Analysis of the long-term dynamics of ungulates

in Sikhote-Alin Zapovednik, Russian Far East

P.A. Stephens, O.Yu. Zaumyslova, G.D. Hayward and D.G. Miquelle

Collaborators:

Sikhote-Alin State Biosphere Zapovednik

Wildlife Conservation Society

University of Wyoming

USDA Forest Service

Analysis of the long-term dynamics of ungulates in Sikhote-Alin Zapovednik,
Russian Far East

A report to the Sikhote-Alin Zapovednik and USDA Forest Service

Philip A. Stephens*

Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071, USA

Olga Yu. Zaumyslova

Sikhote-Alin State Biosphere Zapovednik, Terney, Terneiski Raion, Primorski Krai, Russia

Gregory D. Hayward

Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071, USA;

USDA Forest Service, Rocky Mountain Region, PO Box 25127, Lakewood, CO 80225, USA

Dale G. Miquelle

Wildlife Conservation Society, Russian Far East Program, Vladivostok, Primorye Krai, Russia

2006

* Present address: Department of Mathematics, University of Bristol, University Walk, Bristol,

BS8 1TW, UK;

Executive Summary

Study and findings

  1. The winter transect count involves monitoring game species by counting tracks of animals that intersect with a stable network of transects, surveyed during periods of snow cover. It is the main method of estimating the number of many game animals in the Russian Federation. For over four decades, this approach has been used consistently to monitor a variety of species in Sikhote-Alin Zapovednik (SAZ), Russian Far East. Hitherto, this extensive data set has not been rigorously analysed to assess trends and ecological relationships in a variety of species, or to assess its potential and limitations with regard to informing management of SAZ. We present such an analysis, focused on six of the larger game species occurring in SAZ: red deer (Cervus elaphus), wild boar (Sus scrofa), roe deer (Capreolus pygargus), musk deer (Mochus moschiferus), sika deer (Cervus nippon) and moose (Alces alces).
  2. The principle objectives of this work were to examine spatial pattern in the occurrence of the species of interest; to investigate methods for estimating population densities from the track encounter data; to assess factors underlying temporal changes in populations of the more abundant species; to analyse the survey protocol and recommend practices whereby it might be improved; and to determine the likely impact of Amur tigers (Panthera tigris altaica) on potential prey species. Through these analyses, we aimed to inform understanding of the distribution and dynamics of the ungulates within SAZ, to aid ongoing efforts to manage the area for the benefit of the endangered Amur tiger, and to integrate the disparate Russian and English language literatures on estimating animal abundance from indirect sign, thereby contributing to this important yet contentious field.
  3. Comparisons of track encounter rates among forest types and drainages suggested few consistent patterns of animal distribution beyond those already recognised by accepted divisions of SAZ into three broad habitat zones (the coastal, oak-birch zone; the central belt, dominated by mixed Korean pine and deciduous forests; and the north-western, higher altitude areas, dominated by spruce and fir forests). Within the oak-birch zone, however, sika deer show pronounced differences in their use of the coastal oak forests and mixed birch and aspen forests further inland. Though less marked, the data suggested that other species may also show differences in their use of these areas. Consequently, we recommend that, in future, the oak-birch zone should be divided into two separate survey units, recognising the existence of four (rather than three) broad habitat zones for survey purposes.
  4. Three methods for estimating ungulate absolute population densities from track counts were compared, including a correction factor based on the relationship between track counts and total counts of deer in experimental plots; the established Formozov-Malyeshev-Perelshin (FMP) formula based on records of animal daily movement distances; and a computationally-intensive simulation approach based on two-dimensional records of animal daily movements. The simulation and FMP approaches gave very similar estimates, supporting the existing belief in Russia that the FMP formula is theoretically sound and generally robust to the different movement patterns of ungulate species. The correction factor tended to overestimate densities but this is unsurprising, given that data used to develop the correction factor came from other study areas, where animal movements may be very different.
  5. We stress that no method for estimating density from indirect sign is robust to violations of underlying assumptions. In particular, no method can fully compensate for biases arising from a survey network that does not adequately represent the area of interest. All methods based on indirect sign also require independent validation, ideally using monitoring based on direct sightings. Specific recommendations for enhancing the validity of the track count surveys are given below.
  6. Two of the methods for estimating ungulate abundance from track encounter rate depend on good data on animal 24-hour movements. These data are currently limited for SAZ but preliminary analyses indicated that movement distances may be affected by time of year and group size (for red deer and roe deer), and a combination of habitat type and time of year (for wild boar). Understanding how travel distances are affected by different conditions is essential for improving the accuracy of density estimation and we urge further collection of these data in a range of conditions.
