l ; Err/5

100? L I—o'Ir—o‘u Ll IRYI“
[\L'ILECI 3:33 Stale

.. University

 

This is to certify that the
thesis entitled

RODENTICIDE USE AND SECONDARY POISONING RISKS
TO NON-TARGET WILDLIFE IN CENTRAL MONGOLIA

presented by
ANN M. WINTERS

has been accepted towards fulfillment
of the requirements for the

MS. degree in Fisheries and Wildlife

Major Professor's Signature

74’1441/ 41 2066
7 ,

Date

 

 

 

MSU is an Affimvative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
i To AVOID FINES return on or before date due.
; MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DateDue.indd

RODENTICIDE USE AND SECONDARY POISONING RISKS
TO NON-TARGET WILDLIFE IN CENTRAL MONGOLIA

By

Ann M. Winters

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Fisheries and Wildlife

2006

ABSTRACT

RODENTICIDE USE AND SECONDARY POISONING RISKS
TO NON-TARGET WILDLIFE IN CENTRAL MONGOLIA

By
Ann M. Winters
In 2001, hundreds of non-target wildlife deaths occurred in the steppe region of
Mongolia after aerial applications of bromadiolone, a second generation anticoagulant,
were carried out to control eruptive Brandt’s vole (Microtus brandti) populations in areas
used for livestock grazing. A two-year field study was initiated in 2004 to investigate
bromadiolone application policies and practices and quantify the secondary poisoning
risk that bromadiolone use may pose to non-target wildlife in Mongolia. The study found

that carcasses of voles killed by bromadiolone control are commonly available to

scavengers; a mean of 2.64 carcasses/ 100 m2 (a: 0.47 SE) were found on bromadiolone-

treated field plots. Digital trip-cameras were used to investigate which wildlife species
are likely to scavenge non-toxic vole carcasses; five birds and five mammals were
recorded doing so. High-performance liquid chromatography assays of tissues from live
and dead voles exposed to either 0.05% or 0.005% bromadiolone-treated wheat grain
revealed that these voles contain relatively small amounts (means 11.8 pg i 3.8 SE and
2.3 pg :1: 0.5 SE, respectively) of bromadiolone. It is unlikely that non-target animals
could consume enough of these toxic voles to kill 50% of the population (i.e. to reach an
LDso dose). However, in situations where excessive bromadiolone concentrations are
used, or where the toxicant is reapplied frequently, there may be a significant risk to

some of the valued wild animals present on Mongolia’s rangelands.

Cepyright by
ANN M. WINTERS
2006

For my parents,
John L. Winters & Susan L. Winters
Without the technical know-how and practicality I inherited from Dad, the adventurous

spirit I inherited from Mom, and the perseverance I inherited fi'om both of you, I would
not have succeeded in this challenging endeavor.

iv

ACKNOWLEDGMENTS

I thank my committee: Dr. Graham Hickling, Dr. Scott Winterstein, Dr. Wilson
Rumbeiha and Dr. Amanda Fine for their advice and support. Dr. Hickling, my major
adviser, offered a great deal of guidance and support throughout my master’s program,
and was always ready to work closely together. Dr. Winterstein gave me many uplifiing
pep-talks over the past few years, and helped me work through my final analyses. Dr.
Rumbeiha patiently explained different aspects of toxicology to me and facilitated the
sample analysis. Dr. Fine offered advice on the project’s sample design and the
feasibility of conducting research in Mongolia, not to mention one huge reality check
when it was time to write. I must also thank Dr. Jean Tsao, as a virtual member of my
committee, since she often gave me sound advice during my time here at MSU, and was
always ready to lend an ear. I could not have asked for better mentors to guide me
through my master’s research and professional development over the past few years.

This research was facilitated by Bariushaa Munkhtsog, executive director of
IRBIS Enterprises, Mongolia Program Director of the International Snow Leopard Trust,
and co-founder of the Pallas’ Cat Project. I thank the Pallas’ Cat Project, especially Dr.
Meredith Brown, for providing me with the opportunity to conduct this research and for
collaboration and support in Mongolia. I also thank the Plant Protection and Research
Institute of Mongolia, especially director Dr. B. Battur and Otgonjargal, for assistance in
exporting samples from Mongolia and for collaboration during my field work. Thank
you to those individuals who provided me with information on rodenticide use in
Mongolia; L. Dolgormaa, World Wildlife Fund Mongolia Programme Office, and B.

Mendjargal, Mongolia Ministry of Food and Agriculture.

This research would not have been successful without the support of Jarnsuren
and his family in Altanbulag, Tov aimag. I specifically thank Boldbaatar, Gantulag,
Naranbaatar, Batbold, Tumursukh, Doljin, Hishgee, Davaadulam (Sima), Amarjargal,
Tagta, Urantsetseg, Steve Ross and Ruth Kramnitzer. My warmest thanks go to my
technician, J. Batzorig, and driver, Enkhtuvshin. Their help and enthusiasm during my
2005 field season made the summer both productive and enjoyable. Tuvshin, I miss you
clearly.

I thank Jean Gaymer, University Laboratory Animal Resources, and Michelle
Bennett, Diagnostic Center for Population and Animal Health, Toxicology, for their help
with sample collection training and analysis. I also thank Tom Cooley, Michigan
Department of Natural Resources, for kindly allowing me to work in his lab.

I also thank the Department of Fisheries and Wildlife faculty and staff who often
helped me with travel preparations and other odd tasks, and were always willing to
entertain my numerous questions. The Department of Fisheries and Wildlife offers great
opportunities to graduate students, and I feel fortunate to have been one. Thank you for
this excellent opportunity. Thank you to my fellow graduate students for their advice and
camaraderie, especially my cube-mate, Dave Close, whose sense of humor and advice I
will always appreciate. Finally, I thank my family and Brad Priebe for their never-ending
support.

This research was supported by the Department of Fisheries and Wildlife, USAID
SANREM CRSP, Pallas’ Cat Project, Cincinnati Zoo and Botanical Garden, Graduate

School at Michigan State University, and College of Agriculture and Natural Resources.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES ........................................................................................................... xi
CHAPTER 1
INTRODUCTION ............................................................................................................... 1
Background to the Problem ................................................................................................. 2
Study Objectives .................................................................................................................. 4
CHAPTER 2
BACKGROUND AND LITERATURE REVIEW ............................................................. 7
Introduction .......................................................................................................................... 7
Rodent and Lagomorph Pests: International Perspective .................................................... 7
Rodent and Lagomorph Pests: China, Tibet and Mongolia ................................................. 8
Anticoagulant Rodenticide Control Worldwide ................................................................ ll
Anticoagulant Rodenticide Control: China, Tibet and Mongolia ...................................... 17
China and Tibet ...................................................................................................... 17
Mongolia ................................................................................................................ 17
Current perceptions .................................................................................... 22
Current and future use of rodenticides in Mongolia .................................. 24
CHAPTER 3
FIELD ASSESSMENTS OF THE RISK OF SECONDARY POISONING .................... 26
Introduction ........................................................................................................................ 26
Government Rodent Control: Field Observations ............................................................. 26
Field Trials ......................................................................................................................... 27
Study site and environmental data ......................................................................... 27
Materials and Methods: Vole Bromadiolone Trials ............................................... 30
Study site .................................................................................................... 30
Selection of experimental plots .................................................................. 3O
Trapping and sampling .............................................................................. 3O
Carcass monitoring .................................................................................... 32
Beetle trapping trial .................................................................................... 33
Bromadiolone HPLC analysis .................................................................... 34
HPLC data analysis .................................................................................... 35
Materials and Methods: Scavenger Identification ................................................. 36
Camera trapping ......................................................................................... 36
Image analysis ............................................................................................ 39
Authorizations ........................................................................................................ 40
Results: Vole Bromadiolone Trials ........................................................................ 40
Wheat grain counts .................................................................................... 40
Amount of bromadiolone in voles ............................................................. 41
Carcass monitoring .................................................................................... 47

vii

Results: Scavenger Identification .......................................................................... 50

Camera trapping vs. vehicle surveys ......................................................... 53
Discussion .............................................................................................................. 55
Implications of vole bromadiolone results ................................................. 55
Implications of scavenger identification results ........................................ 59
Risk of secondary poisoning ...................................................................... 61
Conclusions ................................................................................................ 63

CHAPTER 4
OVERALL DISCUSSION AND CONCLUSIONS .......................................................... 65
Introduction ........................................................................................................................ 65
Concern over non-target poisoning .................................................................................... 65
Implications of the field trials ............................................................................................ 66
Exposure factors ................................................................................................................. 68
Further Research ................................................................................................................ 69
Overall Recommendations ................................................................................................. 70
Overgrazing and vole outbreaks ............................................................................ 70
Risk/benefit analyses and alternatives ................................................................... 71
In-use monitoring ................................................................................................... 71
APPENDIX A .................................................................................................................... 73
APPENDIX B .................................................................................................................... 75
LITERATURE CITED ...................................................................................................... 79

viii

Table 2—1.

Table 2-2.

Table 3-1.

Table 3-2.

Table 3-3.

Table 3-4.

Table 3-5.

Table 3-6.

LIST OF TABLES

Wildlife species found dead in areas treated with bromadiolone-
impregnated wheat grain fi'om suspected secondary poisoning in 2002 in
central Mongolia (Tseveenmyadag and Batbayar 2003). .......................... 20

Total hectares (x1000) treated with bromadiolone in 11 provinces of
Mongolia between 2002 and 2004. Total hectares treated are included for
each year. Information provided courtesy of the Mongolian Ministry of
Food and Agriculture, 2005. ...................................................................... 21

Carcass scoring criteria for vole carcasses that were tethered in view of
camera trap-stations (CTSs) at 16 sites from June 12 — July 17, 2005, in a
study area near Altanbulag, Tov aimag, Mongolia .................................... 38

Behavior scoring criteria for wildlife species that visited camera trapping
stations (CTS) from June 12 — July 17, 2005, in the study site near
Altanbulag, Tov aimag, Mongolia. ............................................................ 39

Decline over time in the numbers of wheat grains per m2 in plots
broadcast with two concentrations of bromadiolone-treated grain at the
study site near Altanbulag, Tov aimag, Mongolia. .................................... 40

Mean (i SE) bromadiolone concentrations (pg/g) and amounts (pg) in
liver, stomach contents, body samples, and total burdens — calculated
using measured concentrations and sample weights - for voles sampled
from plots treated with wheat grain impregnated with two concentrations
of bromadiolone. Voles were trapped and sampled in a study area near
Altanbulag soum, Tov aimag, Mongolia. .................................................. 42

Mean (2|: SE) bromadiolone concentrations (pg/g) in livers, and total body
burdens (pg) in live-trapped voles and vole carcasses found on the ground
during the field study (Treatments A and B are combined) ....................... 47

Wildlife species that visited camera trapping stations (CTSs) at 16
locations from June 12 — July17, 2005 during 4,704 hours of camera
monitoring, in a study area 60 km south-west of Altanbulag, Tov aimag,
Mongolia. Birds and mammals are tallied separately and ranked by the
number of times they scavenged, from most to least. Data listed include
the number of day (06:01 to 21 :29) and night (21 :30 to 06:00) visits, total
number of visits, mean visitation rate per 100 hrs of camera trapping, the
number of times scavenging behavior was observed, percentage of visits
in which species exhibited scavenging behavior, and the mean scavenging
rate per 100 hrs of camera trapping for each species recorded. ................. 52

ix

Table 3-7.

Table 3-8.

