Seasonal and annual fluctuations of deer populations estimated by a Bayesian state-space model.

Inoue Mizuki,Atsushi Takayanagi,Daisuke Fujiki,Keitaro Fukushima,Yuuki Okamoto,Masae Iwamoto Ishihara,Hiroki Itô,Shigeru Fukumoto,Masaru Sakai,Michimasa Yamasaki,Shota Sakaguchi,Masaya Katsuki,Hikaru Nakagawa,Yong Zhang

doi:10.1371/journal.pone.0225872

Abstract

Deer overabundance is a contributing factor in the degradation of plant communities and ecosystems worldwide. The management and conservation of the deer-affected ecosystems requires us to urgently grasp deer population trends and to identify the factors that affect them. In this study, we developed a Bayesian state–space model to estimate the population dynamics of sika deer (Cervus nippon) in a cool-temperate forest in Japan, where wolves (Canis lupus hodophilax) are extinct. The model was based on field data collected from block count surveys, road count surveys by vehicles, mortality surveys during the winter, and nuisance control for 12 years (2007–2018). We clarified the seasonal and annual fluctuation of the deer population. We found a peak of deer abundance (2010) over 12 years. In 2011 the estimated deer abundance decreased drastically and has remained at a low level then. The deer abundance gradually increased from April to December during 2013–2018. The seasonal fluctuation we detected could reflect the seasonal migration pattern of deer and the population recruitment through fawn births in early summer. In our model, snowfall accumulation, which can be a lethal factor for deer, may have slightly affected their mortality during the winter. Although we could not detect a direct effect of snow on population dynamics, snowfall decrease due to global warming may decelerate the winter migration of deer; subsequently, deer staying on-site may intensively forage evergreen perennial plants during the winter season. The nuisance control affected population dynamics. Even in wildlife protection areas and national parks where hunting is regulated, nuisance control could be effective in buffering the effect of deer browsing on forest ecosystems.

Highlights

In the past few decades, deer have become increasingly abundant worldwide [1, 2]; this population increases have contributed to the degradation of plant communities and ecosystems [3,4,5,6]
We introduced stochasticity into the deer population dynamics
We modeled the number of deer seen by block count (Bt) as follows: Bt PoissonðytÞ

Summary

Introduction

In the past few decades, deer have become increasingly abundant worldwide [1, 2]; this population increases have contributed to the degradation of plant communities and ecosystems [3,4,5,6]. Large ungulates are able to breed under low food availability [7], the birth rate of deer would not largely decrease even in a degraded forest; the density-dependent decline in the birth rate of deer occurs at a later period of the outbreak stage [8]. The survival rate of adult deer was high even in a poor nutritional environment [9]. Deer is a species that can live in high densities and low-nutrient environments. If predators (e.g. wolves) are absent, hunting is one options to control deer populations under these conditions [10, 11]

Methods

Results

Conclusion