Patch use under predation risk often results in a change of feeding behaviour in the prey animals. However, such changes only appear if the animals are able to assess under which predation pressure they live. We investigated patch use of Mastomys natalensis under different conditions of avian predation pressure. 
 In replicated maize field plots in Morogoro, Tanzania, avian predators were allowed under natural conditions (control), attracted with perches and nest boxes or kept out with nets. During four one‐week periods in late 1999, we measured rodent feeding decisions with the giving‐up density (GUD) method. Trays with known amounts of millet seeds in sand were placed in pairs, one of them under a cover, the other one in the open. M. natalensis mice were expected to give up sooner in the open trays than in those with cover. We hypothesised that M. natalensis mice could assess the ambient predation pressure leading to larger difference in GUD between covered and non‐covered trays in the plots where predators were attracted. We also made video recordings of the rodent activity at a pair of trays in each treatment. The GUD‐values were significantly lower for the covered trays but predation pressure did not affect this difference. The video observations showed that in the control and netted plots the animals visited trays equally frequently regardless of the cover, while the visits in the predator‐attracted plots occurred significantly more often in the covered trays. We conclude that M. natalensis can assess the ambient predation pressure and adapt its behaviour at a feeding patch. However, the variation in predation pressure in our experiment was not obvious from the GUD. Moreover, we found a strong relation between rodent density and GUD, which may mask variations in perceived predation pressure. Similar GUD values may be reached in different ways and we present models to investigate whether animals’ decision to forage at a food patch is only affected by the seed density at that patch, not by that at a neighbour patch.

Context Rodent pests can have severe impacts on crop production in sub-Saharan Africa. In particular, the multimammate mouse Mastomys natalensis severely damages agricultural crops in southern and eastern Africa, leading to significant losses. Both its population ecology and breeding biology have been studied in agricultural and natural habitats. Population numbers erupt depending on the timing and amount of rainfall and may reach plague proportions, especially in agricultural settings, where it may become a serious pest. However, the ecology of this species, in particular its interactions with other species within the context of human settlement, is poorly understood. It may occasionally enter houses, but the degree to which it does so and the factors influencing this movement are not known. Aims We investigated the relationship between Rattus spp. and M. natalensis entering buildings in an agro-ecological setting. We predicted that M. natalensis would enter houses more readily when food availability was lowest in the surrounding fields, and when the larger Rattus spp. were absent. Methods We followed 40 individuals of M. natalensis in Swaziland and Namibia by radio-telemetry. Mice were captured in maize fields within 50 m of a homestead and fitted with radio-transmitters at three different times corresponding to different stages of crop development: pre-harvest, post-harvest and pre-planting. To corroborate the findings of the telemetry study, a non-toxic marker, rhodamine B, was mixed with standard bait and left at bait stations inside houses in 10 homesteads in Swaziland and Tanzania. Key results Mice remained in the fields during the entire period of study in Swaziland, but entered buildings in Namibia during the post-harvest stage, which may represent a period of food shortage for these mice in the field. Rodents captured after baiting with rhodamine B demonstrated that Rattus spp. predominated within the houses. A small number of rhodamine B-marked M. natalensis were captured outside the houses, the proportion declining with distance away from the houses. Conclusions These results suggest that in a typical rural African setting dominated by subsistence agriculture, Rattus spp. (when present) competitively exclude the smaller M. natalensis from entering houses. Implications Interactions between rodent pest species may be important in determining which rodent species enter houses in rural African landscapes. Consideration of such interactions may play an important role when developing pest management strategies.

Rodents are of concern as reservoirs and transmitters of pathogens that cause zoonotic and other diseases in domestic animals and humans. The contact between wild and urban rodent fauna is increasing in expanding cities in Africa, and this arguably increases the risk of disease transmission to urban populations. When rodents from rural areas are accidentally introduced into cities and encounter urban rodents, for example, in markets, grain mills and butcheries, there is a potential that pathogens are transferred between rural and urban rodent populations. Using a non-toxic biomarker, Rhodamine B (RB), we investigated the distances over which rodents are active around such potential introduction sites. Animals feeding at such sites were traced up to 100 m distant within a period of 10 days. We found that RB is a practical alternative for studying rodent movements. These results may be useful in designing management strategies to reduce the risk of the introduction of new rodent-borne pathogens in cities.