Comparison of night visitation rates for the camera trapping study and
night sightings during vehicle transects*, of 12 mammal species in the
study area near Altanbulag, Tov aimag, Mongolia .................................... 54

Comparison of estimated maximum tolerated dose (MTD) values for two
species, the total amount of bromadiolone required for an MTD in each
species given an average weight, and the estimated number of voles these
species would have to consume to reach their MTD. Two estimates are
given, one using the mean (11.7 pg i3.8 SE) bromadiolone content of
voles from high-dosed (A = 0.05%) plots, and one using the maximum
bromadiolone content of a vole encountered during the study (151.4 pg).62

LIST OF FIGURES

Figure 2-1. The chemical structure of bromadiolone (WHO/FAQ 1996). ................... 17

Figure 3-1. A map of the study area located 65 km south-west of Altanbulag, Tov
aimag, Mongolia, depicting a) the study area’s location in Mongolia (red
dot near Ulaanbaatar); b) an enlarged map of the island where 15 vole
sampling plots were established; and c) the entire study area with locations

of the 16 camera-trapping stations, and the study island in relation to Base
Camp and the Tul River. ............................................................................ 29

Figure 3-2. A beetle trap constructed over a partly buried water bottle, half firll
of water, with a vole carcass tethered over the hole to attract
invertebrate scavengers. ............................................................................. 34

Figure 3-3. Photographs of vole carcasses in three stages of decomposition/
absence; far left: score = 3, middle: score = 4, far right: score = 5. .......... 38

Figure 3-4. Box and whisker plot of bromadiolone concentrations (ppm) in the
livers of voles collected from high-dose (A = 0.05%) and low-dose
(B = 0.005%) research plots. One outlying value (24.37 ppm) has
been truncated to 10 ppm in treatment A. .................................................... 43

Figure 3-5. Frequency distribution of bromadiolone concentrations in the livers of
voles exposed to a) Treatment A — 0.05% bromadiolone-treated wheat
grain, and b) Treatment B — 0.005% bromadiolone-treated wheat grain.
Voles were trapped 2 and 4 days post-exposure fiom fenced plots in a
study area near Altanbulag soum, Tov aimag, Mongolia. ........................... 44

Figure 3-6. Linear regression of liver and body bromadiolone amounts (pg),
calculated using measured concentrations and sample weights, for
eight voles sampled fiom plots treated with bromadiolone-impregnated
wheat grain in Altanbulag soum. ................................................................. 46

Figure 3-7. Number of carcasses present at 12 hour intervals after voles were

exposed to bromadiolone-treated wheat grain in six 100 m2 plots
located near Altanbulag, Tov aimag, Mongolia. ......................................... 48

Figure 3-8. Mean decomposition/absence (scored as follows: 0 = dying vole, 1 =
fresh carcass, 5 = carcass absent) of 18 vole carcasses over a 7 day
period after bromadiolone-treated wheat grain had been spread in six

100 m2 plots, located near Altanbulag, Tov aimag, Mongolia. The
dotted line corresponds to absence of the carcass ........................................ 49

xi

CHAPTER 1

INTRODUCTION

Non-target wildlife deaths often occur when toxic compounds are used in open
areas to control outbreaks of rodent pests. Such deaths can result from direct
consumption of toxic bait, or through ‘secondary poisoning’ when a predator or
scavenger consumes tissues of poisoned target species. The risk of secondary poisoning
is recognized as a growing threat to valued wildlife populations (Poche 1988, Shore et al.
1996, Stone et al. 1999, Fournier-Chambrillon 2004), and reports of wildlife
contamination and toxicoses after the use of second-generation anticoagulants (SGAs) are
increasing worldwide (Eason et al. 2001). The impact of this collateral damage may
outweigh the benefits of SGA use. Wildlife deaths occurred recently in Mongolia as a
result of secondary poisoning after an SGA was aerially spread over large areas.

Instances of suspected secondary poisoning were reported in Mongolia in 2002
after SGA baits were spread over large areas of Steppe grassland. During the fall of
2001, the Mongolian government began treating areas with bromadiolone-laced bait to
control outbreaks of Brandt’s voles (Microtus brandti), a rodent pest. Bromadiolone is
an anticoagulant (AC) compound that acts in the liver to inhibit normal synthesis of
vitamin K-dependent clotting factors, eventually leading to hemorrhaging in the animal
(Eason et al. 2002).

In the spring and summer of 2002, following this poison campaign, Mongolian
biologists became alarmed when they found numerous non-target birds and mammals

dead or dying from suspected anticoagulant poisoning in the treated areas

(Tseveenmyadag and Batbayar, 2003). Carcasses of these animals showed evidence of
hemorrhaging in the neck and mouth, indicating likely anticoagulant poisoning from
primary or secondary poisoning. Among the dead predators and scavengers were two
Saker falcons (F alco cherrug), an endangered species throughout its range; two Corsac
foxes (Vulpes corsac); and two Pallas’ cats (Otocolobus manul), a near-threatened
species. Since it is unlikely that these species would have directly ingested treated wheat
grains, these deaths are thought to have been caused by secondary poisoning that resulted
from consumption of poisoned prey items.
Background to the Problem

Throughout Mongolia’s Steppe, Brandt’s voles invade overgrazed pastures and
further contribute to pasture degradation, as well as compete with livestock for forage.
Consequently, the government attempts to control vole outbreaks through large-scale
aerial applications of bromadiolone baits. Studies in Inner Mongolia have documented a
relationship between decreased vegetation height, either from overgrazing or low rainfall,
and Brandt’s vole outbreaks (Zhang et al. 2003, Zhong et al. 1999). These voles are one
example of the pests that cause major destruction of stored food and agricultural crops
(Byers et al. 1976, Jackson 1977, Marsh et al. 1980, Lund 1994), transmit diseases
(Nelson 1980, Gratz 1994), cause erosion and pasture degradation (Lund 1994, Wood
1994, Fan et al. 1999), and contribute to the decline of threatened and endangered
wildlife populations in many parts of the world (Howland et al. 1999, Alterio and Moller
2000, Eason et al. 2002). The economic and environmental damage that rodent pests
cause has lead to control efforts that utilize anticoagulants (AC3) to control pest

outbreaks.

Second-generation anticoagulants (SGAs) were developed in response to strains
of rats and mice that became resistant to first-generation anticoagulants (Marsh et al.
1980, Jackson and Ashton 1992, Buckle 1994). These compounds are effective against
first-generation AC-resistant strains of rodents because of their distinct molecular
structure. The most widely known compounds in the SGA group are difenacoum,
bromadiolone (Maki®), and brodifacoum (Talon®). A single dose of SGA bait is usually
lethal, although rodents may continue feeding on the bait for several days afier ingesting
the initial lethal dose, with death fi'om hemorrhaging resulting five or more days later
(Jackson and Ashton 1992). Thus, SGAs are more potent than first-generation ACs and
can be used as ‘single-application’ rodenticides.

When SGAs are used to control pest populations, the compounds can bio-
accumulate in, or be relayed to, natural predators and scavengers of the target species
(Greaves 1994, Eason and Spurr 1995, Fournier-Chambrillon 2004). This is especially
true for extensive control programs that expose predators and scavengers to poisoned
prey for several days (Mendenhall and Pank 1980, Godfrey 1985). These toxicants have
much longer half-lives in tissues because they have a greater binding affinity in the liver
than first-generation ACs (Parmar et al. 1987, Poche 1988, Jackson and Ashton 1992,
Eason and Spurr 1995, Erickson and Urban 2004, Harrell et al. 2003). Consequently,
these toxicants can accumulate in predators and scavengers that feed on poisoned prey.
Accumulation of SGAs can eventually lead to death, or to a sub-lethal burden that
increases an animal’s vulnerability to other causes of death (Townsend et al. 1981,

Fournier-Chambrillon et al. 2004, Brakes and Smith 2005).

Accidental secondary poisoning of non-target animals fiom SGAs has been
reported in many parts of the world. Polecats were found dead with signs of secondary
SGA poisoning in Britain (Shore et al. 1996), and subsequent surveys recorded the
secondary poisoning of various small mammal predators (Shore et al. 1999). Secondary
poisoning of at least five bird species, including weka (Gallirallus australis), a threatened
scavenger species, occurred in New Zealand during brodifacoum applications aimed at
controlling introduced predator populations (Eason and Spurr 1995). In New York,
researchers documented various cases of non-target wildlife poisoning, most of which
had SGA residues (Stone et al. 1999). Shawyer (1987) reported a ‘mass mortality’ of
various raptors and red foxes after a Hampshire farm was treated with brodifacoum baits
in 1981. In France, researchers conducted a survey to evaluate the secondary poisoning
hazard of SGAs to non-target species (Bemy et al. 1997) and found evidence of
secondary poisoning in foxes and buzzards. One of the most common SGAs identified in
these poisonings was bromadiolone.

Instances of secondary poisoning from bromadiolone were documented in
Switzerland after a rodent control campaign in which numerous raptors and several
predatory mammals were found dead (Beguin 1983 and Pedroli 1983 cited in Shawyer
1987). This compound has been used extensively for rodent control in France and other
countries in Europe (Bemey er al. 1997, Morin et a1. 1990, Shore 1999) and was
introduced to the US. market in 1980. However, it is not registered for field use.

Study Objectives
The conservation community in Mongolia remains concerned about the

consequences of having a persistent toxicant such as bromadiolone in the environment

(WWF, WCS and ESBP 2004). In particular, the Pallas’ Cat Conservation Project
(PCCP; http://www.fw.msu.edu/Pallas%20Cat%20Proiect/index.htm) sought to
investigate the possible risk bromadiolone use poses to the Pallas’ cat population in
Mongolia. In response to these concerns, the PCCP sought advice and collaboration from
researchers at Michigan State University’s Department of Fisheries and Wildlife. This
research project was subsequently developed, with the broad aim of investigating the
secondary poisoning risk that large-scale bromadiolone applications pose to non-target
predator and scavenger wildlife species in Central Mongolia. The specific objectives of
the project were to:
1) Review bromadiolone application policies and practices from 2001 to 2005 in
Mongolia.
2) Quantify the secondary poisoning risk to wildlife by:
a. Determining if carcasses of poisoned rodents are available to scavengers
in areas treated with bromadiolone.
b. Determining which native wildlife species are most likely to scavenge
rodent carcasses.
c. Quantifying the amount of bromadiolone in live voles and vole carcasses
after poisoning campaigns.
The following Chapters present the results of this work. Chapter 2 summarizes
background information on pest problems and anticoagulant use world-wide and in
Mongolia. Chapter 3 summarizes the field methods and results, concluding with a

discussion on the implications of these findings (this chapter is structured as a stand-

alone manuscript for submission to a scientific journal). Chapter 4 discusses the results

of the study in the broader context of sustainable ecosystem management in Mongolia.

CHAPTER 2

BACKGROUND AND LITERATURE REVIEW

Introduction

The following chapter begins with a literature review concentrating on pest
damage and control strategies in several countries, with a focus on China, Tibet, and,
most specifically, Mongolia. Instances of secondary poisoning are also reviewed, with
particular attention accorded to the mass mortalities of wildlife species after initial
applications of bromadiolone in Mongolia. The chapter concludes with a detailed
overview of the bromadiolone application policies and practices in Mongolia from 2001
to 2005, and current and future use of bromadiolone.
Rodent and lagomorph pests: International perspective

World-wide, significant amounts of food resources are destroyed or contaminated
annually by vertebrate pests in the field and during storage (Jackson 1997). Rodent pests
often infest private dwellings, food-manufacturing plants and public institutions as well
as agricultural crops and pasture lands (Wood 1994). Estimates of damage to standing
crops by rodent pests have ranged from 9 to 15% in Pakistan and Australia, respectively,
during years of serious outbreaks (Lund 1994). In addition to contaminating and
damaging food sources, rodents often harbor and transmit disease. For example,
Spirillum minus and Streptobacillus mom'liformis are two disease-causing bacteria that
can be transmitted by rat bites in the United States and countries in south-east Asia (Lund

1994). Infectious agents such as Yersinz'a pestis can be maintained in reservoir

populations of sylvatic (wild) rodent species for long periods of time until contact with
commensal species, which can then vector the disease to humans (Gratz 1994).

Eruptive pest populations are a serious problem on several continents. Half the
food produced in India has been destroyed in some years by a combination of rats, rot,
birds and insects (Proctor 1994). The loss in potential production of crops, livestock, and
forests is estimated at 30% annually, with 50% losses common in developing nations
(Jackson 1977). In eastern and southern Australia, outbreaks of house mice (Mus
domesticus) occur periodically throughout the cereal production areas, causing high
economic losses to grain-growers. Researchers have linked these outbreaks to years with
above-average crop production, which in turn is highly correlated with winter-spring
rainfall (Pech et al. 1999). Extensive use of chemicals to control outbreaks in Australia
has resulted in serious environmental problems (Pech et al. 2002).

