Research Article |
Corresponding author: Shinsuke Koike ( koikes@cc.tuat.ac.jp ) Academic editor: Jesus Maldonado
© 2022 Mitsuko Hiruma, Kahoko Tochigi, Ryosuke Kishimoto, Misako Kuroe, Bruna Elisa Trentin, Shinsuke Koike.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Hiruma M, Tochigi K, Kishimoto R, Kuroe M, Trentin BE, Koike S (2022) Long-term stability in the winter diet of the Japanese serow (Artiodactyla, Caprinae). ZooKeys 1122: 39-51. https://doi.org/10.3897/zookeys.1122.76486
|
The winter diets of northern ungulates are sensitive to changes in environmental conditions and ungulate population densities. We hypothesized that the winter diets of smaller browser ungulates might not readily change in response to fluctuating environmental conditions. We analyzed long-term trends in the winter diet of the Japanese serow (Capricornis crispus) by analyzing rumen contents of 532 individuals over a span of 16 years among five populations along with changes in the population densities of sika deer (Cervus nippon) in Nagano Prefecture, central Japan. The winter diet composition of the serow was stable over the long term despite the increase in deer population density. The little-flexible nature of the serow diet may explain the long-term stability in the winter diets.
browser, Capricornis crispus, Cervus nippon, population dynamics, sika deer, ungulate
The winter diets of northern ungulates are sensitive to changes in environmental conditions, and the compositions of their diets undergo change to adapt to the nutritional restrictions of winter (e.g.,
Previous long-term studies on the winter diet of white-tailed deer (Odocoileus virginianus) have shown that changes in vegetation alongside changing population densities and increasing snow depth can influence the winter diet (
The Japanese serow (Capricornis crispus) is an endemic ungulate species in Japan. The Japanese serow was designated as a Special National Treasure in 1955; hunting this animal is illegal. In the early 20th century, the number of serows decreased dramatically due to poaching, but their populations gradually recovered after protective legislation was passed after World War II (
We examined the changes in serow food habits by analyzing long-term (2000–2015) trends in winter food habits in central Japan along with changes in the environmental conditions. Particularly, we focused on the population density of sika deer as an environmental condition. The serow is a browser that mainly feeds on woody plants and is considered to have a narrow food habit, whereas the sika deer is considered an “intermediate feeder” among cervid species with a very flexible food habit (
Nagano Prefecture, Japan (35°11'–37°01'N, 137°19'–138°44'E) (Fig.
The estimated total serow population in Nagano Prefecture was 9340 ± 1630 in 2000 but has hovered around 7738 ± 6420 in 2018 using the block-count method (
Estimated densities (serows / km2 ± SD) of the serow populations in 2000, 2004, and 2014 (
EM | SA | YA | NA | CA | |
---|---|---|---|---|---|
2000 | 2.48 ± 3.43 | 0.57 ± 0.83 | 0.47 ± 1.04 | 1.70 ± 1.99 | 4.16 ± 3.23 |
2004 | 0.51 | 0.33 ± 0.67 | 0.46 ± 0.73 | 0.51 ± 0.61 | 2.40 ± 2.37 |
2014 | 0.83 ± 1.11 | 0.25 ± 0.70 | 0.28 ±0.49 | 0.59 ± 1.00 | 2.22 ± 2.84 |
Since 2000, approximately 10% of the rumen contents has been collected randomly from each culled serow for monitoring purposes (Fig.
We analyzed the reported frequencies as percentages of each food group using the point-frame method (
For data regarding sika deer population density, we referred to the official surveys conducted by Nagano Prefecture (Table
Estimated number and density (median values) of sika deer in each area of Nagano Prefecture in 1999, 2010, and 2015 (
YA | SA | Other | |
---|---|---|---|
1999 | 8,657 (6.2) | 18,858 (11.2) | – |
2010 | 48,527 (20.7) | 33,787 (11.4) | 8,644 (3.9) |
2015 | 128,598 (51.4) | 30,812 (12.7) | 19,795 (5.2) |
The results of an official survey using the block count and fecal pellet count methods in the fall of each year (1999–2015) revealed that there was a gradual long-term increase in the deer population in YA, whereas the densities of the SA population remained stable, and the densities of the NA, CA, and EN populations remained low. Additionally, most areas were not inhabited by deer until the 2000s (Fig.
For analysis, years potentially have both a direct effect through the decrease of food availability and an indirect effect thorough deer density on the winter diet of serows. We employed Bayesian regression models with paths to verify the existence or non-existence of this relationship.
First, we converted the food item frequency of each food group to proportions and modeled them with Dirichlet regression to account for the composition data. Since the number of rumen content counts at cross-sectional points is equal to the discrete level, and the total number of counts for each subsample is summed up to a fixed number, the probability of the occurrence of one group is influenced by increases or decreases in other groups (i.e., multivariate analysis). A disadvantage of this transformation is that the information from the data is changed or lost when skewing the data structure. In contrast, the Dirichlet distribution is appropriate for count-based data summed up to a fixed value.
To build a model for estimating the factors influencing diets of serows, we set the proportion of the six food groups as the response variables. In food groups, we set leaves of deciduous broad-leaved trees as the reference category. We defined year (the years 2000–2015 were converted to years 1–16) and deer density as the explanatory variables. Food composition is possibly different among serow populations due to differences in vegetation and food availability. Therefore, we included the serow population as a random effect. We assumed that response variables follow a Dirichlet distribution.