  7. Differences between ungulate densities in the three major habitat zones of SAZ are pronounced and we assumed that data would always be stratified at this level, at the very least. Finer levels of stratification, including stratification by drainage basin and by forest formation were compared. These different types of stratification seldom had strong effects on estimates. However, analyses indicated that stratification by forest formation could be vulnerable to outliers and, consequently, stratification by drainage basin is recommended. It remains to be seen whether this will be necessary, if a four zone approach to the surveys is adopted (see further below).
  8. Non-parametric bootstrapping was used to derive confidence intervals around estimates of ungulate densities. This method is free from many of the assumptions required by other suggested methods for estimating confidence intervals about estimates derived using the FMP formula. Using non-parametric bootstrapping also avoids the requirement for estimates of parameters such as average group size and average crossing rate for the paths of individual animals, both of which can be very difficult to obtain.
  9. Overall, densities of ungulates tend to be highest in coastal areas and lowest in the spruce-fir, montane forests. Red deer were the most abundant species (1.5 to 3.0 km-2 throughout SAZ), followed by roe deer (1 to 2.5 km-2). At present, sika deer occur only in the oak-dominated forests on the coast but their population appears to be growing rapidly (now exceeding 1 km-2 in that area). Less is known about musk deer daily movements but analyses indicated that this species shows the opposite trend to the other ungulates in SAZ, with the highest densities in montane, spruce-fir forests, and the lowest densities towards the coast. Overall, mean musk deer density throughout SAZ is approximately 1 km-2. Wild boar show substantial fluctuations throughout the coastal and central areas but are at generally low abundance in both, seldom exceeding 0.1 to 0.5 km-2. Finally, moose tracks are encountered too rarely to analyse. That moose track encounters have virtually ceased since 1980, suggests that this species (which is at the southern extent of its range in SAZ) may have shifted northward in response to increasing temperatures.
  10. Analyses of changes in track encounter rates within years suggested that encounters of the tracks of several species (including red deer, roe deer and musk deer) show pronounced declines from early to late winter. Although this results partly from changes in travel distances as winter progresses, it is also possible that species distributions shift throughout winter. Survey routes that accurately represent the entire area of interest are essential if this phenomenon is to be understood (see further below).
  11. In spite of the rigour with which SAZ is surveyed, the track data are prone to census error and resultant estimates of density are noisy. This leads to difficulties in determining the major factors dictating the dynamics of each species. Nevertheless, evidence for density dependent processes was found in several populations. Additionally, climate, competition, quality of mast crops and protection from poaching all influence the studied populations. Dynamics of the red deer and sika deer populations are currently best understood. There is evidence for competition between these species and, also, for climate effects acting in different directions. In particular, increasing temperatures appear to have a positive effect on the sika deer population but a negative effect on red deer populations in the coastal and central zones. By contrast, red deer in the spruce-fir zone are positively affected by increasing temperatures, suggesting that the species may be shifting its distribution northwards and to higher altitudes as mean temperatures increase.
  12. Assessments of the survey protocol and of the relationship between survey effort (kilometres of transects conducted per year) and precision, emphasised two major points. First, that survey design depends critically on the goals of monitoring, in particular, whether relative indices of abundance are sufficient or absolute estimates of abundance are desired, and whether it is necessary to detect trends in animal abundance. If the detection of trends is a goal of the monitoring, then it is important to establish what magnitude of trend should be detectable and over what time period. Secondly, analyses also highlighted the fact that, for a given set of goals, required survey effort is affected by the density, movement behaviour and grouping behaviour of the species considered. Low densities, short daily travel distances and clumped distributions all increase the uncertainty in abundance indices. Thus, some species will be harder to monitor as accurately as others. Different monitoring goals will be relevant to different species (see further below).
  13. Simple analyses of the likely effects of tigers on prey populations indicated that these are likely to be small relative to the estimated effects of other large carnivores on their prey. This accords with findings from the temporal analyses, which showed no evidence for a strong effect of tigers on prey populations. Although the social intolerance of tigers may play a role in limiting their local density and, hence, their effects on prey, this is unlikely to be important at the relatively low prey densities in SAZ. More likely, the generally low impact of tigers on prey results from their relatively low energetic requirements when compared to many other large carnivores.

Specific recommendations

This study highlighted a variety of improvements that could be made to the monitoring work conducted in SAZ. Ultimately, the goals of Zapovednik monitoring are for the managers of the Zapovednik system to designate. However, some suggested goals and recommendations include:

