Using artificial food patches to investigate the foraging behaviour of Rattus rattus L., and testing for suitable study populations of Mus musculus L., in New Zealand.

Abstract

Introduced rodents are pest species in New Zealand, in particular Rattus rattus L. (the ship rat), and have contributed to the decline in some native forest biota. Within New Zealand native forests, there is little knowledge of the microhabitats ship rats use to forage or how predation risk from Mustela erminea L. (stoats) and other predators affects these choices. This thesis investigates the influence of predation risk on ship rat foraging behaviour in southern beech forest and mixed podocarp–broadleaved forest. A subsidiary investigation looks at whether the activity of Mus musculus L. (house mice) in grasslands is adequate to measure their foraging behaviour. To investigate ship rat foraging, artificial food patches (seeds embedded in a non–food sand matrix in trays) were used to quantify foraging behaviour by measuring Giving–Up–Densities (GUDs), which are the number of seeds remaining after a night of foraging; food patches with a lower GUD (i.e. less food remaining) indicates more foraging in a ‘safer’ microhabitat. Artificial food patches were first deployed in Nelson Lakes National Park southern beech forest, and then in Orongorongo Valley and Kaimai Range mixed podocarp–broadleaved forest. Within each forest type, artificial food patches were placed in three microhabitats (covered (by vegetation), uncovered, and tree trunk (1.5 m up a tree trunk)), in areas of relatively low predation risk (stoat control areas) and relatively high predation risk (no predator control). To investigate house mouse activity and distribution, ChewTrack Cards (CTCs; a monitoring device) were deployed on eight Otago grassland grids. It was found that in Nelson Lakes National Park southern beech forest, ship rat foraging behaviour (i.e. their GUD) could not be measured because no ship rats foraged in the artificial food patches; this was possibly because of low ship rat activity at the time. In the Orongorongo Valley and Kaimai Range mixed podocarp broadleaved forests, ship rat GUDs either showed: (1) a tendency to prefer foraging (i.e. lower GUDs) in kiekie (Freycinetia banksii A.Cunn.) covered microhabitats, or (2) no foraging preference (i.e. similar GUDs) between covered (by kiekie), uncovered or tree trunk microhabitats. However, these foraging behaviours could not be attributed solely to the level of predation risk, because of variations in ship rat densities, background food availability, surrounding microhabitat variables (up to 20 m in radius), and rainfall at the time these foraging behaviours (GUDs) were measured and they may also explain the foraging behaviour. On Otago grassland grids, house mouse ii Preface populations were found to have low levels of activity and therefore small scale deployment of artificial food patches may not successfully measure their foraging behaviour (i.e. GUDs). The two main conclusions are: firstly, that the use of artificial food patches to measure the foraging behaviour of ship rats in southern beech forest, and that of house mouse in Otago grasslands, will require a large deployment of seed trays specifically where rodents have previously been detected. Secondly, while the observed ship rat foraging behaviour in mixed podocarp–broadleaved forest could not be unequivocally attributed to predation risk, nor readily extrapolated to other forest locations or seasons, ship rat control programmes might benefit from locating control devices in kiekie covered microhabitats. This research was done in association with the Department of Conservation and Landcare Research.