Spatiotemporal changes in antlerless proportion of culled Sika deer in relation to deer density

Kei K.Suzuki·Teruki Oka·Masatoshi Yasuda

Abstract Population management of herbivores is widely applied to decrease damage to forests in the northern hemisphere.Culling more antlerless deer,including females and fawns,is an effective means of reducing the population and helps to improve management efficiency.To increase the efficiency of antlerless-biased culling,we assessed the spatiotemporal distribution of the antlerless ratio in culled sika deer (Cervus nippon) in relation to population density.We hypothesized that the antlerless ratio is higher at the center of the distribution than at the margins of the deer’s range,because dispersal processes differ according to sex and age:Young male deer first disperse into an unoccupied area,and then males may become established in the new area for approximately 10–15 years before the number of females increases.A statistical model revealed spatiotemporal changes in the antlerless ratio of culled sika deer.The change in the ratio basically fit the density distribution of the deer,which was estimated independently on the basis of fecal pellet count surveys.The antlerless ratio estimated from the model increased asymptotically as deer density rose.The results support our hypothesis and suggest that antlerless sika deer will be more easily culled at the center rather than at the margins of the distribution range.These findings should help to increase the efficiency of managing the deer via antlerless-biased culling.

Keywords Antlerless·Culling·Density·Distribution·Large herbivore·Wildlife management

Introduction

Browsing by abundant large herbivores causes serious damage in both anthropogenic and natural environments (Gill 1992;C?té et al.2004;Putman et al.2011b).For example,browsing of crops and tree seedlings by herbivores results in economic losses to farmers and foresters (Putman and Moore 1998;Putman et al.2011a).In addition,long-lasting overbrowsing by herbivores can cause recruitment failure of trees and thus inhibit the natural regeneration of forests(Whigham 2004;Putman et al.2011b;Ramirez et al.2018).Therefore,decreasing browsing damage is necessary from both economic and ecological sustainability points of view.

Population management of herbivores is widely applied to decrease forest damage in the northern hemisphere,including Europe,North America,and Asia (McCullough et al.2009;Apollonio et al.2010;Hewitt 2011;Putman et al.2011a).The free culling of individuals is sometimes harmful to population management,however,selective culling is essential (Swan et al.2017).Culling of more female or antlerless deer including females and fawns is an effective means of population reduction and helps to increase management efficiency (Matsuda et al.1999;Brown et al.2000;Ueno et al.2010).Survival of adult females is one of the most important parameters for population growth (Gaillard et al.1998;DeCesare et al.2 012),because female abundance determines recruitment (Solberg et al.1999),which is directly linked with population dynamics (Gaillard et al.2000;Raithel et al.2007).Moreover,culling mature males in a high-density population can increase recruitment by improving the nutritional status of females and fawns (Clutton-Brock et al.2002).Previous studies reported that culling an entire matriline kept density low for 2–10 years (Porter et al.1991;McNulty et al.1997),implying that culling of antlerless deer limits recolonization in a local area.

Animals often move toward unexploited areas to locate resources (Bhat and Huffaker 2007;Nathan et al.2008;Abrahms et al.2021),and the movement of individuals is an important aspect to consider for wildlife management(Allen and Singh 2016).Although some ungulate species have similar dispersal characteristics in both sexes (Coulon et al.2006;Gaillard et al.2008;Bonenfant et al.2009),the dispersal processes in sika deer (Cervus nippon) differ between males and females (Ratcliffe 1987;Swanson and Putman 2009).Usually,young male sika deer,at 1–3 years of age,are the first to disperse into new areas,and adult males typically become established in the new areas approximately 10–15 years before females are present.The number of males increases and territories are established,and then a few females penetrate the area.Finally,the number of females also increases and the population size increases rapidly.

In light of these dispersal processes,it is likely that the female ratio gradually decreases with increasing distance from the major high-density population center.In addition,fawns increase in proportion to the number of females owing to the stable pregnancy rate in sika deer,regardless of density (Putman and Clifton-Bligh 1997;Ueno et al.2010).Therefore,a higher ratio of antlerless deer,including both females and fawns,will exist at the center of the distribution range than at the margins (Fig.1).

Fig.1 Predicted sexual difference in the density distribution of sika deer

In this study,to increase the efficiency of population management of sika deer,we considered antlerless-biased culling by focusing on the different dispersal processes between the sexes.To test our hypothesis,we assessed spatiotemporal variation of the antlerless ratio of culled sika deer and compared the ratio with the population’s density distribution.

Materials and methods

Sika deer

Cervus nipponis native to eastern Asia,and the species has been widely introduced into many parts of the world,including Europe,North America,and New Zealand (McCullough et al.2009).In some areas,sika deer populations have increased to the point of causing damage to habitats,agricultural crops,and commercial forests.An additional problem with this species is that it hybridizes with native deer species in Europe (Abernethy 1994;Putman and Hunt 1994;Barto? 2009;Pérez-Espona et al.2009).Therefore,population management of sika deer is an important issue across several continents (McCullough et al.2009;Pérez-Espona et al.2009).