  1. Use a four-zone classification of SAZ for ungulate monitoring. This is discussed in more detail in the report but the zones would include the coastal zone (dominated by oak forests), the inner-coastal zone (dominated by birch-aspen forests), the central zone (dominated by Korean pine-deciduous forests) and the montane zone (dominated by spruce-fir forests).
  2. Define an overall goal for monitoring ungulates. This should specify whether monitoring should produce only an index of relative abundance, or estimates of absolute abundance also. It should also specify the units of interest (both species and zones) and whether trend detection is important. If trend detection is important, the magnitude of trends and the time periods over which these should be detected must also be defined. We suggest that the goal be defined as follows: Ungulate monitoring in SAZ will provide estimates of the absolute abundance in winter of ungulates in the four major habitat zones. At least 1000 km of surveys will be conducted annually, distributed equally over the four zones. The aim of this will be to give the maximum power to detect trends in numbers of the more abundant species in the habitats most important to that species. A design capable of detecting a 15% annual decline after 5 years of monitoring will be achievable for most species.
  3. Recognise limitations and adapt to priorities and changing conditions. It is vital that the limitations of the monitoring be recognised including, in particular, that density estimates are associated with considerable uncertainty, and that species at lower abundance, with shorter daily travel distances and with highly clumped behaviour will be subject to greater uncertainty, such that trends are harder to detect with confidence. The monitoring protocol should also be adaptable to changing priorities and to changes in conditions (such as increasing or decreasing densities of certain species).
  4. Validate the relationships between track counts and density estimates. Independent estimates of density must be generated using alternative methods, in order to indicate how accurately density is estimated by current methodologies. In particular, we recommend the use of distance samplingand aerial surveys (combined with sightability models) as potential methods for validating the track count index.
  5. Assess bias in transect network. Assess bias in the transect network using GIS analyses and by comparing results of randomly placed transects to the existing network within a number of basins of the reserve. If a significant bias is detected, there are two alternatives to address this bias: (i) if the bias is stable and predictable across all areas and all conditions, apply a simple correction factor; (ii) if the bias is not stable and is difficult or impossible to predict, relocate transects to approximate a random sampling effort.
  6. Improve data base on daily travel distances. Daily travel distances must be collected during the time frame in which surveys are conducted, as there is evidence that travel distance drops in late winter. Data on travel distances must also be collected across the range of environmental parameters that are likely to affect movements. Our initial analyses suggest that group size, time of year and habitat type are the primary drivers of daily travel distance. Collecting data over the full range of possible values for each of these parameters will be important in deriving appropriate estimates of travel distance for ungulates within the Zapovednik.
  7. Collect data on the numbers of animals that made each set of tracks encountered. To collect data not only on the number of sets of tracks of each species encountered on transects but, also, on the number of these that were made by single animals or groups of various sizes, is likely to be awkward, especially from the point of view of data storage. Nonetheless, our analyses showed that size of the travelling group may be important in dictating the travel distance of some species. Consequently, collecting such data will be helpful for improving the accuracy of density estimates. The data could also be useful for determining group size distributions, which will have important implications for error calculations and other aspects of understanding demography of the studied species.
  8. Eliminate the recording of “nabrods”, or multiple, uncountable crossings. Eliminate records of “nabrod” in SABZ dataset by training all observers to circle nabrods and report actual numbers of tracks to the best of their ability.

Acknowledgements

This monograph is the result of a long-term collaborative effort between Sikhote-Alin Zapovednik and the Wildlife Conservation Society. We thank A.A. Astafiev, Director of Sikhote-Alin, for continual support of these mutually beneficial, ongoing efforts. We also thank M. Hornocker and H. Quigley, who had the wisdom and courage to initiate the Siberian Tiger Project, and select Sikhote-Alin Zapovednik as its base. M.N. Gromyko, and L.V. Potikha have both acted as Assistant Directors of Science for Sikhote-Alin Zapovednik and facilitated our collaborative efforts. E.N. Smirnov, our scientific collaborator for the Amur Tiger Project, has been instrumental in all phases of the work. A.E. Myslenkov provided much data on daily travel distances of ungulates, a pivotal part of the database which is used here. T. Merrill provided the first GIS database training for Zapovednik personnel, and helped design and implement database development. We thank other members of the Zapovednik staff for their help and advice including, especially, Luba Khubotnova, whose assistance with translation during meetings was invaluable.

This work was funded by the U.S. Forest Service, International Programs, part of the U.S. Department of Agriculture, to whom we are most grateful. In particular, we would like to thank Liz Mayhew, Lara Peterson and Jen Peterson for all of their support and logistical advice throughout the project.

The bulk of the analytical work was conducted at the University of Wyoming, USA, and we thank the University for support. In particular, we are grateful to heads of the Zoology and Physiology Department, N. Stanton and G. Mitchell, as well as S.D. Hutton in International Student Services.

Translations were completed by E. Nikolaeva, A. Murzin, and D. Karp. G. Contraras facilitated the translation of Russian scientific articles into English. Many others provided advice on research approaches, translation of materials and statistical techniques. In particular, we thank the following: C. Nations, C. Martínez del Rio, S. Buskirk, R. Freckleton, R. King, A. Cardinali, C. Anderson, J. Crait and K. Gerow.

Finally, PAS and GDH would like to thank all those in Terney and Vladivostok who made their visits so enjoyable and useful. Many of those are already listed but, additionally, we thank John Goodrich, Marina Miquelle, Galia Safanov, Zheny Gishko, Kolya Reebin, Sasha Reebin, and Volodia Melnikov.