O. Gy. Dely szerk.: Vertebrata Hungarica 20. (Budapest, 1981)

Demeter, A.: Small mammals and the food of owls (Tyto and Bubo) in Northern Nigeria 127-136. o.

DISCUSSION A determined trapping effort to collect small mammals was rather unsuccessful. Trap success, measured as the number of captures per total trap nights, was extremely low in all habitats, ranging from 0 % in cultivated fields around Futuk to about 2 % in transitional vegetation between savanna woodland and riparian forest around Wikki. HAPPOLD (1970) found low densities in the natural habitats in Yankari, with an overall trapping success of 3.4 % for savanna, woodland and rocky habitats. In the rainy season when I collected, rodents seemed to be even fewer. It Is well known that in the seasonal tropical climates of Africa reproduction by many rodents is also seasonal, with rainfall as the major regulating factor (for references, see NEAL, 1977). This is also true for West Africa (HAPPOLD, 1974; HUBERT, 1977). In Senegal, HUBERT (loc. cit.) found that savanna rodents bred at the end of, or shortly after, the rains and lowest densities were at the beginning of the rainy season. A similar demographic pattern may have been one of the factors responsible for the extremely low densities observed, although with such a variation in patterns of rainfall and reproduction, Senegal data may not be relevant to conditions in Yankari. What must also be taken into consideration is trappability

. Small mammal faunas in Africa display a wide range of sizes, and NEAL & COCK (1969) have documented for snap-traps that the choice of traps used may bias catches. It is desirable to use traps of different sizes to ensure good representation of all size classes of small mammals. In Yankari small mammals weighing from 3 g (Crocidura pasha

) to 103 g (

Tatera kempi

) were captured, and the traps I used were probably not fully effective for the larger size categories greater than about 70 g. Of the larger rodents Tatera kempi

, with a mean weight of 96 g for the specimens I collected, featured prominently in the diet of the owls around Wikki, whereas only 6.8 % of the total number of rodent specimens consisted of this species in the collection. Prey selection of such a degree does not seem to be a plausible explanation and the differences must at least partly be attributed to trap selection. Another point in trappability is bait effectiveness. The "kuli-kuli" I used is in taste and smell much like peanut butter, which ANADU (1979) found to be quite effective for savanna rodents in south-western Nigeria. Therefore I rule out the possibility that incorrect bait was responsible for the low capture rates, but the abundance of natural food in the rainy season may have adversely affected the response of small mammals towards the traps. I do not know whether bait was responsible for the extremely few captures of shrews, for they must be quite common as evidenced by the prey of the owls. Long-term effects of burning may also suppress rodent populations, and may be the probable cause for the low densities in Yankari (HAPPOLD, 1970). This aspect of small mammal ecology has been investigated in East Africa by NEAL (1970) and CHEESEMAN & DELA NY (1979). Immediately after a fire both numbers and total biomass of small rodents decreased, but 4-6 months after the burn densities were higher than before the burn and in unburnt areas. This is due to the numerical increase of a few, but not all, species, which indicates that some rodents are better adapted to pre-burn conditions, while others flourish in post-burn enviroment (CHEESEMAN & DELANY, 1979). However, these findings cannot be directly related to conditions in Yankari, for GEERLING (1973) noted down-grading conditions of the savanna woodland in the reserve. He recommended burning early in the dry season to avoid "hot fires" which scorch trees, but this is also the time when perennial grasses are the most sensitive to fire. Uncontrolled burning, accidental or not, occurs all through the dry season. Ground cover in the savanna woodland had become quite scarce by 1978, and there were some areas, e.g. around Kaiban Hills, where the woody vegetation showed signs of severe burning and the ground cover between the trees and bushes was practically absent. Here even such efficient predators as Spotted Eagle-Owls caught only Taterillus gra cilis

, the only species which I captured in really open conditions. I believe that the degraded nature of the habitat is responsible for the large proportion of non-mammalian (mainly arthropod) prey of the owls In that area. Table 2 shows the habitat distribution of the various species. Except for Mus musculoides and Myomys daltoni, which have wide ecological tolerances (HAPPOLD, 1975), most of the species are restricted to one or two habitats. Riparian woodlands and fadamas are inhabited by much fewer species than the savanna. Except for the gerbil Taterillus gracilis

, savanna rodents occurred only in patches of dense ground cover which were rather few and far inbetween. Dasym.ys incomtus

, found in the pellets, has never been trapped in Yankari, but Its occurrance was expected as it mainly inhabits swampy areas that are plentiful in the reserve, although it has occassionally been found in drier habitats (DEMETER, 1978; HAPPOLD, 1975). All the shrews seem to occupy habitats close to the river.