In Europe, voles cause serious damage annually by taking grain from fields,
girdling young trees, and destroying root systems in orchards. Damage to forests in
Germany reached 50% on more than 1000 ha in 1998 and 1999, with an estimated cost of
10.5 million Euros (Pelz 2002). In the Central-Eastern United States, Pine voles
(Microtus pinetorum) girdle apple trees during the winter by removing bark from around
the trunks of trees along trail systems the voles construct under the snow during the
winter months (Byers and Young 1976).

Rodent and lagomorph pests: China, Tibet and Mongolia

Rodent and lagomorph pest problems in China, Tibet and Mongolia share many

similarities, and are therefore considered together. On the Qinghai-Tibet Plateau, plateau

pikas (Ochotona curzom’ae) and zokors (Myospalax baileyi) compete with livestock for

food, reportedly consuming about 150 million tons of fresh grass every year (Fan et al.
1999). They contribute to soil erosion and firrther damage pastures by uprooting and
destroying vegetation through their fossorial lifestyle (Zhong et al. 1999). These
activities make areas vulnerable to more serious erosion when seasonal high winds
remove large amounts of the top soil.

Brandt’s voles (Microtus brandti) are another pest species that causes major
destruction of pasturelands in parts of northeast Asia (Zhong et al. 1999, Sarnjaa et al.
2000, Peck et a1. 2002). Over the period 1950 — 1996, 15 different outbreaks of these
rodent pests were documented at four sites in Inner Mongolia (Zhang et al. 2003).
Rodent pests infest 10 to 20% of China’s grasslands every year, with an annual loss of
vegetation biomass estimated at 40 — 50 million tons (Zhang 2002). These rodent
populations are reservoirs of various zoonoses, including bubonic plague, and thus
present a serious threat to local human populations. The damage Brandt’s voles inflict on
pasturelands, and their contribution to plague outbreaks, has led to control programs in
China that utilize various un-named rodenticides.

An eruptive Brandt’s vole population has been of great concern in Mongolia for
the past 50 years. Voles move into overgrazed pasture and compete with livestock for
forage. It is estimated that Brandt’s voles inhabit 40 million hectares of Mongolian
grassland, extending fi'om the Khan-Khokhii Mountain range to the eastern Mongolian
steppe (Tseveenmyadag and Batbayar, 2003). They have a reproductive rate of 3-5 litters
per year, producing between 21 and 25 offspring. Thus, they can occupy grazing areas
very quickly, where they often have a devastating impact on the pasture. Mongolian

scientists estimate that during infestations, voles damage two-thirds of pasturelands, and

destroy 90-95% of all infested pastures (Tseveenmyadag and Batbayar, 2003). The
extensive devastation caused by vole outbreaks has led to the implementation of wide-
scale poisoning efforts by the Mongolian government.

Recent studies in Qinghai-Tibet and northern China show that pika and Brandt’s
vole densities increase with decreasing plant height (Fan et al. 1999, Zhong et al. 1999).
As grassland becomes degraded through overgrazing, the biomass of preferred forge of
the Daurian pika (Ochotona daurica) increases. When vegetation cover is further
reduced through continued grazing pressure, the vegetation species composition changes
and the habitat becomes unsuitable for the pika; its abundance decreases, and Brandt’s
voles begin to dominate (Zhong et al. 1999). This precipitates a Brandt’s vole outbreak
where densities can reach 1,300 voles per hectare. Erosion then becomes an added
degradation problem, coupled with the invasion of undesirable plant species, when voles
construct burrowing systems and loosen the soil.

Environments with short, sparse grass are the preferred habitat of Brandt’s voles
(Zhong et al. 1999). Hence, when livestock graze pastures heavily and reduce the grass
height to below 130 mm, the ideal habitat created attracts Brandt’s voles and conditions
become favorable for an outbreak. However, when vegetation is above 190 mm high,
voles are found in relatively low densities because high or dense grass disrupts social
interactions and interferes with detection of predators (Zhong et al. 1999). This
relationship has many repercussions for grazing practices in Mongolia, indicating the
need for intensive management of overgrazed pastures to increase the vegetation height,

thereby producing sub-optimal habitat for Brandt’s voles.

10

When there is ample precipitation for high plant biomass, or when flooding from
snowmelt or summer rain causes high mortalities in burrow systems, circumstances
become unfavorable for vole outbreaks (Zhang 2003). Years with low precipitation are
thought to correspond to vole outbreaks because grasses remain shorter, enabling voles to
visually communicate with each other and thereby avoid potential predators. Researchers
in Inner Mongolia are developing a comprehensive model to predict Brandt’s vole
outbreaks based on the Southern Oscillation Index (Pyper 2001, Zhang et al. 2003).
However, existing models tend to predict more outbreaks than actually occur, so
additional factors that have not yet been identified may be preventing outbreaks. It is
difficult to identify all of the factors that influence outbreaks and, eventually utilize those
that prevent them.

Anticoagulant rodenticide control worldwide

A variety of toxic chemicals have been used to control eruptive rodent
populations throughout history. Traditionally, compounds with acute effects such as
zinc phosphide, sodium monofluoroacetate, and strychnine were used to control rodent
pests (Kaukeinen 1982, Jackson and Ashton 1992, Buckle 1994). These compounds
posed significant disadvantages such as the need for relatively large concentrations (> 2%
of the active compound) to cause sufficient mortalities, and their acute action, which left
little time for treatment when baits were accidentally ingested by non-target animals
(Jackson and Ashton 1992). Additionally, sub-lethal doses of these toxicants produce
‘bait shyness’ in surviving target animals, because the animals become ill from a sub-
lethal dose and avoid the baits after their initial negative experience (Macdonald and

Penn 1994).

ll

The first anticoagulant rodenticide, Warfarin, was developed in the 19403
(Osweiler et al. 1985, Jackson and Ashton 1992). Similar compounds such as
chlorophacinine, dipacinone and pindone were subsequently developed. Collectively,
these compounds became known as the ‘first-generation’ anticoagulants (Jackson and
Ashton 1992). Anticoagulants (ACs) interrupt the vitamin K cycle in the liver, thereby
blocking the mechanism that produces clotting factors in the blood (Buckle 1994). A
sufficient amount of clotting factor remains in the blood for some time after ingestion of
the toxicant but is eventually depleted, ultimately leading to hemorrhaging and
subsequent death in the animal. An animal must ingest small quantities of first-
generation AC baits over several days before it will die fi‘om hemorrhage. First-
generation ACs offered a safer means of rodent control because of the multiple-dose
requirement, and one major advantage — a delay of several days between ingestion of bait
and the onset of symptoms (Jackson and Ashton 1992). These properties decreased the
risk of mortalities to non-target animals by accidental ingestion of baits, and eliminated
‘bait shyness’ in target species.

Rodents exhibiting inherited resistance to first-generation compounds were first
observed in Scotland in the late 19503. Resistant rodents were subsequently observed in
other parts of Britain and Canada. Rats that were able to eat first-generation AC baits
and survive had an evolutionary advantage and became the parents of the next generation,
passing resistant genes on to at least some of their offspring (Jackson and Ashton 1992).
In response to this crisis, researchers identified compounds similar to first-generation
ACs that were more active in resistant rodents. Difenacoum was the first compound

commercialized for use against warfarin—resistant rodents, closely followed by

12

brodifacoum in the late 19703 (Buckle 1994). Chemists in France synthesized
bromadiolone at about the same time. Later, flocoumafen and difenthialone were added
to the list of compounds that became known collectively as the second-generation

anticoagulants (SGAs). Brodifacoum is considered the most toxic of the SGAs (acute

LD50 in rat = 0.27 mg/kg; Godfrey 1985), and many instances of secondary poisoning

have been attributed to this compound (Erickson and Urban 2004). Bromadiolone is not

as toxic (acute LD50 in rat = 0.65 mg/kg; Godfrey 1985), but its use has also resulted in

secondary poisoning of non-target animals (Erickson and Urban 2002). SGAs offer a
distinct advantage over first-generation AC3 because only a single dose is needed to
produce mortalities, but with the same delay in onset of symptoms. Thus, SGAs became
known as ‘single-application rodenticides’.

Secondary poisoning of non-target species has been an issue of growing concern
in those parts of the world where anticoagulants are extensively used to control pest
populations (Mendenhall and Pank 1980, Godfrey 1985, Lund 1986, Poche 1988, DuVall
et a1. 1989, Cox and Smith 1990, Bemy et al. 1997, Howland et al. 1999, Shore er al.
1999, Stone et al. 1999, Mcdonald and Harris 2000, Eason et al. 2002, Fisher et al. 2003,
Foumier-Chambrillon er al. 2004, Brakes and Smith 2005). Primary poisoning results
from the direct ingestion of toxic baits, whereas secondary poisoning occurs when tissues
of poisoned target species are consumed by predators or scavengers. Non-target species
are at risk of secondary poisoning when they feed on poisoned carcasses or live, lethally
and sub-lethally poisoned prey during control programs that utilize AC3.

SGAs are especially dangerous because they are excreted very slowly,

accumulating in the body, and therefore are likely to be passed on to animals that eat

13

poisoned prey and carrion (Eason and Spurr 1995, Alterio and Moller 2000, Foumier-
Charnbrillon et al. 2004). These toxicants are more lipophilic than first-generation
compounds (Lechevin and Vigie 1992), and undergo slow metabolism and extensive
enterohepatic recirculation. For these reasons, they have long half-lives in vertebrate
tissues — several months compared to several weeks with first-generation AC3 (Jackson
and Ashton 1992, Hawkins et al. 1991 cited in Erickson and Urban 2004). The risk of
secondary poisoning is further increased by the fact that lethally-poisoned target animals
survive for 4 — 9 days before dying, and are available to predators. Cox and Smith (1992)
found that rat behavior changes significantly after ingestion of anticoagulant rodenticides.
They observed that live, lethally-dosed rats spend more of their time in open areas, are
more sluggish, and leave blood trails, circumstances that make them more susceptible to
predation. Also, poisoned carcasses offer a source of toxicant available to scavengers
after target animals die above ground. The phenomenon of secondary poisoning has
negatively affected non-target wildlife populations worldwide.

Control strategies in many countries utilize SGAs to combat outbreaks of rodent
pests. For example, SGAs are used in and around farm buildings to control common rats
in Britain, and may be used in fields to protect crops and feed hoppers from various
rodent pests (Shore et a1 1999). However, two cases of secondary poisoning by SGAs
have been documented in polecats (Mustela putorius), a rare species in Britain. These
small predators often hunt around farm buildings where prey is most abundant during the
winter. Poche (1988) found that 25 — 35% of small mammal predators are secondarily
exposed to rodenticides such as difenacoum and bromadiolone in Britain. These findings

were confirmed by Shore (1999) in two subsequent surveys of small mammal predators,

14

where similar rates of exposure were observed. Similarly, brodifacoum baits are used to
control rodent populations in apple orchards and near poultry farms in the United States,
where several instances of secondary poisoning from SGAs have been documented in
great-horned owls (Bubo virginianus) and red-tail hawks (Buteo jamaicensis) (Stone et
al. 1999). However, SGAs are not registered for field use in the United States.

Infestations of introduced Norway rats (Rattus norvegicus) have impacted seabird
populations on Langara Island, located at the north-westem tip of British Columbia’s
Queen Charlotte archipelago. Rats predate on eggs and breeding adult birds, causing
serious population declines (Howland et al. 1999). Brodifacoum baits were applied
during intense rodent control programs modeled after pest control techniques used in
New Zealand, resulting in the successful removal of rats from the island. However, an
intensive study demonstrated a clear risk of secondary poisoning to avian scavengers
such as ravens and bald eagles when brodifacoum baits were used to control rat
populations on the island (Howland et al. 1999).

In New Zealand, brodifacoum has been extensively used in efforts to eradicate
both introduced rodents and introduced predators. These species have devastated
endemic bird populations, placing many species in danger of extinction. A novel
approach to these operations has been to kill the predators of rodent pests that feed on
brodifacoum baits through secondary poisoning. The slow metabolism and persistent
properties of this toxicant allow it to acctunulate in top carnivores that scavenge on
poisoned carcasses or feed on live prey containing sub-lethal or lethal doses of poison
(Alterio and Moller 2000). Thus, both rodent pest and predator populations are

controlled during poisoning campaigns. Alterio and Moller (2000) demonstrated

15

effective control of introduced stoats (Mustela erminea) through secondary poisoning
when the stoats preyed on mice that had been primarily poisoned with brodifacoum.
During this study, brodifacoum applications successfully killed 100% of stoats in a South
Island beech forest along with various rodent species.