To build a model for estimating annual changes in deer density, we set the deer density as a response variable and year as an explanatory variable. Density may differ among deer populations due to differences in hunting pressure and food availability. Therefore, we included the serow population as a random effect. We assumed that response variables follow a normal distribution.
To evaluate whether there are significant paths among serow dietary composition, year, and deer density, we checked whether 95% credible intervals (CIs) for estimate values of each variable were greater than zero. We standardized the estimated unstandardized coefficients to compare the impact of factors in which the unit and size differed. We used a Bayesian approach, implemented using the BRMS (Bayesian Regression Model Stan) package (
According to the model analysis results, the 95% CIs of coefficients of year and deer density on all food groups had zero overlap (Table
Model coefficients for six the food plat groups of the serow rumen modeled with Dirichlet distribution for deer density and with normal distribution. Leaves of deciduous broad-leaved trees were set as the reference category. We presented estimates in 95% credible intervals (CIs) for each model parameters. All response variables had a random effect on population. Estimated coefficients were standardized to have zero mean with a standard deviation 1.
Response variables | Parameter | Estimate | SE | Lower CI | Upper CI | Standardized estimate |
---|---|---|---|---|---|---|
Fixed effect | ||||||
Leaves of broad-leaved evergreen trees | Intercept | −1.032 | 0.163 | −1.360 | −0.737 | 0.000 |
Year | 0.019 | 0.014 | −0.008 | 0.046 | 0.640 | |
Deer density | −0.004 | 0.010 | −0.024 | 0.017 | −0.217 | |
Leaves of planted conifers | Intercept | −0.342 | 0.135 | −0.628 | −0.085 | 0.000 |
Year | −0.013 | 0.014 | −0.042 | 0.013 | −0.326 | |
Deer density | 0.008 | 0.010 | −0.011 | 0.028 | 0.317 | |
Leaves of natural conifers trees | Intercept | −0.734 | 0.211 | −1.139 | −0.320 | 0.000 |
Year | 0.024 | 0.014 | −0.003 | 0.051 | 0.559 | |
Deer density | −0.003 | 0.010 | −0.021 | 0.017 | −0.094 | |
Leaves of graminoids | Intercept | −0.312 | 0.243 | −0.792 | 0.210 | 0.000 |
Year | −0.005 | 0.014 | −0.031 | 0.022 | −0.095 | |
Deer density | 0.008 | 0.010 | −0.012 | 0.027 | 0.250 | |
Other food material | Intercept | 0.491 | 0.166 | 0.139 | 0.789 | 0.000 |
Year | 0.013 | 0.012 | −0.011 | 0.038 | 0.246 | |
Deer density | −0.003 | 0.010 | −0.022 | 0.017 | −0.085 | |
Deer density | Intercept | 1.033 | 0.888 | −0.964 | 2.780 | 0.000 |
Year | 0.058 | 0.009 | 0.039 | 0.075 | 0.037 | |
Random effect | ||||||
SD (Leaves of broad-leaved evergreen trees) | Intercept | 0.187 | 0.189 | 0.011 | 0.705 | |
SD (Leaves of planted conifers trees) | Intercept | 0.129 | 0.114 | 0.005 | 0.407 | |
SD (Leaves of natural conifers) | Intercept | 0.317 | 0.208 | 0.083 | 0.873 | |
SD (Leaves of graminoids) | Intercept | 0.421 | 0.277 | 0.134 | 1.200 | |
SD (Other food material) | Intercept | 0.219 | 0.195 | 0.015 | 0.760 | |
SD (Deer density) | Intercept | 1.864 | 0.898 | 0.800 | 4.204 |
Percent composition of the six major food groups in Japanese serow rumen samples of the six populations in Nagano Prefecture, pooled across the populations from 2000 to 2015. The leaves of different plants are depicted as follows. Aqua: deciduous broad-leaved trees, orange: broad-leaved evergreens trees, gray: planted conifers, yellow: natural conifers, blue: graminoids, yellow: other plant foods. The populations are North Alps (NA), Central Alps (CA), Echigo–Nikko–Mikuni (EN), South Alps (SA), Yatsugatake (YA), and pooled across all populations (All).
Directed acyclic graph depicting the year and deer density that are not important in the path analysis of serow food habits. Solid lines: 95% credible intervals (CIs) whose path coefficients do not overlap with zero (see Table
The serows’ winter diet did not readily change in response to increasing deer density. This may be a result of their ecological characteristics, such as small size, solitary nature, and territoriality, which characterized highly selective foraging habits. Therefore, the results support our hypothesis.
It is well known that when deer density rises, vegetation, particularly forest floor vegetation, declines or disappears (
First, spatial partitioning between sika deer and serows may progress. Although there are some regions where the diets of both species overlap, distinctions in habitat use between the two species are also known (
In this study, we could not clearly evaluate the relationship between the increase in deer and the decrease in serows because we did not investigate changes in the vegetation structure and winter food habits of deer and serows’ behavior. However, decrease in serow food supply as a result of increased sika deer density could have resulted in a spatial partitioning between sika deer and serow or decrease in serow density.
We thank K. Edo of the Agency for Culture Affairs, Japan, and T. Miura and K. Kodama of the Japan Wildlife Research Center for providing data regarding serow and sika deer surveys. Additionally, we thank the Nagano Prefectural Government for collecting serow rumen contents with the cooperation of municipal government offices and local hunting associations for providing forestry data. We thank the two reviewers for their valuable suggestions and comments on the earlier versions of this paper.
This work was supported partly by a JSPS Fellows (no. 16H02555) and the Institute of Global Innovation Research in TUAT (2018–2020).