Analysis on Population Level Reveals Trappability of Wild Rodents Is Determined by Previous Trap Occupant

Abstract

Live trapping is central to the study of small mammals. Thus, any bias needs to be understood and accounted for in subsequent analyses to ensure accurate population estimates. One rarely considered bias is the behavioural response of individuals to the trap, in particular the olfactory cues left behind by previous occupants (PO). We used a data set of 8,115 trap nights spanning 17 separate trapping sessions between August 2002 and November 2013 in Wytham Woods, Oxfordshire, UK to examine if the decision to enter a trap was affected by the PO, if this was detectable in traditional Capture-Mark-Recapture trapping data (i.e., individuals not uniquely marked), and if it was possible for this effect to bias the population estimates obtained. Data were collected on Apodemus sylvaticus, Myodes glareolus, and Microtus agrestis. Three Generalised Linear Models revealed a significant tendency for the three species to enter traps with same-species PO. With, for example, A. sylvaticus 9.1 times more likely to enter a same species PO trap compared to one that contained a M. agrestis in the grassland during the nocturnal period. Simulation highlighted that, when all other factors are equal, the species with the highest PO effect will have the highest capture rate and therefore return more accurate population estimates. Despite the large dataset, certain species-, sex-, and/ or age-combinations were under-represented, and thus no effects of any additional individual-specific characteristics could be evaluated. Uniquely marking individuals would allow for the PO effect to be disentangled from other biases such as trap-shyness and spatial heterogeneity, but may not be possible in all cases and will depend on the aims of the study and the resources available.

Fig 1. The proportion of each species caught.

Depending on the previous occupant in the trap. The values were calculated from the raw data.

Fig 2. Previous Occupant, habitat interaction plot for wood mice.

The interaction would suggest that the effect the PO is less pronounced in the woodland compared to the grassland Habitat. Wood mice are less likely to enter traps that previously contained field voles compared to those that contained bank voles.

Fig 3. Simulated captures showing the proportion of voles caught.

(a) with a 1:1 species ratio of mice to voles, (b) with a 1:4 species ratio of mice to voles in the grassland and a 4:1 ratio of mice to voles in the woodland. The red dashed line indicates the expected proportion of voles caught during the trapping simulations based just on the species ratio E _v. The blue dotted line represents the expected proportion of voles caught during the trapping simulations based on the capture chance without any previous occupant effect C _v. Capture proportion being defined as the number of individuals of a particular species caught divided by the total number of individuals of both species caught during the trapping session.

Fig 4. Repeat runs of simulated captures varying mouse same species multiplier.

The proportion of mice caught compared to voles, for increasing values of mouse same species PO multiplier (C _s). The initial capture chance was 20% and the other PO multipliers were set to 1.0 for both species and kept constant for all runs of the simulation.