However, this secondary effect also influenced non-target species in New
Zealand. Instances of secondary poisoning were documented in five bird species after
islands were treated with brodifacoum baits for rat and rabbit control (Eason and Spurr
1995). Species found dead included native raptors such as the Australasian harrier
(Circus approximans) and morepork (Ninox novaeseelandiae), as well as Western weka
(Gallirallus australis australis), Steward Island weka (G. australis scotti), and southern
black-backed gulls (Larus dominicanus). Some studies involving radio-tagged
moreporks have documented mortality rates fi'om secondary poisoning as high as 21%
after aerial distribution of brodifacoum baits (Eason et al. 2002).

Bromadiolone is commonly used in France against pests such as coypu
(Myocastor coypus), muskrat (0ndathra zibethicus), and water vole (Arvicola terrestris)
during the fall and winter months (F ournier-Chambrillon et al. 2004). Baits are applied
in areas commonly used by rodents such as wetlands, marshes and ponds in western
France. Bemy et al. (1997) analyzed tissues from carcasses that were submitted over a 3-
year period. Bromadiolone residues were detected in 22/34 red fox (Vulpes vulpes),
15/ 16 buzzard (Buteo buteo), one lynx (Lynx Canadensis), and one badger (Meles meles)
carcasses. Since most of these predators were found in areas where bromadiolone had
been applied two weeks earlier, they most likely succumbed to secondary poisoning (> 4

days), and not primary poisoning. In Switzerland, mass mortalities of non-target species

16

occurred after bromadiolone was used to control water voles. An estimated 185 buzzards
(Buteo buteo), 25 kites (Milvus milvus) and one goshawk (Accipiter gentilis), as well as
several predators, were found dead after the control operation (Beguin 1983 and Pedroli
1983 cited in Shawyer 1987).
Anticoagulant rodenticide control: China, Tibet and Mongolia
China and Tibet

More than 10 different rodenticides have been used to control small mammal
outbreaks on the Qinghai-Tibet Plateau since the 19603 (Fan et al. 1999). In this region,
eruptive plateau pika and zokor populations contribute to pasture degradation and erosion
after areas have been overgrazed. A decline in the number of predator species fiom
secondary poisoning was documented on the Plateau after non-specific poisons were
spread. Secondary poisoning of natural predators has also been documented in Inner
Mongolia, China, after field treatments to control Brant’s voles (Zhong et a1. 1999).
Mongolia

Rodenticides have been used to control outbreaks of rodent pests in areas used for
livestock grazing for more than 40 years in Mongolia. In 2001, zinc phosphide was
replaced by bromadiolone (C30H23Br04, Figure 2-1) as the primary rodenticide used for
Brandt’s vole control. Bromadiolone is an anticoagulant rodenticide that is particularly
palatable to rodents, reported to be 90% effective against them (Batbayar and
Tseveenmyadag 2002). This toxicant is absorbed through the gastrointestinal tract, skin
or respiratory system and accumulates in the liver as the unchanged parent compound,
and is eventually eliminated from an animal’s system via the feces. Bromadiolone is a

potent, slow-acting toxicant that has a rapid initial elimination from the blood after

17

ingestion with an a half-life of 2 — 8 days, and a slower, terminal phase with a B half-life
of 170 — 318 days in tissues (Nahas 1987, WHO/FAO 1996, Erickson and Urban 2004).
The rate of elimination varies according to the dose ingested and particular species.
Nevertheless, its relatively long half-life makes bromadiolone a very likely candidate for
secondary poisoning of predators or scavengers that feed on poisoned rodents and
carcasses. In a study carried out with 14C-bromadiolone in four types of soil, the
compound was readily degraded (its half-life ranged from 1.8 to 7.4 days; WHO/FAO

1996), thus it does not accumulate in the environment.

0 /0
O ‘9“
OH

Figure 2-1. The chemical structure of bromadiolone (WHO/FAQ 1996).

In 2001, the Mongolian Ministry of Food and Agriculture (MFA) imported 13.5
tons of bromadiolone from China in the form of a 0.5% solution bottled in 20 L
containers (Tseveenmyadag and Batbayar 2003). The manufacturer-recommended
concentration is 0.005%-treated baits, intended for use indoors (WHO/FAO 1996).
Bromadiolone-laced baits prepared with this stock solution may, however, have been
mixed at higher than recommended dosages during the initial poisoning campaigns.
Initially, private contractors were hired to spread these baits, which were broadcast both
aerially and by hand in the fall of 2001 and spring of 2002. Bromadiolone applications

occurred over an estimated area of 350,000 hectares in 2002 (Tseveenmyadag and

18

Batbayar 2003), however, the MFA later reported that 494,000 hectares were treated that
year. Thus, some discrepancy exists about how many hectares were actually treated in

2002.

The recommended distribution of bromadiolone baits is 6 — 8 grains per m2 for

vole control (Damdin and Dagva 2004). Local herdsmen received large amounts of
bromadiolone-laced grain, estimated at 10 tons per county, from local authorities to be
spread by hand in areas with high vole densities. During the 2002 control campaign, one
human death and three serious poisonings resulted from ingestion of poisoned grain
because of improper instructions, markings and storage of the chemical (Tseveenmyadag
and Batbayar 2003).

In the spring and summer of 2002, biologists became alarmed when they found
over 340 non-target birds and eight mammals dead or dying in the Steppe region of
Mongolia (Tseveenmyadag and Batbayar 2003). Most of these deaths (over 340)
involved primary poisoning of birds that directly ingested toxic baits. In addition, it is
suspected that some predators found dead in the treated areas had succumbed to
secondary poisoning. Among these were two Saker falcons (Falco cherrug), an
endangered species throughout its range; two Corsac foxes (Vulpes corsac); and two
Pallas’ cats (0tocolobus manul), a near threatened species (Table 2-1). It appears that
when bromadiolone-laced baits were first spread in 2001, excessive dosages were used in
some areas (Anonymous 2004). The potential impacts of this practice were not

understood at the time.

19

Table 2-1. Wildlife species found dead in areas treated with bromadiolone-impregnated
wheat grain from suspected secondary poisoning in 2002 in central Mongolia
(Tseveenmyadag and Batbayar 2003).

 

 

 

Species Latin Name # Dead
Red Fox Vulpes vulpes 1
Corsac Fox Vulpes corsac 2
Pallas' Cat Otocolobus menu! 2
Black Kite Mi/vus migrans 2
Golden Eagle Aquila chrysaetos 2
Steppe Eagle Aquila nipalensis 1
Upland Buzzard Buteo hemi/asius 2
Saker Falcon Falco cherrug 2
Corvus spp. 23
Total 37

 

The mammal and raptor carcasses found in 2001 and 2002 exhibited
hemorrhaging in the cranium and coronary veins, and blood clots in the mouth, consistent
with anticoagulant poisoning (Batdelger and Potapov 2002). Most instances of
demoiselle crane deaths (through consumption of toxic grain, i.e. primary poisoning)
occurred in the Orkhon river basin and around lakes Ogii, Zegst, Doitiin and Tsagaan in
Arkhangai aimag. The chemical used — identified later as bromadiolone — appeared to be
a persistent poison that could be relayed between organisms along the trophic chain,
characteristics typical of SGAs.

Rodenticide use in Mongolia has steadily risen over the past ten years, increasing
from 105,500 hectares treated in 1997 to over 493,000 hectares treated in 2001 (Table 2-
2). In addition to bromadiolone, other rodenticides are likely to be in use in areas
occupied by wildlife, but there has been minimal research on their effects in Mongolia.
Newly imported rodenticides are currently being tested by the Plant Protection and

Research Institute (PPRI) in Mongolia. The PPRI’s testing for bromadiolone’s effect on

20

rats and Brandt’s voles began in 2000. This organization also undertook all rodenticide
treatments in 2004.
Table 2-2. Total hectares (x1000) treated with bromadiolone in 11 provinces of

Mongolia between 2002 and 2004. Total hectares treated are included for each year.
Information provided courtesy of the Mongolian Ministry of Food and Agriculture, 2005.

 

 

 

 

 

 

 

N°- No. Hectares Treated by Year (x1000)
Province Name Counties
Treated 21% 1203 2004
Arkhangai 4 158.3 0 0
Ovorkhangai 6 94.6 24 0
Selenge 1 20 0 0
Dundgobi 6 77.6 100 0
Gobi - Altai 2 5 0 0
Tov 13 101 .8 88 0
Khentii 6 28.5 8.4 0
Bulgan 1 4 0 0
Gobisumber 1 4 8 0
i Sukhbaatar 3 0 0 183.7
Dornogobi 1 0 80 0
TOTALS 493.8 308.4 183.7

An inspection report on manual methods for vole control in Khentii aimag
(province) — written jointly by officials from the MFA, Ministry of Nature and
Environment (MNE) and Food Security and Agricultural Protection Bureau — was
published in June of 2002 (Davaasuren et a1. 2002). Numerous livestock deaths were
reported in this document, including over 100 goats, 17 sheep, and at least one cow, all of
which showed signs of anticoagulant poisoning at necropsy, including bleeding from the
nose, and obvious hemorrhages in the neck and thoracic cavity. These deaths were
reported in Kherlen, Darkhan, and Bayankhutag soums. The poisoning of four children
and subsequent death of one child in Ulziit bag (district), Kherlen soum (county) — after
they had stolen and cooked bromadiolone-laced wheat grain — were also reported in this

document. The report briefly mentioned various technical problems encountered when

21

spreading baits by hand, including how heaps of baits were left behind when they were
spread with shovels by workers in vehicles, and the difficulty of properly training
workers to spread baits evenly.

After spending more than a million dollars on rodenticides since 1958, a decrease
in the Brandt’s vole populations still has not occurred. However, use of rodenticides has
continued, in part because impoverished soum governments, with little or no income and
few jobs, are paid to distribute poison. In November of 2003, discussion arose at the
ministries about application methods and it was suggested that different methods be used.
It was agreed to decrease the use of chemicals and instead use bacterial methods (referred
to as ‘biological methods’) to control Brandt’s vole outbreaks (B. Mendjargal pers.
com. 2004). Indeed, aerial bromadiolone treatments were limited to one province in
2004, and none in 2005, although areas may still have been treated by hand. The
ministries also decided to renew a 19803 program supported by the government for
controlling voles and conserving predators, which involved the training of local people to
used traditional methods including drowning out vole colonies with water, smoking-out
voles, and making conditions more favorable for natural predators of voles.

Current perceptions

In May of 2004, I met with Dr. Bold, a senior professor at the Institute of Biology,
Mongolian Academy of Sciences, to hear his views on Brandt’s vole outbreaks and large-
scale bromadiolone use in Mongolia. Dr. Bold believes that there has been a decline in
carnivore numbers over the past three years. Of special concern are the deaths of Saker
falcons in areas treated with bromadiolone. Mongolia exports these birds of prey to

countries in the Middle East for a considerable profit. Falcons caught in treated areas

22

may have compromised health, not to mention the decrease in population numbers that
secondary poisoning may cause (Bold pers. com. 2004).

Officials argue that animals found dead in treated areas may have died from
causes other than secondary poisoning, pointing out that dead wildlife are found in areas
that have not been treated. Both government officials and biologists are interested in
obtaining more conclusive evidence that will help determine whether primary and
secondary poisoning is indeed occurring in areas treated with bromadiolone (B. Battur
pers. com. 2004, Bold pers. com. 2004).

There are many issues influencing the perception of why vole outbreaks occur and
how they should be managed. Local herders believe that Brandt’s vole outbreaks lead to
poverty situations, and herders often solicit the government directly for help in stemming
outbreaks. In addition to overgrazing, degradation caused by nomadic camps may be
another cause of vole outbreaks (Dolgorrnaa pers. com. 2004). When families migrate
to new pastures, they often leave behind old camps with accumulated livestock dung and
household waste on which weed species grow. This produces conditions favorable for
Brandt’s voles.