Sika deer mature at 1 year of age.The pregnancy rate of 1-year-old females reaches approximately 80%,and that of adult deer (2+years) is generally even higher (Putman and Clifton-Bligh 1997;Koizumi et al.2009).These high pregnancy rates are maintained regardless of population density(Putman and Clifton-Bligh 1997;Ueno et al.2010).Moreover,even at high densities,yearly natural mortality rates in female sika deer are low,at approximately 5–15% (Uno and Kaji 2006;Ueno et al.2018).

Study area

Our study area was Kumamoto Prefecture on Kyushu Island in western Japan.Kumamoto Prefecture contains large areas of forest (total of 462,965 ha).Natural forest vegetation differs above and below an elevation boundary at 800–1000 m,with deciduous broad-leaved forests at higher elevations and evergreen broad-leaved forests at lower elevations.In addition,plantations comprising Japanese cedar (Cryptomeria japonica) and Japanese cypress (Chamaecyparis obtusa)cover approximately 60.5% of the prefecture’s forested area.Sika deer have seriously damaged natural and anthropogenic vegetation over wide areas within the prefecture (Ohashi et al.2014;Suzuki et al.2021b).

Deer density and culling data

In Kumamoto Prefecture,sika deer densities were estimated by using the fecal pellet count method in 2006,2009,2010,2014,and 2019.Suzuki et al.(2021a) standardized the densities by using the fecal pellet count and the vector autoregressive spatiotemporal (VAST) model (Thorson and Barnett 2017;Thorson 2019) and estimated spatiotemporal c hangesinthe d(eerde)nsity acrossKyushuIsland.Expect ed deerdensitiesdsi,tifor eachsampleiwere estimated by using a log-linked linear predictor and a lognormal distribution with the following formula:

whereβis the intercept for yeart,andωandεare spatial and spatiotemporal random effects for yeartand locations(latitude and longitude).

Sika deer have been culled by game hunting,nuisance control,and survey harvesting in Kumamoto Prefecture.In these culling efforts,nearly all the deer have been captured by using traps,and guns are occasionally used to kill trapped individuals.Age and sex data for deer captured in traps,which are randomly encountered by individuals,may more accurately reflect population structure than do data based on results of gun-hunting,which can target certain individuals.The locations and sexes of culled deer recorded by trappers were reported to the Kumamoto prefectural government,which tallied these data for 5-km2grid cells from 2009 to 2017.In our study,females and males in the records were treated as antlerless and antlered,respectively.Fawns were not sexed in the data reported to the government,so it is possible that antlerless male fawns were included among those recorded as females.We used data from 2009 (115 grid cells),2010 (71),and 2014 (157)–the years when fecal pellet groups were also monitored by systematic survey.

Spatiotemporal changes in the sex ratio in relation to the density distribution

First,we calculated the sex (antlerless and antlered) ratioSR(t,m) in yeartin 5 km2grid cellsmfrom the numbers of female (antlerless),F(t,m) ,and male (antlered),M(t,m),culled deer.To remove bias due to sample size differences,we treatedSR(t,m) in R ver.3.5.2 (R Development Core Team 2018) as follows:

where c bind is a function indicating the ratio of antlerless to antlered culled deer.

Second,we estimated the deer density index (DDI)in yeartat locationsas follows:

From these data,we evaluated the effect ofonSR(t,m) by using a generalized additive model (GAM).We included the effects of year and location within the model to account for spatiotemporal changes in the sex ratio.The expectedsexratiofor each sampleiwas estimate d by using a logit-linked linear predictor and a binomial distribution with the following formula:

whereβis the intercept,γ andρare smooth termed fixed effects ofand meshm(tensor production of latitude and longitude at the center of the mesh),ηis a fixed effect of yeart,andφis a fixed effect of the interaction between yeartand grid cellm.The best model was selected on the basis of the Akaike information criterion (AIC).We ran the GAM by using the“gam”function of the“mgcv”package (Wood 2002) in Rver 3.5.2 (R Development Core Team 2018).

Results

Spatiotemporal changes inSR(t,m) are shown in Fig.2.In 2009,higher antlerless ratios occurred in the center and southeastern regions of Kumamoto Prefecture than in the northeastern region.A similar trend was also observed in 2010 and 2014.However,the antlerless ratios were high in the northern part of the study area in 2010,but only in three grid cells.Low antlerless ratios were observed in the southwestern region in 2014.

