper reports a comparative study of the altitudinal distribution, habitat occurrence, feeding ecology and niche overlap of these two coexisting, insectivorous small mammals on Mt. Elgon, and presents new information on the ecology of C. montis.
Methods
The study area and trapping regime
The vegetation formations on Mt. Elgon exhibit distinct altitudinal zonation. Below 3200 m are mixed montane forests of Prunus africyana, Ilex mitis, Podocarpus latifolius and forests dominated by Hagenia abyssinica, Rapanea melanophloeus and bamboo (Sinarundinaria alpina). Above 3200 m are areas of heather forest (Erica excelsa), tussock grassland and moorland dominated by Festuca spp., Koeleria spp., Carex runssoroensis, Alchemilla elgonensis, Dendrosenecio elgonensis which have been subject to human impact through grazing and burning. For further details of the vegetation see Wesche (2002). The climate is dominated by seasonally alternating moist south-westerly and dry north-easterly air streams.
Thirteen study sites were established, mainly on the western slopes of Mt. Elgon. They followed an altitude gradient from 2900 m to 4200 m a.s.l and incorporated representative vegetation formations, dry and wet sites. The transect commenced in the Sasa Valley in montane forest (up to 3100 m) and progressed to the Simu Valley in Erica forest (up to 3500 m) and terminated in montane grassland in the Dirigana Valley from 3600 m upwards. Precipitation in the study area was 911 mm per annum at 3750 m, with the wettest months being April and September. Average temperature was 5.4°C.
At each study site, a transect of 30 trapping points was established, each 20 m apart. At each trapping point, one large and one small break-back traps were set, baited with a mixture of cassava flour, peanuts and vegetable oil. They were left in operation for three days and two nights consecutively, and checked once each day. Trapping was carried out every two months between October 1996 and October 1997 at six of the study sites, and once only (between February–October 1997) at the remaining seven sites. Details of the trapping regime are shown in Table 1. All trapped specimens were preserved in formal-saline solution followed by 70% alcohol to await further analysis.
| Site | Vegetation type | Altitude (m) | Hydrology | No. trap nights | No. sampling occasions | Total captures | |
|---|---|---|---|---|---|---|---|
| C. montis | L. flavopunctatus | ||||||
| 1 | Montane forest | 2875 | wet | 600 | 4 | 0 | 10 |
| 2 | Montane forest | 2875 | dry | 40 | 1 | 0 | 0 |
| 3 | Sinarundinaria forest | 3100 | dry | 240 | 2 | 0 | 2 |
| 4 | Hagenia forest | 3250 | dry | 40 | 1 | 0 | 0 |
| 5 | Erica forest | 3300 | wet | 120 | 1 | 1 | 4 |
| 6 | Erica forest | 3460 | dry | 720 | 6 | 20 | 25 |
| 7 | Montane grassland | 3600 | dry | 120 | 1 | 0 | 2 |
| 8 | Grassland/Erica forest | 3200 | wet | 120 | 1 | 4 | 6 |
| 9 | Montane grassland, bog | 3590 | wet | 120 | 1 | 5 | 12 |
| 10 | Montane grassland | 3870 | dry | 720 | 6 | 37 | 20 |
| 11 | Carex bog | 3750 | wet | 720 | 6 | 4 | 60 |
| 12 | Alchemilla shrub | 3850 | dry | 720 | 6 | 20 | 9 |
| 13 | Montane grassland, bog | 4120 | wet | 120 | 1 | 2 | 5 |
Invertebrate sampling
To gauge the availability of invertebrate prey, three pitfall traps were set in six of the study sites representative of each of the vegetation formations, namely montane forest, Sinarundinaria forest, Erica forest, montane grassland, Alchemilla shrub and Carex bog. Pitfalls had a diameter of 5 cm at the rim and were partially filled with formalin solution. They were left open for two consecutive days and nights in each trapping session. In the montane forest and Sinarundinaria forest, where C. montis and L. flavopunctatus were absent or few, pitfall trapping was limited to four and two sessions, respectively, in different months. At the other sites, where these small mammals were more abundant, six pitfall trapping sessions were conducted. While pitfalls do not adequately sample for soil-dwelling invertebrates, they do give an indication of the diversity and relative abundance of invertebrates active on the ground surface.
Diet analysis
Stomach and intestine contents were removed from trapped mammals and food remains were examined microscopically. Invertebrates were identified with the aid of a reference collection which also permitted prey to be categorized according to size. Results were expressed in terms of the percentage frequency of occurrence of different food types (the proportion of specimens containing a named food taxon) and the percentage dietary occurrence of food types (the number of occurrences of a named food taxon as a proportion of the total occurrences of all food taxa).