Another issue is that of public land. No private pasture exists in Mongolia as of
yet, so nomadic herders are free to graze their livestock on all rangeland. There is a
reduced sense of responsibility in this situation, because herders can move to better
pastures when their current land becomes overgrazed and degraded — they have no ‘stake’
in the rangeland’s health (Dolgonnaa pers. com. 2004). The government is left to
manage rangelands, but currently does not have the infrastructure and economic

resources to do so effectively.

23

Current and fixture use of rodenticides in Mongolia

Bromadiolone is on the list of chemicals sanctioned for pest control by the World
Health Organization (WHO). However, WHO states that anticoagulant rodenticides are
not intended for direct application to growing crops (Tasheva 1995). Despite these
recommendations, and assertions from the MFA that only proper methods are used, aerial
treatments have been the chosen mode of bromadiolone applications since its initial use
in Mongolia in 2001. According to the MFA, methods other than aerial application have
proven to be inefficient in Mongolia. In forty minutes 300 — 400 hectares can be treated
with rodenticide using aerial methods, whereas it takes one person a whole day to treat
one hectare. Distributing baits by hand with ten people is too time consuming and not
useful (B. Mendjargal pers. com. 2004). Soum administrative officers distribute grain
to bag (district) leaders and herdsmen; however, no specific methodology on how to
broadcast it exists. Aerial methods are preferred because they result in more consistent
application of baits.

I spoke with the local environmental inspector and his employees while observing
a bromadiolone-treatment campaign during May 2004. One individual had observed the
mixing of baits from April 30 to May 9. Workers mixed baits by dumping buckets of
bromadiolone solution onto large piles of wheat grain, and then mixed it by hand with
shovels. This resulted in unevenly-concentrated grain, distinguished by patches of
lightly-colored, pink grain and very dark, red-colored gain. The inspector also said that
he thought the vole population had peaked during the summer and fall of 2003, and
crashed over the winter, so there were few voles left. His perceptions were verified when

we attempted to locate vole colonies in an area that had been aerially treated with

24

bromadiolone baits. We were not successful in finding active vole colonies after
searching for close to 12 hours, although we would occasionally see a lone vole or hear
one of their alarm calls. We observed several abandoned colonies, and the area showed
signs of extreme degradation. The environmental inspector believed that even though the
contracts to spread the bait had been signed more than a year earlier, the PPRI, under the
direction of the MFA, was still obligated to carry out the work despite the absence of any
signs of a continuing vole outbreak.

Some discrepancy exists as to whether or not bromadiolone is to be phased out of
use in Mongolia after 2004. Both the MFA and PPRI, the two organizations that
implement aerial application of bromadiolone baits, indicate that bromadiolone will not
' be used widely for rodent control after 2004. However, stockpiles of bromadiolone
solution and baits still exist in provincial centers and in Ulaanbaatar, and it seems likely

that these will indeed be used if there are any vole outbreaks in the near future.

25

CHAPTER 3

FIELD ASSESSMENTS OF THE RISK OF
SECONDARY POISONING

Introduction

In this chapter I describe the field studies that I undertook to assess the risk of
secondary poisoning to non-target wildlife species in central Mongolia. Initially, I
planned to trap and sample Brandt’s voles (Microtus brandti) in areas subjected to
Govemment-funded rodent control campaigns. In the summer of 2004, several such
areas were scheduled to be aerially treated with bromadiolone-laced wheat grain by Plant
Protection and Research Institute (PPRI), under contract from the Mongolian Ministry of
Food and Agriculture (MFA). Unforeseen changes in the control campaign schedule
meant that we were not able to carry through with this initial plan. Therefore,
independent field experiments were undertaken to address the research objectives (see
Chapter 1, Study Objectives, pp. 4-5). The field trial design, methods, observations and
results are reported in this chapter, concluding with a discussion of the risk that wide-
scale bromadiolone use may pose to non-target wildlife species in Mongolia.
Government Rodent Control: Field Observations

Since 2001, the Mongolian government has undertaken annual regional rodent
control operations using bromadiolone (Tseveemnyadag and Batbayar 2002). During
May of 2004, bromadiolone-treated grain was scheduled to be spread by the Plant
Protection and Research Institute (PPRI) in three soums (counties) — Uulbayan,

Tuvshinshiree and Talbulag — of Sukhbaatar aimag (province), under the direction of the

26

Ministry of Food and Agriculture (MFA). The PPRI planned to aerially treat 500,000 ha
of rangeland with bromadiolone-laced wheat grain in an attempt to control recent
outbreaks of Brandt’s voles. Three airplanes were sent to Sukhbaatar to spread the
treated grain. Upon our arrival in Ulaanbaatar in late May 2004, the MFA advised us that
these aerial bromadiolone applications had begun on May 13, and were scheduled to be
completed by May 25 (MFA official, pers. com. 2004).

Based on this information, Dr. Hickling and I traveled to Sukhbaatar on May 22
to observe the poisoning campaign, and to trap and sample voles in treated areas. We
were unable, however, to locate an area with active vole colonies that had been treated
with bromadiolone baits during the May poisoning campaign. The local environmental
inspector believed that the vole population had peaked during the summer of 2003 and
then crashed over the winter, so few voles were present by the time we reached the area
in spring. Furthermore, even though PPRI technicians showed us areas on maps that had
supposedly been aerially treated with rodenticide, we were unable to find significant
amounts of bait on the ground at any of the recommended field locations.

Therefore, instead of trapping for voles in an area being treated by the PPRI, we
decided to conduct our own field experiment to determine the amounts of bromadiolone
present in tissues of lethally and sub lethally poisoned voles.

Field Trials
Study area and environmental data

Our field trials were conducted at a site 100 km south-west of Ulaanbaatar, the

capital of Mongolia, in Altanbulag soum. The Pallas’ Cat Conservation Project has used

this area — centered on grid reference 47°32'27.30"N, 105°55'43.20"E — as a study site

27

since 2000. The project has a good relationship with a local herder family who camps
along the Tul River each summer. Family members often help collect data or are
employed as field technicians, and the field team could purchase meat and milk products
from the family. Also, the study area was located within the buffer zone of Hustai Nuruu
National Park, offering some protection to native wildlife species. This, coupled with the
fact that few herders inhabit the hilly region of the study area during the summer months,
offered an ideal location to conduct field research.

The region is characterized by rolling grassland with occasional hills reaching to
1656 m above sea level. The Tul River flows through the large, flat portion of this area,
with occasional braided channels creating small islands. Steep ravines and gullies are
found throughout the hilly region, offering ideal corridors for wildlife. Hill slopes and
ravines vary from being slightly rocky, to consisting almost entirely of rocky outcrops
and boulders. The vegetation was composed of grass species including Stipa bylovii,
Cleistogenes squarrosa, Achnatherum splendens, S. sibirica, Elymus chinensis,
Agropyron cristatum, Poa attenuata, Carex durisculla, and F estuca dahurica, various
sedge species such as Artemisia fiigida, and A. adamsii, and spiny shrubs such as
Caragana.

During the field assessment, minimum and maximum temperatures were recorded
daily with a digital min-max thermometer, and meteorological data for the general area
(daily mean temperature, percent cloud cover, average wind speed and rainfall) were

obtained from the local meteorological station in nearby Altanbulag (47°42’13.6”N,

106°25’17.7”E).

28

 

a) M w b) ISLAND
RUSSIA /
, / \
‘\

 

 

 

 

  
  

  

0/
_------__, _I g0

 
     

,4 ‘ .
\ // \\
N Study Hga\

I Vole mm

A Camera Sltes
'\

@ Base Ca p \I’\
/\/ Roads \\ \
N Rlvers " \ ‘1‘

0

 

(K \\\ ‘
f

\\

 
  

."\\ I i

5 10 15 an 3 so lfllm

r

Figure 3-1. A map of the study area located 65 km south-west of Altanbulag, Tov
aimag, Mongolia, depicting a) the study area’s location in Mongolia (red dot near
Ulaanbaatar); b) an enlarged map of the island where 15 vole sampling plots were
established; and c) the entire study area with locations of the 16 camera-trapping stations,
‘ and the study island in relation to Base Camp and the Tul River.

29

Materials and Methods: Vole Bromadiolone Trials

Study site
The vole toxicant study was conducted on a 4 km2 island in the Tul River

(47°32'09.10"N, 105°54'14.20"E), just west of our base camp. The topography
surrounding the river consists of flat grassland as described above. The island offered an
ideal location for the toxicant trial because the river created a natural barrier that helped
to keep livestock and other animals away from the research plots. These plots could be
monitored frequently since the island was only 2.5 km from base camp.
Selection of experimental plots

In early June 2005, nine vole colonies with numerous vole burrows and high vole
activity (i.e. fresh dirt around burrows, active pathways, and audible vocalizations) were
selected for plot establishment, and their locations recorded using a GPS unit. A 10 x 10
m research plot was established around each colony by erecting a 1 m tall fence made of
plastic window screening, supported by metal and wooden stakes. Occasionally livestock
crossed the river, so these fences prevented livestock from entering the plots and helped
deter potential mammalian scavengers from interfering with carcasses and bait.
Similarly, nylon line flagged with small pieces of aluminum was strung across the top of
each plot to discourage birds from interfering with the carcasses and toxic wheat grain.
Trapping and sampling

In each of the nine plots, 10 Sherman live traps (H.B. Sherman Traps,
Tallahassee, Florida, USA) were placed near active burrows. These traps were baited

with a mixture of peanut butter, oats, and anise oil, and were left wired open for 5 days to

30

familiarize voles with the traps and non-toxic bait. All traps were closed on Day 5, and
remained closed until two days after the plots were treated with poisoned wheat grain.
Wheat grain treated with two different dosages of bromadiolone solution was
prepared at the Plant Protection and Research Institute (PPRI) in Ulaanbaatar. Samples
of the treated grain were kept for later assay to confirm toxic loadings. On June 25, 2005,
four plots were treated with 0.005% wt:wt bromadiolone, four plots with 0.05% wt:wt
bromadiolone, and one plot with non-toxic wheat grain. Approximately 800 grains of

wheat, by weight, were broadcast manually in each plot to obtain a rate of 6 — 8

grains/m2 (the sowing rate employed for vole control by the PPRI; B. Battur pers. com.

2004, Damdin and Dagva 2004). Plots were thereafter checked periodically for 8 days to

count the number of wheat grains remaining in five 1 m2 quadrats placed in each plot,

and to collect voles that died above ground as a result of the treatment. Liver and
stomach samples were collected from vole carcasses and stored in liquid nitrogen for
later analysis.

In addition to the carcass-gathering, live voles were trapped and sampled
approximately 2, 4 and 8 days after bromadiolone-laced wheat grain had been broadcast
on the research plots. Traps were set and provided with fresh peanut butter bait mixture.
They were checked between 5 and 12 hours after being set, and four live-trapped voles
were collected from each plot. If fewer than four voles were trapped in a plot, traps were
baited, reset, and checked approximately 6 hours later to obtain the quota for each plot.

Live-trapped voles were anesthetized with a halothane-saturated (Halocarbon
Laboratories, River Edge, New Jersey, USA) cotton ball, and then euthanized via cervical

dislocation. During necropsy, their whole livers and stomachs were collected for later

31

bromadiolone quantification. One body sample was also reserved from each plot during
each trapping session for later comparison of bromadiolone quantities in the liver to body
amounts. Body samples were prepared by removing the pelt and head fiom the vole, and
all four paws. The gastrointestinal tract, stomach and liver from these voles were stored
in separate 2 or 10 ml cryogenic vials. All samples were frozen in liquid nitrogen, in a
shipper, for importation into the United States, and transport to Michigan State
University’s Diagnostic Center for Population and Animal Health (DCPAH), East
Lansing, MI. Samples were stored in a -20 C freezer at DCPAH until they could be
processed.

Carcass monitoring

During the scavenger identification trial (Camera Trapping p. 33), beetles were
observed feeding on many of the vole carcasses, which contributed to their rapid
disappearance. This raised questions about how long poisoned carcasses were likely to
remain available above ground after an area had been treated with bromadiolone. To
address this question, we monitored carcasses on several additional experimental plots
that we treated with bromadiolone-laced wheat grain.