Fig.2 Spatiotemporal changes in the antlerless-to-antlered ratio of sika deer in each 5-km2 grid cell.Number at the top left is the year

The full model (adjustedR2=0.276,deviance explained=31.4%) with the minimum AIC was selected as the best model (Table 1) and all fxied effects were significant(Table 2).Thus,the result of our GAM analysis showed spatiotemporal changes in the sex ratio in relation to the DDI estimated from the VAST model.

Table 1 Model structures and selection based on the Akaike information criterion (AIC)

Table 2 Summary of generalized additive model evaluating the effects of population density and spatiotemporal changes on the ratio of culled antlerless sika deer

The antlerless ratio estimated from the GAM asymptotically increased with DDI (Fig.3).The sex ratio basically fitted the DDI (Fig.4).In 2009,DDI values in the southeastern region were higher than those in the northeastern region.This trend was also observed in 2010.In contrast,although DDI values in 2014 were high in the center of the study area,those in the northeastern region and across the southern region were low.Therefore,in the southeastern region,mismatches between the sex ratio and DDI were observed.

Fig.3 Regression curve (solid line) and 95% confidence interval(dashed lines) of the relationship between the deer density index(DDI) and rate of antlerless individuals among culled sika deer

Fig.4 Spatiotemporal changes in the deer density index (DDI) in each 5-km2 grid cell.Number at the top left is the year

Discussion

Our analysis revealed an asymptotic positive regression between the DDI and antlerless-to-antlered ratio in the sika deer population (Fig.3).This finding supports our hypothesis that the rate of antlerless deer decreases as the distance from the main population center increases (Fig.1).In 2014,however,a mismatch was observed between the antlerless ratio and DDI in the southeastern region.This area has

been the center of the sika deer population in Kumamoto Prefecture for a long time (Suzuki et al.2021a).In 2014,however,the population decreased not only in Kumamoto Prefecture but also across the island (Suzuki et al.2021a).This mismatch probably reflected episodic decrease in the population density at the center of the distribution range.Because the mismatch may cause an overestimation of the antlerless to antlered ratio in low-density areas,the actual ratio may be lower than the estimate.After removal of the overestimated value,however,our hypothesis was even more strongly supported.Thus,our results indicate that antlerless deer were more easily culled at the center of the distribution range than at the margins.

It is known that intensive gun-hunting in high-density areas can effectively cull deer in the initial phase of population management (Iijima 2017).Although our results are based mainly on trapping data,this information may also help to increase the culling of antlerless deer by gunhunting because the rate of encounters with antlerless deer is higher at the center of the distribution range than at the margins.Because individual hunters have difficulty in estimating deer density,however,national and local governments should constantly survey the density distribution and encourage active culling of antlerless deer by hunters in high-density areas.Beside hunting,the employment of professional sharpshooters should also be considered (DeNicola et al.1997;DeNicola and Williams 2008).Although we did not have separate data for females and fawns in this study,fawns basically remain with their mothers.Because lactating female sika deer tend to frequent artificial salt licks(Ping et al.2011),it may be effective to cull the population of antlerless deer by employing professional sharpshooters and using salt as a bait.

Culling in high-density areas often causes increased recruitment (Russel 1931;Graham 1935).For example,in organisms with a Ricker type of stock–recruitment relationship,although recruitment is enhanced with the increase from low stock levels,recruitment starts to decline when the stock exceeds a certain threshold (Ricker 1954).Although the stock–recruitment relationship of sika deer is unknown,we consider that the species has a Beverton-Holt (Beverton and Holt 1957) or hockey stick (Barrowman and Myers 2000) type of relationship,and not the Ricker type,for the following reasons.Studies have shown that population density has a very small effect on the pregnancy rates of sika deer,with rates over 80% maintained regardless of density(Putman and Clifton-Bligh 1997;Ueno et al.2010).In addition,although the fawn survival rate in high-density populations could be decreased by density effects,the impact of such a decline on population growth is limited (Ueno et al.2010).Kaji et al.(2004) reported that the population growth rate of high-density populations remains over 1.0.Therefore,culling antlerless deer at the center of the distribution range should be effective for decreasing recruitment in the next year.

Conclusions

Our findings provide information for the efficient culling of antlerless sika deer on the basis of the population’s density distribution.Although the importance of antlerless-biased culling for population management has been explained (Porter et al.1991;McNulty et al.1997;Matsuda et al.1999;Ueno et al.2010),it has not worked very well to increase antlerless culling because the applicable method has not been established.In addition,because the number of hunters in Japan has declined,a more efficient management scheme is required.We conclude that,if population density is taken into account and professional sharpshooters are employed,antlerless-biased culling of sika deer can become a more efficient management strategy.

AcknowledgementsWe are especially grateful to the Kumamoto prefectural government for providing the cull records and pellet count survey data.We also thank the staff of the Forestry and Forest Products Research Institute for their helpful comments.