Niche overlap
Niche overlap between C. montis and L. flavopunctatus was calculated as follows (after Pianka, 1973):
where Oab = mutual overlap between species a and b; pia = proportion of the total catch represented by species a or resource used by species a; pib = proportion of the total catch represented by species b or resource used by species b.
Results
Distribution and abundance of C. montis and L. flavopunctatus
A total of 93 C. montis and 155 L. flavopunctatus were captured. Crocidura montis was found in small numbers at eight of the thirteen study sites while L. flavopunctatus was more widely distributed, occurring in all but two sites and in generally larger numbers (Table 1). Lophuromys flavopunctatus was found throughout the altitude gradient. It occurred in all vegetation types except Hagenia forest, and in both wet and dry sites. C. montis was found only at altitudes above 3300 m (maximum 4180 m) in the montane grassland sites and Erica forest, and was notably absent from the montane forest sites at lower altitudes. It, too, occurred in both dry and wet, boggy sites but 85% of C. montis captures were in dry sites whereas 63% of L. flavopunctatus were captured in wet sites. Crocidura montis was most abundant in the dry montane grassland while L. flavopunctatus was most abundant in the wetter Carex bog in the montane grassland zone. Habitat overlap (Oab) between the two species was 52.5%. The numbers of each species captured per month were too small to elucidate seasonal trends in population dynamics.
Diets of C. montis and L. flavopunctatus
A total of 40 stomachs of C. montis and 159 stomachs of L. flavopunctatus, captured from the different study sites, were examined for food remains. Crocidura montis subsisted mainly on invertebrates and sixteen different invertebrate taxa were positively identified in the diet, most of which were arthropods (see Table 2). The number of food types identified per specimen ranged from one to nine, with a median of five. The most important prey, in terms of both frequency of occurrence and dietary occurrence, were Coleoptera, Araneae and Lumbricidae, followed by Opiliones, Lepidoptera larvae and Heteroptera (Fig. 2). Formicidae and Gastropoda were rarely eaten. Prey ranged in size from <3 mm in length (such as Formicidae and Araneae) to >20 mm (Lumbricidae and larger Diplopoda) but 55% of dietary occurrences were arthropods of 3–5 mm (see Fig. 3). In addition to invertebrates, plant material (unidentifiable seeds and fragments of leaf/stem) was found in 63% of the samples and comprised some 13% in terms of dietary occurrence.
| Food type | C. montis | L. flavopunctatus | ||
|---|---|---|---|---|
| % Frequency of occurrence | % Dietary occurrence | % Frequency of occurrence | % Dietary occurrence | |
| Coleoptera adults | 75.0 | 19.9 | 70.4 | 8.3 |
| Coleoptera larvae | 15.0 | 2.5 | 0.0 | 0.0 |
| Diptera adults | 17.5 | 2.9 | 0.6 | 0.1 |
| Diptera larvae | 12.5 | 2.1 | 82.4 | 22.2 |
| Lepidoptera adults | 2.5 | 0.4 | 0.0 | 0.0 |
| Lepidoptera larvae | 40.0 | 6.6 | 0.0 | 0.0 |
| Homoptera | 17.5 | 3.3 | 8.2 | 0.6 |
| Heteroptera | 25.0 | 5.0 | 0.0 | 0.0 |
| Dictyoptera | 7.5 | 1.2 | 0.0 | 0.0 |
| Gryllidae | 5.0 | 0.8 | 0.0 | 0.0 |
| Acrididae | 5.0 | 0.8 | 0.0 | 0.0 |
| Formicidae | 5.0 | 0.8 | 0.0 | 0.0 |
| Collembola | 0.0 | 0.0 | 0.0 | 0.0 |
| Insects indet. | 2.5 | 0.4 | 52.1 | 4.0 |
| Araneae | 87.5 | 18.6 | 0.0 | 0.0 |
| Opiliones | 47.5 | 7.9 | 0.0 | 0.0 |
| Isopoda | 5.0 | 0.8 | 0.0 | 0.0 |
| Chilopoda | 7.5 | 1.2 | 0.0 | 0.0 |
| Diplopoda | 17.5 | 2.9 | 0.0 | 0.0 |
| Gastropoda | 2.5 | 0.4 | 30.2 | 3.3 |
| Lumbricidae | 52.5 | 8.7 | 64.8 | 17.3 |
| Vertebrata | 0.0 | 0.0 | 18.2 | 7.2 |
| Seeds | 42.5 | 7.0 | 35.2 | 7.5 |
| Hepaticae | 0.0 | 0.0 | 45.3 | 12.1 |
| Monocots leaf/stem | 0.0 | 0.0 | 63.5 | 12.3 |
| Dicots leaf/stem | 0.0 | 0.0 | 25.2 | 2.2 |
| Plant leaf/stem indet. | 35.0 | 5.8 | 0.0 | 0.0 |
| Roots | 0.0 | 0.0 | 31.5 | 2.9 |
The percentage occurrence of major food types in the diets of C. montis (black) and L. flavopunctatus (hatched) from Mt. Elgon, Uganda
The percentage occurrence of invertebrate prey of different sizes in the diet of C. montis from Mt. Elgon
Invertebrates were also a major part of the diet of L. flavopunctatus, contributing 63% of dietary occurrences (see Table 2). Most important of these were Lumbricidae and Diptera larvae (especially Tipulidae), both of which were taken in greater numbers than by C. montis, and the majority of samples contained these prey. Coleoptera were eaten often but not as frequently as by C. montis. Notable was the presence of vertebrate (mammal) remains in stomachs of L. flavopunctatus but not in those of C. montis. In contrast to C. montis, plant material formed an important part of the diet of L. flavopunctatus, and comprised Hepaticae, other plant leaf/stem (mostly monocotyledons and rarely dicotyledons) and seeds. Dietary overlap (Oab) between C. montis and L. flavopunctatus was 36.4%.