To select suitable vole colonies, we used soil to close vole holes in 10 different
colonies, and then assessed what percentage of these holes had been re-opened 24 hours
later. The six vole colonies with the most re-opened holes were then fenced (as described
in the previous section) and used as research plots. Three plots were treated with
0.005%, and three with 0.05%, bromadiolone-treated wheat grain. Plots were searched at
12 h intervals for 7 days to monitor the persistence of the carcasses of voles that died

above ground. Each vole carcass that we located was marked with a wooden stake,

32

photographed, and given a decomposition/absence score using the scale described in
Table 3-1 (Camera Trapping, p. 35). On subsequent visits, at 12 h intervals, these
carcasses were relocated, photographed, and re-scored until the completion of the 7-day
experiment.

The number of carcasses found above ground was totaled for each plot and
summary statistics were calculated for these data to determine the overall number of
carcasses potentially available per hectare for a 7-day period, after an area had been
treated with bromadiolone.

Beetle trapping trial

To gain a better understanding of what species of beetles and other invertebrates
were contributing to carcass disappearance, a short invertebrate-trapping experiment was
undertaken. Beetle traps were constructed at six old camera trapping sites. For each
invertebrate trap, a 10 x 20 cm hole was dug, and a 10 cm water bottle with the top out
off placed in it. This was half-filled with water, and a few drops of soap were added to
break the surface tension of the water. A collapsed Sherman trap was wired to wooden
stakes, left 10 cm above the ground, over the hole to reduce evaporation and to deter
birds from removing invertebrates from the trap (Figure 3-2). A fresh vole carcass was
tethered near the water hole, and hung over it. When invertebrates came to scavenge on

the carcass, they fell into the water and were drowned.

33

 

Figure 3-2. A beetle nap constructed over a partly buried water bottle, half
full of water, with a vole carcass tethered over the hole to attract invertebrate
scavengers.

Beetle traps were checked approximately 12 hours after being established. The
number and different types of invertebrates trapped was recorded at each beetle trap, and
traps were subsequently removed.

Bromadiolone HPLC analysis

Samples were prepared by removing stomach contents from stomachs, and
weighing all liver, body (with pelt, head, paws and gastro-intestinal tract removed) and
stomach contents samples to 0.01 g accuracy with an Ohaus Explorer Pro scale (model
EP6102C, Switzerland). A total of 79 samples were then shipped on dry ice to Iowa
State University’s Veterinary Diagnostic Lab in Ames for bromadiolone analysis.

Liver, body and stomach contents samples were analyzed via High Performance
Liquid Chromatography (HPLC) to determine concentrations of bromadiolone in the
respective tissues. The HPLC analysis was done at Iowa State University’s Veterinary
Diagnostic Laboratory under the supervision of Dr. Paula Irnerrnan, following procedures
described in Chalermchaikit et al. 1993. The average recovery rate for bromadiolone was

75 :l: 9%. The minimum level of detection was 0.025 parts per million (ppm) for a 2 g

34

sample, so values below this were designated ‘NDL,’ meaning ‘no detectable level.’
During statistical analyses, a value of 0 ppm was assigned to each of these NDL data
points.

HPLC data analysis

Total amounts of bromadiolone in liver and body samples were calculated by
multiplying the concentration of bromadiolone in ppm by the weight of each sample.
The relationship between liver and body amounts was assessed using a least squares
linear regression (Statistix 8.0 2003). Only amounts calculated from measured
concentrations in both liver and body samples were used during this analysis. The
resulting regression equation was then used to estimate amounts of bromadiolone in
bodies of voles fi'om which only livers were sampled. Known total amounts of
bromadiolone in the liver and body were also compared to calculate the average
percentage of bromadiolone sequestered in the liver.

Bromadiolone amounts were calculated for stomach contents samples by
multiplying measured concentrations from the HLC analysis with original stomach
contents weights. Total bromadiolone amounts were calculated for each vole as the sum
of liver, body and stomach contents (when available) amounts. If there were no, or very
little, stomach contents available for a particular vole, we assumed they had no
bromadiolone in their stomachs. These values were compared for the two bait
concentrations (A = 0.05% and B = 0.005%) and two trapping periods (2 and 4 days) by
assigning ranks to the data and performing a Randomized Complete Block Analysis of

Variance on the ranks (Analytical Software 2003).

35

Liver concentrations were compared between treatments using a Wilcoxon Rank
Sum test (Analytical Software 2003). This test was also used to compare the total
amount of bromadiolone in livers and bodies between treatments.
Materials and Methods: Scavenger Identification
Camera trapping

Six digital camera-trap stations (CTS) were used to monitor scavenger activity at
a total of 16 sites from June 10 through July 17, 2005. Local herdsmen, especially
J. Batzorig and J. Gantulag, who were knowledgeable of the area and animal movements,
identified sites suitable for establishment of CTSs. The locations of these sites, which

were mostly in vole colonies or ravines, were recorded using a GPS. Habitat variables

measured in a 100 m2 area around each CTS included slope; aspect; percentage of

vegetation cover, litter, bare ground and rock cover; vegetation species composition; and
general topography type. Topography types were separated into four categories; 1) rocky
hill slope, 2) ravine, 3) steppe, and 4) valley. Location photographs were taken from two
different perspectives near the CTS at each site.

Each CTS consisted of a motion-sensitive digital camera in a protective ‘trailcarn’
housing wired to stakes driven into the ground, and aimed down at a vole carcass placed
approximately 1 m from the camera. Two types of remote cameras were used: a
Cuddeback Digital Scouting Camera Model C-lOOO (1.3 megapixel image sensor; Non
Typical Inc.) and a Bushnell Trail Scout Digital Scouting Camera (2.1 megapixel image

sensor; Bushnell Performance Optics Go).

On the day each CTS was established, a fresh vole carcass was tethered to the

ground with a buried metal stake and wire 30 that scavengers could not easily remove the

36

carcass before triggering the camera Cameras were mounted so as to obtain a clear view
of visiting scavengers and the surrounding area. They were camouflaged with rocks and
bushes to avoid alarming scavengers and to deter thieves. During the day, Cuddeback
cameras took a still, color photograph when tripped, followed by 30 s of 15 frame per
second (fps) video; after a minimum 30 s delay, if tripped again, the cycle was repeated.
When tripped at night the video feature became disabled and they took only still images -
illuminated with a visible-light flash — at 30 3 intervals. The Bushnell cameras were set
to take 15 s of 15 fps color video when tripped, with a minimum 30 s delay between
successive recordings. The Bushnell cameras provided both day and night video, with
illumination provided via built-in infra-red LEDs. Both cameras were programmed to
record images continuously. Every 2 to 7 days the images and videos were downloaded
ffom the cameras’ 256MB digital media cards and stored on a laptop computer. If no
scavengers were recorded at a site, and the rodent carcass had not been disturbed for 7
days, the CTS was moved to a new location and provided with a fresh carcass. Cameras
that remained at the same location for 3.5 weeks were then moved to new CTS locations

regardless of scavenger activity at the previous site.

Non-toxic rodent carcasses were obtained using snap-traps in an area near the
study site, and placed in view of cameras within 6 hours of being trapped. A total of 71
carcasses were used to replenish 17 different CTSs during the study. To minimize human
scent that might deter scavengers, carcasses were handled with gloves and stored in Zip-
loc bags. To help attract potential scavengers to CTSs, a dab of Steppenwolf 11 scent lure
(Milligan Brand, Chama, New Mexico, USA) was smeared on a small rock and placed

near CTSs each time a new rodent carcass was exchanged for an old one.

37

Each time CTSs were checked, the condition of rodent carcasses was recorded on
a scale fi'om 1 — 5 (Table 3-1), with 1 equal to very fresh and 5 equal to Z 90% absent.
This scale offered a standard method for rating carcass condition between observers, and
provided information on whether a carcass had merely decomposed, or had been
removed. The visual scores were validated with photographs taken at the time of scoring
(Figure 3-3).
Table 3-1. Carcass scoring criteria for vole carcasses that were tethered in view of

camera trap-stations (CTSs) at 16 sites from June 12 — July 17, 2005, in a study area near
Altanbulag, Tov aimag, Mongolia.

 

Score Description
1 Fresh carcass, not penetrated or bloated
Recently penetrated carcass. some bloating and

 

 

 

 

2 insect activity
Flesh present on carcass; maggots may be

3 present; and/or insects actively feeding on
carcass

4 Dehydrated carcass consisting mostly of fur and

bones; and/or Insects feeding on carcass
5 Carcass is 2 90% absent. almost nothing left

 

 

 

 

 

 

ate JUN ~05 F'j;

 

 

 

Figure 3-3. Photographs of vole carcasses in three stages of decomposition/
absence; far left: score = 3, middle: score = 4, far right: score = 5. Altanbulag,
Tov aimag, Mongolia.

38

Image analysis

The CTS image data were summarized by counting the number of visits for each
species (recorded separately for daylight and night time hours) and scoring the behavior
of each individual on a scale from 1 to 5 (Table 3-2). Multiple images of the same
individual taken within 30 min of each other were counted as a single event. Visits more
than 30 min apart were counted as separate events.
Table 3-2. Behavior scoring criteria for wildlife species that visited camera trapping

stations (CTS) fi'om June 12 — July 17, 2005, in the study site near Altanbulag, Tov
aimag, Mongolia.

 

Score Behavior

 

Visited the CTS and an image of the animal
was recorded

 

Investigated the carcass

 

Ate insects & maggots off the carcass

 

Consumed part of the carcass

 

 

 

 

mew»

Consumed or removed the whole carcass

 

The visitation rate for each species was calculated per 100 camera-trap hours,
given the number of visits and total number of monitoring hours per site. A mean
visitation rate per 100 hours was calculated for each species. A mean scavenging rate
was calculated for each species in the same manner, and summarized by species. The
percentage of visits that resulted in scavenging was calculated for each species. In
addition, the number of visits during the day and night were counted for each species, and

summarized for ‘all birds’ and ‘all mammals.’

39

Authorizations

Trapping and euthanizing Brandt’s voles for both field studies was approved by
MSU’s Institutional Animal Care & Use Committee (Application 03/04-056-00).
Permission to conduct the field experiments was obtained from the Ministry of Nature
and Environment in Ulaanbaatar, and supported by the Mongolian Academy of Sciences,
Mammalian Ecology Laboratory.
Results: Vole Bromadiolone Trials

Wheat grain counts

It was difficult to achieve the recommended broadcast rate of 6 — 8 grains/m2

when initially treating plots. On average, plots were actually broadcast with 21

grains/m2. The amount of wheat grains per m2 declined at a steady rate for 5 days post-

treatrnent (Table 3-3), after which almost no wheat grains remained.

Table 3-3. Decline over time in the numbers of wheat grains per m2 in plots broadcast
with two concentrations of bromadiolone-treated grain at the study site near Altanbulag,
Tov aimag, Mongolia.

 

 

 

Plot Days after bait application
0 0.5 1 1.5 2 3 4 5
,\° vcso 19.6 12.4 9.0 4.4 4.2 3.4 2.2 1.8
§ vcss 21.0 14.6 10.4 11.0 8.0 ~ 1.2 0.4 0.2
o vcs4 14.2 9.0 7.6 3.8 3.2 0.6 0.6 0.4
3° vcss 34.0 19.2 17.4 13.6 7.4 4.0 4.0 3.8
3 vcs7 20.6 13.6 11.4 10.4 5.8 2.8 1.4 1.0
o vcss 17.2 13.4 12.6 11.8 11.0 9.0 7.0 6.2
Mean 21.1 13.7 11.4 9.2 6.6 3.5 2.6 2.2

 

40

Amount and concentration of bromadiolone in voles

Bromadiolone was detected in 32 of 37 livers, 5 of 19 stomach contents, and 4 of
12 bodies of voles sampled from treated research plots. The highest concentrations
detected in liver, stomach contents and body samples were 24.37, 33.54, and 0.71 ppm,
respectively (Appendix A). In general, liver samples contained the greatest amounts of
bromadiolone and stomach samples the least, although the amount of bromadiolone in
stomach contents was highly variable. No bromadiolone was detected in samples taken
from voles trapped on the control plots, where non-toxic wheat grains had been spread
(Table 3-4).

