Globally, the genus Rattus is one of the most influential exotic species due to its high rates of competitive exclusion and large dietary breadth. However, the specific foraging strategies of urban and urban-adjacent populations remain largely unknown. We examined Rattus spp. dependency on human food supplementation in a peri-urban population. Through a natural experiment made possible by the COVID-19 shelter in place order in Santa Cruz California, USA, we measured changes in activity between invasive rats and native rodents with and without human supplementation. We measured invasive rat presence in normal (pre-COVID) conditions near dining halls and similar waste sources, and again under COVID lockdown conditions where all sources of human supplementation were removed. We found a decrease in Rattus presence after the removal of human refuse (p < 0.001), while native small mammal presence remained unchanged. These results have strong conservation implications, as they suggest that proper waste management is an effective, targeted, and less-invasive form of population control over conventional forms of poison. 

All fieldwork was conducted from November 23, 2019, to February 14, 2021 in two trapping periods. Trapping in Period 1 (November 2019 – March 2020) served as a control for Rattus presence under normal (pre-COVID) conditions and lasted for 14 weeks. Trapping during Period 2 (October 2020 – February 2021) took place during the shelter in place order and was conducted at the same sites as Trapping Period 1. Trapping Period 2 lasted for 14 weeks, and served as our experimental treatment approximately one year later during the COVID lockdown when human supplementation was absent. To test our hypothesis that invasive rats are dependent on human supplementation, the first site in each trapping transect originated at a semi-exposed source of human refuse (dining hall dumpster, compost pile, etc.). We placed each sequential site in a direction away from the initial refuse source, and towards natural habitat. To reduce the probability of capturing the same individual, each transect contained three sites spaced 75 m apart each, for a total transect length of 150 m (N = 15 sites) (Oca et al. 2017). We used camera traps (Bushnell Trophy Cam; Bushnell Corp., Overland Park, KS, USA) that were left on-site for four consecutive nights to document Rattus presence (N = 60 trap nights per trapping period). We programmed the cameras to capture three images every 30 seconds once movement was detected. Each site was baited with 30 grams of sunflower seeds (Helianthus annuus). All sites were originally open to the public (vehicles not allowed) but were closed from March 31, 2020 to September 2, 2020.

To test the influence of habitat on Rattus presence and to account for errors in camera trap detection, we used occupancy modeling in the unmarked package in R (Fiske 2011). We binary-coded invasive Rattus presence or absence with a 1 or a 0, respectively, and indexed habitat composition at each site as a covariate (1 = mixed redwood, 2 = mixed oak, 3 = meadow) (Fiske 2011).  We treated each individual night as a survey trial, which yielded four repeated surveys at each site. Finally, we back-transformed our detection and occupancy estimates and fitted 95% confidence intervals. To test the influence of habitat on Rattus presence, we created three models: one with variable occupancy, one with variable detection, and one with both variable occupancy and variable detection as a function of habitat.