Sample numbers of L. flavopunctatus were sufficient to allow some analysis of seasonal trends in the diet. The dietary occurrence of invertebrates, particularly as earthworms and Tipulidae larvae, reached a maximum in the wettest months (April and August). The occurrence of plant material in the diet reached a maximum (50% of dietary occurrences) in February, the driest month when soil invertebrates such as earthworms and Tipulidae larvae were scarce.
Prey abundance, diets and numbers of small mammals
The incidence of many invertebrate taxa in the diet of C. montis and L. flavopunctatus reflected their abundance in pitfall traps, particularly Coleoptera, Diptera larvae, Heteroptera and, in the case of the shrews, Araneae and Opiliones (see Tables 2 and 3). With the exception of Opiliones which had a patchy occurrence, these prey were abundant in each of the habitats sampled (Table 3). Collembola were abundant in pitfall samples but were consumed by neither small mammal. Lepidoptera larvae and Lumbricidae were common as dietary items but not in pitfall samples which underestimate numbers of soil-dwelling invertebrates. While the diet of C. montis comprised mainly invertebrates in all months sampled, these prey decreased in importance in the diet of L. flavopunctatus during the dry season. Numbers and biomass of invertebrates captured in pitfalls were found to be positively correlated with rainfall (Clausnitzer, in press (b)).
| Montane forest | Sinarundinaria forest | Erica forest | Montane grassland | Carex bog | Alchemilla shrub | |
|---|---|---|---|---|---|---|
| No. trap days | 24 | 12 | 36 | 36 | 36 | 36 |
| Coleoptera | 38 | 18 | 13 | 45 | 23 | 19 |
| Diptera | 25 | 33 | 13 | 14 | 12 | 12 |
| Lepidoptera | 1 | 0 | 0 | 0 | 1 | 0 |
| Homoptera | 2 | 1 | 4 | 6 | 12 | 3 |
| Heteroptera | 0 | 0 | 0 | 1 | 1 | 0 |
| Blattidae | 1 | 8 | 0 | 0 | 0 | 0 |
| Formicidae | 0 | 107 | 0 | 0 | 5 | 0 |
| Collembola | 30 | 7 | 19 | 15 | 46 | 9 |
| Araneae | 79 | 7 | 16 | 24 | 14 | 27 |
| Opiliones | 3 | 4 | 10 | 10 | 0 | 1 |
| Isopoda | 0 | 0 | 1 | 4 | 0 | 0 |
| Diplopoda | 1 | 0 | 0 | 0 | 1 | 2 |
| Gastropoda | 0 | 0 | 0 | 1 | 0 | 0 |
| Lumbricidae | 2 | 2 | 1 | 0 | 1 | 0 |
| Biomass (g wet weight) | 1.24 | 0.58 | 1.64 | 3.4 | 1.76 | 0.71 |
Known prey taxa were most abundant in the montane forests at lower altitudes, with large numbers of Araneae, Coleoptera, Diptera (mainly adults) and Formicidae, but biomass was relatively low here. Crocidura montis was absent from these sites while L. flavopunctatus was present in small numbers. Of the habitats occupied by C. montis, montane grassland had the highest numbers of prey and the greatest biomass (Table 3). Coleoptera and Araneae were particularly abundant here. Montane grassland also had the largest number of C. montis, plus many L. flavopunctatus. Despite relatively low invertebrate numbers and biomass in the Alchemilla shrub, C. montis were abundant there.