No significant difference was detected between liver bromadiolone concentrations
of voles trapped on days 2 and 4 after treated grains were spread (P > 0.05); however, a
significant difference was observed in liver bromadiolone concentrations of voles from
treatments A and B (Figure 3-4; P < 0.05). Overall, liver bromadiolone concentrations
from voles trapped on high-dose plots were four times greater than from voles on low-
dose plots (Table 3-4). A two-way analysis of variance on the ranked data showed no
significant interaction between treatment and day trapped (P > 0.05). Outlier values for
two voles, one in each treatment, skewed the data significantly for each of the treatments
(Figures 3-5 a and b); therefore, it was necessary to use non-parametric statistics for

comparisons.

41

.mcoEeQEoo 05:8.—
265 05266 3 £82: 05 mo macaw—=28 Sch 3682 :03 022 Am: 36 .50 I boon ”w: ~_.mm .3»: emdm u .30 528% ”w: >2. .3? 26 I .53:
m can Aw: 3.3 .60 n Econ ”w: 3.: .wh: 8.: .1. .28 goes—Sm ”w: 84% .whz nmém I :55 < 3:25.85 89¢ :03 one .mo_o> 23 how 82.? warn—SO...

 

 

eoec8v

m 8.0 m 8.0 23 m 8.0 86 m. 86 86 o seesaw:
A So a and 3 :3 a cad 4; 63 a and e are d A3689

em RN 6 5.0 Nod a 36 26 em Se «3 m 5958:
83 a an.“ “a $3 a sod .3 A86 a 82 e 83 a A389

: «a: o me...” mad a mod 8.6 : 8s 8...” Seesaw:

: a: .395 a as 353 a 63 333 a as A99:
525 Eek 68 .20 :85on .02..

 

.ezomcoE $65.5
>8. .538. wean—=82 he: «one .895 e E 3383 use Seaman 203 86> .oeo_o%mfiob mo mucus—E886 95 £3, coeeewoafim
58w 308$ 53> coach mac—m 80¢ BEES 83> com .. $363 2998 was 3235588 35308 wfims teem—3:8 I $823
38 was .3383 Son .3538 sumac; .83— 5 31v 3:508“ 98 Awhzv meoaebeooaoo 0:223:85 ma 3 .802 Tn 033—.

42

 

 

 

 

 

 

 

 

 

10-
9.
8.
E71
e
2 6‘
2 *-
0
=6 5‘
E
e 4‘
m -)(-
3 3‘
.J
2.
1.
o- |
A B
0.05% 0.005%

Treatment

Figure 3-4. Box and whisker plot of bromadiolone concentrations (ppm) in
the livers of voles collected from high-dose (A = 0.05%) and low-dose (B =
0.005%) research plots. One outlying value (24.37 ppm) has been truncated
to 10 ppm in treatment A.

43

a) Treatment A

 

 

 

WW7 V/flr/M

 

”ma/AZ

MWmW/vflfl

///.,////

 

       

Oran/[vmmha'o

 

4 3
Sconce".

Bromadiolone Concentration (ppm)

b) Treatment B

 

 

 

 

 

7 6 5 4 3
5:032“.

Bromadiolone Concentration (ppm)

Figure 3-5. Frequency distribution of bromadiolone concentrations in the livers of voles

exposed to a) Treatment A — 0.05% bromadiolone-treated wheat grain, and b) Treatment

B — 0.005% bromadiolone-treated wheat grain. Voles were trapped 2 and 4 days post-
exposure from fenced plots in a study area near Altanbulag soum, Tov aimag, Mongolia.

44

Whole livers were collected from all of the voles trapped and found dead on the
ground during this study (11 = 42); however, I was able to transport body samples of only
15 voles back to the United States fiom Mongolia for HPLC analysis. It was therefore
necessary to determine the relationship between measured amounts of bromadiolone in
the liver and bodies of sampled voles in order to estimate bromadiolone amounts, and
ultimately total body burdens, in the bodies of voles that were not sampled.
Approximately 53% (i 8% SE) of the ingested bromadiolone accumulated in the liver,
based on a comparison of total bromadiolone in livers and bodies of 19 voles. An un-

weighted least squares linear regression showed that amounts of bromadiolone found in

vole bodies were strongly correlated with liver levels (Figure 3-6; R2 = 0.98, P < 0.0001),

30 the regression equation (Carcass = 0.664*Liver - 0.369) was used to estimate the
amount of bromadiolone in those vole bodies that were unavailable for HPLC assay. In

contrast, bromadiolone amounts in stomach contents exhibited no consistent relationship

with liver amounts (R2 = 0.09, P > 0.05), so it was not feasible to estimate the missing

stomach contents data. Since the amount of bromadiolone in stomach contents was
typically very low (median = 0), the amount of bromadiolone in stomach contents were

assumed to be 0 ppm for voles from which stomachs were not sampled.

45

12.00

 

    

1000 y = 0.664x - 0.369

R2 = 0.979
8.00 -

Body Sample (pg)

 

 

 

0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00
leormo)

Figure 3-6. Linear regression of liver and body bromadiolone amounts (pg),

calculated using measured concentrations and sample weights, for eight voles

sampled from plots treated with bromadiolone-irnpregnated wheat grain in

Altanbulag soum.

Significantly more of bromadiolone was found in the livers (P < 0.05) and bodies
(P = 0.05) of voles sampled from the high-dose plots (A) than those from the low-dose
plots (B; Table 3-3). Total liver bromadiolone content (pg) was almost five times higher
in voles from high-dose plots than those from low-dose plots, and body amounts were
more than 11 times higher. This test agrees with our earlier comparisons of liver and
body concentration values (ppm) for each of the treatments. Bromadiolone was detected
in the stomachs of 3 of 9 (33%) voles recovered from high-dose plots, compared with
only 2 of 10 (20%) voles recovered from low-dose plots. However, given the small
sample sizes, this difference was not statistically significant (Fisher Exact test, P = 0.35).

The two voles with the highest total burdens of bromadiolone were both found

dead above ground in the high-dose plots. More voles were found dead in the high-dose

46

plots (n = 5) than in the low-dose plots (n = 3). Data pooled from both treatments
showed that the concentration of bromadiolone in livers from carcasses of dead voles was
almost four times greater than the concentration in livers of live voles (Table 3-5); this
difference was statistically significant as a one-tailed test (P < 0.05). This was consistent
with the estimated total body burden of toxicant in vole carcasses, which was almost
eight times higher than the body burden in live voles.

Table 3-5. Mean (i SE) bromadiolone concentrations (pg/g) in livers, and total body

burdens (pg) in live-trapped voles and vole carcasses found on the ground during the
field study (Treatments A and B are combined).

 

 

Carcass (Dead) n Live 11
Liver (09/9) 6.03 8 1.54 29
(t SE) (r 2.84) (i 0.36)
Body Burden (pg) 32.31 8 4.25 29
(E SE) (1: 18.06) (i 1.30)

 

Carcass monitoring

During our l2-hourly visits to the carcass-monitoring plots we encountered four
lethally-dosed voles that exhibited sluggish behavior and reacted slowly to our presence.
These voles were observed moving slowly in open areas, away from their burrows,
during daylight hours. At times these voles left blood trails because they were bleeding

from the nose or anus.

Twenty-nine voles were found dead above ground in a 1100 1112 area after

treatment with bromadiolone-laced wheat grains, a mean of 2.6 carcasses/100m2 (.1: 0.5

SE), when assuming adjacent colonies. During the carcass monitoring experiment, voles

were discovered dead above ground between 60 and 108 hours after being exposed to

47

bromadiolone baits (Figure 3—7). These carcasses disappeared 12 to 96 hours after being

discovered (Figure 3-8).

 

168

421/.

132 144 156
Hours Post Treatment

120

108

        

A, M. /%M%//////////////

   

72

60

 

 

9 8 7 6 5 4 3 2 1 0
20on 0.08850 Co .02

Figure 3-7. Number of carcasses present at 12 hour intervals after voles were

2

exposed to bromadiolone-treated wheat grain in six 100 m plots located near

Altanbulag, Tov aimag, Mongolia.

48

0.00

 

1.00 it,

N
o
0

Mean Decomposition!
Absence Score
00
O
O

<11 .5
o o
o o

r—f—fi
H—.

|

Ire—I
E11
.4.

l
PH

 

 

6.00

 

Time Since Carcass Detected (Hours)

Figure 3-8. Mean decomposition/absence (scored as follows: 0 = dying vole,
1 = fresh carcass, 5 = carcass absent) of 18 vole carcasses over a 7 day period

after bromadiolone-treated wheat grain had been spread in six 100 m2 plots,
located near Altanbulag, Tov aimag, Mongolia. The dotted line corresponds

to absence of the carcass.

The primary reason carcasses disappeared so quickly was the scavenging activity
of carrion-eating invertebrates. In particular, burying beetles (Nicrophorus spp.) often
buried carcasses or consumed them in a short period of time. During the beetle trapping
trial, different types of invertebrates were trapped that had supposedly scavenged on a
vole carcass. Out of the six beetle traps, a total of 31 Nicrophorus species (N.
investigator and N. germanicus) and 37 Carabidae species (T aphoxenics spp., Harpalus
spp., Amara fodinae and Ptherostiches ravus) were trapped over a 12 hour period, along
with various spiders, grasshoppers and flies including: Dennestidae, Myriapoda,

Opiliones, Blaps gudosa, and Anatolica spp.

49

Results: Scavenger Identification

During 4,704 hours of camera monitoring, 20 wildlife species visited CTSs on a
total of 276 occasions (Table 3-6; see Appendix B for images). This represents a mean
visitation rate of 5.9 visits per 100 hours of observation. Twelve of the 20 species
camera-trapped were mammals, with over half of the mammal visits (52%) occurring at
night, whereas most bird visits (81%) occurred during the day. Of the bird species,
Northern wheateaters visited CTSs the most frequently (2.53 visits/ 100 hrs), followed by
Pere David’s snow-finches (0.29 visits/ 100 hrs) and Asian short-toed larks (0.245
visits/ 100 hrs). Corsac foxes had the highest visitation rate for mammals (0.57 visits/ 100
hrs), followed by Mongolian silver voles (0.36 visits/ 100 hrs), and Brandt’s’ voles (0.17
visits/ 100 hrs).

Scavenging was observed during 13% of bird visits and 28% of mammal visits.
Northern wheateaters were the most likely bird species to scavenge (0.206 times/ 100 hrs),
followed by Asian short-toed larks (0.029 times/ 100 hrs) and little owls (0.026 times/ 100
hrs). Of the mammals, Corsac foxes scavenged most frequently (0.187 tirnes/ 100 hrs),
followed by Daurian hedgehogs (0.112 times/ 100 hrs) and ground squirrels (0.061
times/ 100 hrs; Table 3-6).

The species most likely to scavenge during a visit were little owls (100% of
visits), ground squirrels (60%), Daurian hedgehogs (57%) and steppe polecats (50%).
Scavenging behavior was not observed for seven of the 12 mammal species, and three of
the eight bird species, that visited CTSs. Most of the bird species that scavenged
appeared to be eating insects off the rodent carcass along with small pieces of tissue.

With the exception of little owls, no raptors were recorded at CTSs. On a few occasions,

50

herbivorous mammals such as jerboas, argali sheep, and red deer visited CTSs; however

these species made no visible attempt to scavenge carcasses (Table 3-6).

51

 

 

 

 

 

 

 

 

2 09. SEE ....<mm>o

N 8 9 3 0.2.5...

- - 0 No.0 N 0 N .0603. 0305.0:

- - 0 08.0 N 0 N 8000.6 8200 .80 8...

- - 0 0.00 P 0 F 8550 0.30 82.0 =00...

- - 0 0.. F .0 N. 0 0 8.8.5 8.20.: 20> 0.0.080

- - 0 80.0 N 0 N .0800 08.080 3.... 02.0002...

- - 0 0 8.0 F v 0 8.2.0.0 000.00.? 03.0..

r - o NNod —. —. o @905 $20: uOGUGm :m_nu._:m
No.0 0000 N N80 4 v 0 30050.80 0.0.002 086.0“. 2.030
08.0 0:: 0 N80 0. N 3 808.58 88...... 20> 32.0 02.0000...
08.0 8.00 m 08.0 m 0 m 8.0.800 3.2.6 .8300 00:20
8 :0 3.3 v 08.0 N N 0 08.0008 80.8862 02.008: 00.500
00 F .0 8.8 5 80.0 00 N 0 0860 80.0.. Noumeaww
08:38..