Discussion
Crocidura montis is a relict species of isolated volcanic massifs in eastern Africa (see Fig. 1 ). Its species limits, as defined by Dippenaar & Meester (1989 ), are not well understood. It may well be that some of the isolated populations ( Fig. 1 ) will deserve species rank after a careful genetic analysis. On Mt. Meru, it was found at 2550–2750 m in the montane forest and bamboo zone ( Demeter & Hutterer, 1986 ) and was assumed to be a forest-dwelling shrew. However, in our study, C. montis inhabited only the high-altitude afro-alpine zone above 3200 m, particularly montane grassland, and was not found in any of the montane forest sites. It resembles most shrews in being a generalist, opportunist insectivore, eating commonly available prey ( Bever, 1983 ; Churchfield, 1990 ). Unlike many African shrews which eat many Formicidae and Isoptera ( Churchfield, 1982, 1985a,b ), these prey were rare or absent from the diet of C. montis , reflecting their patchy occurrence or absence in the study sites. Neither prey were eaten by L. flavopunctatus , but ants are known dietary items ( Hanney, 1964 ).
Shrews eat only small quantities of plant material, usually taking seeds when other foodstuffs are in short supply (Dokuchaev, 1989). Seeds and other plant parts had a relatively high frequency of occurrence in C. montis diets which may indicate that invertebrate prey were in short supply. Indeed, numbers of prey obtained in pitfall traps were rather few.
Lophuromys flavopunctatus is clearly an omnivore, eating both plant material and invertebrates. While invertebrates were a major contributor to its diet, fewer taxa were eaten in comparison with C. montis , which indicates prey selection. The presence of intact Homoptera (aphids) in stomachs suggests that the mice either picked these prey off plants or ingested them accidentally while eating plant material on which aphids perched. The occasional consumption of insects, especially during the breeding season, has been recorded in many different African rodents, including Rhabdomys pumilio ( Perrin, 1980 ; Churchfield, 1985b ), Lemniscomys striatus ( Field, 1975 ), Arvicanthis niloticus ( Taylor & Green, 1976 ) and Mastomys natalensis ( Churchfield, 1985a ; Leirs, 1995 ) but the extensive use of invertebrate prey by L.flavopunctatus , as seen here, is unusual and may allow this species to occupy habitats and niches not usually suitable for rodents. Indeed, L. flavopunctatus can be found at high altitudes where no plants occur but where soil-dwelling invertebrates are still abundant ( Misonne, 1963 ; Salt, 1987 ).
Notable was the occasional occurrence of vertebrate remains in stomachs of L. flavopunctatus. These appeared to be the cannibalized remains of other snap-trapped mammals.
The consumption of liverworts (Hepaticae) by L. flavopunctatus is unusual for rodents. Mosses and liverworts are considered unfavourable food for mammalian herbivores since they have low nutrient value and high concentrations of secondary metabolites (Zimov et al., 1995). These fleshy plants, which occurred mostly in wet habitats (Wesche, 2002), may have been consumed for their liquid content since they were eaten mostly in the dry season and lack of water is a major factor in montane environments (Wesche et al., 2000; Clausnitzer, in press (b)).
Crocidura montis and L. flavopunctatus occurred together in most habitats above the tree-line on Mt Elgon but, in contrast to L. flavopunctatus , the shrew was found only in small numbers in wet areas. The relatively large numbers of L. flavopunctatus in wet sites may reflect its dependence on soil-dwelling prey such as Lumbricidae and Tipulidae which were more abundant in these sites (pers. obs.). These prey were not well-represented in pitfall samples. Crocidura montis also ate these prey, but in smaller numbers and amongst a diversity of other invertebrates ( Fig. 2 ). Crocidura montis subsisted mostly on prey active on the ground surface and in the litter layer while L. flavopunctatus ate mostly soil-dwelling prey. These differences between the two mammal species may reflect some habitat and trophic separation between them. Indeed, dietary overlap was small, indicating little competition for food between the two species.
Acknowledgements
The authors thank Dr F. Dieterlen for interesting discussions and comments on the collections, Z. Gibaba and B. Moguli for their assistance in the field, Dr K. Wesche for company on Mt Elgon and for providing vegetation and climate data, and Prof Miehe for supervising parts of the work. Thanks are extended to the German Academic Exchange Service (DAAD) and the German Scientific Foundation (DFG) who financed V. Clausnitzer in the field, and to the Uganda National Council for Science and Technology and the Uganda Wildlife Authorities who granted permission for this study.