N 8. 0» B. WEE

- - o «Nod w o F mqoqo 08:3 goo... cause-0m

- - 0 80.0 0 0 m 0.08058 85.8 «.0... >038.

- - 0 08.0 F F 0 0.0.80.0 080052.“. 0:3 .850:
08.0 8... P N80 N N N 8.800 0305.0:
N00 3. 2 N 08.0 8 N 0. 000.0300 888:0 0600.38» 0.0.80 20..
No.0 0.8: N No.0 N 0 N 8.80 82...... .36 2:...
:80 8.2 0 8N0 8 N 5 6.88.88 80.08.00 0.3.. 082.20 00.2
0 N0 :8 t 80N «N: N 8 058:8 05880 5028...... 52:32
moms.

8.. 00:88. 3.05 8002.80 2.. 00:38. 8.65 83> 83> use: 0:3 8.08m

00.0028» 00.00280 .0 85:... 02.3.65 353 2.0.: a as: a
cue! cue:

 

000.800.. 00.00% £000 .8: 88%.: 80800 mo 0:: o3 :0: 0:8 88:02:00
8.08 05 0:0 .8320: 88:028.... 035.88 00.00% 5023 8 83> :0 088080: 8030on 08> 830:0: w8w:0>0om 008:
(«o 0028:: 05 $838: 80800 we we: 2: :0: 0:8 8:86? :008 83> me 028:: .88 83> Sofie 8 omgmv 8mm: .08. Emma
8 568 me: me 0008:: 05 0.0208 005: 0:5 .800. 8 :88 80¢ 8080800 >05 008: we 0098:: 08 .3 00x88 0:0 b88800.
“0058 0.8 £08808 0:0 8:5 8.0802 808.8 >3. $088810. mo 803-580 5 co 00.8 beam 0 8 $808808 80800 .«o 8:0:
85. 8.30 88 .28: 1 N. as: as. 8080. 0. .8 8.0.80 .808... 8.88 86:86 08...... as 8.8% 6.0.2.3 0.... 8.80

52

Occasionally, tracks of foxes and other animals were visible around a CTS, yet no
images of these individuals were recorded by the camera. Cautious behavior by animals
while in the vicinity of a CTS, combined with limited sensitivity-adjustment of the
triggering sensor on the camera, are likely explanations for these ‘missed’ visitations.
During the initial setup of this study, the Bushnell cameras’ sensors were often overly
sensitive, so that the camera would be triggered by grass blowing in the wind and other
movements, which would fill up the media cards over a short period of time. On these
occasions, animals may have visited the CTS without being recorded because of lack of
storage space on the media card. In addition, Cuddeback cameras often recorded images
that were so over- or under-exposed that it was not possible to determine if an animal was
present at the time. Over the course of the study, two cameras malfunctioned so that
monitoring of those sites ended prematurely. One Cuddeback camera malfunctioned
because a small grasshopper short-circuited the flashcard socket. One Bushnell camera
stopped recording reliably for an unknown reason, and thereafier would only function
intermittently.

Camera trapping vs. vehicle surveys

Corsac foxes and Daurian hedgehogs were seen much more frequently during
vehicle surveys than during passive camera trapping (Table 3-7). However, elusive
species such as steppe polecats and Eurasian badgers were recorded at camera traps

without ever being seen during vehicle surveys.

53

.6008 000M 0>0:m .3 000.88 0:03 0:00 008:0 0.0.:0> ..

 

 

.NN. 0000 8 N00 08.800 8.80.0. 2.08%

.0. 2.0 8.8.0 8.0.0.... 20> 0.0.00.0

00.0 .N: 0.00 E 00.0 8.0.0.0 80.8.... «00.0..

00.0 .0. N00N 0.0. 800.. 20.. ._0..

.5 00.0 00.02. 00.03. .0003 00.085

Avv Nwd .E:mEm..0>o 0.0.033. econ—On— 233m

.3 00.0 0800.800 0.00.0.0. 20> .050 00:09.0!

00.0 3 N00 .N. 00.0 80.088 80.00000... 0000000.. 08000

00.0 .5 N .0 .88. 88.0.8.0 .00 .08....

.N. 0N... .0. N00 88.0.80 :03 :20

.N. 0N... .0. 00.0. 80.0.. 08.0.. .8“. .020

.0. SN .00. 5.09 .0N. 03 000.00 80.0.. 00.. 00200

0... 00:00:80.0 2.. 00:00:08.0 2.. 03.0.00 0.52 0.0.... 8.0000
.80020002 .0000...» 0.0.: 00.05.05 20.2

 

moom 5080000 or - V moom 808082 on - N moom 3:... t - 00:... 2 Hm._.<n_

.0..0w:02 $08.0 >0 ... $030832 :00: 00:0 00:00 05 :. 00.0000 .08808
N. .0 {0.00808 0.0...0> w:.::0 08.0.3.0 Emu: 0:0 >080 8.8.0:: 0:0800 05 :0. 00:0: :0.:0:.0.> £8: .0 0000000800 .hum 030,—.

54

Discussion
Implications of vole bromadiolone results

This study documented that dead voles are commonly available to scavengers
above-ground afier areas are treated with bromadiolone-laced grain, and thus refutes past
claims that almost all poisoned voles die below ground in their burrows. Furthermore,
these carcasses do contain measurable amounts of bromadiolone (mean total body burden
= 32.3 pg 0: 18.1 SE). Depending on the density of vole colonies, the number of toxic
carcasses potentially available above-ground could be as high as 264 carcasses/ha, in
areas where colonies are dense and in close proximity. For a more realistic scenario, we
might assume that colonies are at least 10 meters apart. In this case, the density of
carcasses would be 3.8 carcasses/ha.

When large areas are aerially treated with bromadiolone, the potential for a

scavenger to encounter numerous toxic carcasses and lethally-poisoned live voles is

great. The home range size of corsac foxes (Vulpes corsac) may reach 30 km2 in some

areas, although home range size in this species varies widely according to habitat quality
(Heptner and Sludskii 1992). In this case, a fox could potentially encounter as many as
11,400 vole carcasses when roaming its entire range, given our estimate of 3.8

carcasses/ha. Recent studies have shown that manul cats (more commonly known as

Pallas’ cat, Otocolobus manul), also maintain home ranges that can reach 30 km2 or more
(Munkhtsog and Ross 2006) and, therefore, could potentially encounter a similar number
of toxic carcasses. Furthermore, bromadiolone is a cumulative toxicant, so if treated

grains are spread multiple times in an area, the risk of increased toxicant levels and of

secondary poisoning becomes much greater.

55

Findings from previous studies suggest that the number of target species that die
above ground after broad-scale anticoagulant poisoning is highly variable. Cox and
Smith (1992) observed that half of lethally-poisoned rats appeared to move deliberately
into open areas to die during an enclosure experiment. Alterio and Moller (2000)
observed numerous poisoned rodent and possum (T richosurus vulpecula) carcasses on
the ground in areas treated with brodifacoum in New Zealand. Researchers also
discovered carcasses of lethally-poisoned Norway rats above ground during a similar
study in Canada (Howland et al. 1999). Other studies have reported that the majority of

target species die underground after being exposed to SGAs (Fenn et al. 1987, Harrison et

al. 1988, Taylor 1993). Indeed, the 2.6 vole carcasses per 100 m2 that we recorded

during this study are almost certainly a minority of the voles in a colony. We trapped 71
voles and discovered 11 vole carcasses on the ground — a total of 82 voles — in 9 colonies

that were approximately 10 x 10 m, which represents a density of at least 9 voles per

100 m2.

Carrion beetles contributed significantly to carcass disappearance. On several
occasions, burying beetles successfully buried whole vole carcasses within 24 hours after
the carcass was placed (during the camera study) or discovered (during carcass
monitoring). So, even though many carcasses of poisoned voles may be available afier
poisoning operations, they may only be available for a few days after voles succumb to
poisoning. This same phenomenon was observed during a study on the environmental
fate of ground squirrel carcasses in Montana (Sullivan 1988). The amount of time

squirrel carcasses remained above ground (mean = 3 days) was commonly determined by

56

the presence of carrion-eating invertebrates that quickly attacked and consumed
carcasses.

In addition to the risk from poisoned carcasses, 86% of the live voles trapped in
treated areas contained bromadiolone residues at levels that were generally lower than
those found in the carcasses. Wheat grain counts showed that voles steadily consumed
bromadiolone-laced wheat grains for 5 days post-treatment, at which time nearly all of
the grain was gone (Table 3-3). Some intoxicated individuals exhibited sluggish
behavior and were observed sitting out in the open, bleeding from the mouth and anus,
and would have been easy targets for predators. These same behaviors were observed by
Cox and Smith (1992) during a controlled experiment to investigate the pre-lethal effects
of anticoagulant rodenticide on the behavior of wild rats in cages and enclosures.
Bleeding in rats was observed 48 hrs after they had ingested anticoagulant-treated food.
When the rats were closer to death, they exhibited sluggish behavior and would fieeze in
the open when alarmed instead of bolting to cover. Sluggish behavior and bleeding in
live, lethally-poisoned animals has been observed in multiple instances. Rodents
poisoned with brodifacoum in New Zealand became lethargic and disoriented, and left
blood trails during terminal stages of poisoning, leading to a concentration of hunting by
stoats in the treated zones (Alterio and Moller 2000). These behaviors make prey such as
voles and rats particularly vulnerable to predation. After observing the hunting behavior
of several raptor species, Rudebeck (1950) concluded that predatory birds have twice as
many successful attacks on injured or abnormal prey.

Even though the least squares regression between total liver and body

bromadiolone levels was significant, we do not have data for the middle of the curve.

57

Thus, we assumed a linear relationship existed between the total amount of bromadiolone
in the liver and body of voles in order to estimate bromadiolone amounts in body
samples.

We also assumed that the amount of bromadiolone a vole ingests, and the amount
that is sequestered in the liver exhibits a linear relationship. This assumption is supported
by findings from Morin et al. (1990) who observed this relationship when conducting a
kinetics study on bromadiolone in various tissues of coypu (Myocastor coypus). They
found that bromadiolone quantified in livers was linearly correlated with the total
quantities of bromadiolone consumed by each of 15 animals during the study.

Data from the vole bromadiolone experiment were not normally distributed, so it
was necessary to use non-parametric statistics for comparisons. Notably, while the
median level in the carcasses was 10.7 ug, two voles found dead on the ground during
this study contained much larger amounts of bromadiolone (151.4 and 47.3 ug,
respectively); such ‘outliers’ are not common, but may be of particular concern with
regard to the risk to non-target species.

A significant difference between the bromadiolone concentrations in vole tissues
existed according to the dosage of bromadiolone-laced wheat grain used to treat plots,
although this difference was less than the 10-fold difference we might have expected
given the ten-fold increase in dosage (0.005% compared to 0.05%). It is likely that we
underestimated the total body burden of voles since we were not able to estimate the
amount of bromadiolone in the stomachs of those voles that ‘were not sampled, and
assumed that the stomach concentration for these voles was 0 ppm. Nevertheless, it is

clear that a consequence of using higher than recommended concentrations of

58

bromadiolone-treated baits is that increased amounts of the toxicant will be present in the
carcasses that end up lying above-ground in treated areas.

The amount of bromadiolone in stomach contents of voles was highly variable,
ranging from 0.14 — 35.12 pg, and did not contribute much to the total bromadiolone load
in most of the voles sampled during this study. In a few instances, when a vole is preyed
upon or scavenged soon after it has consumed a significant meal of treated grain, the
amount of bromadiolone relayed to a predator or scavenger may be significant. Two
voles found dead on the ground during this study contained relatively large amounts
(11.84 and 35.12 ug) of bromadiolone in their stomachs. These voles would pose a much
higher risk of secondary poisoning to non-target scavengers and predators than an
average-dosed vole.

Implications of scavenger identification results

The species most likely to consume poisoned vole carcasses was the little owl.
This conclusion is based on the fact that little owls consumed part, or all, of non-toxic
rodent carcasses that were tethered in view of camera traps. However, only two visits by
little owls, eight days apart, were documented. Since these visits occurred at the same
location, it is highly likely that the same individual visited both times. Therefore, we are
not able to conclude that all little owls scavenge on vole carcasses.

Of particular interest was the evidence that steppe polecats scavenge on rodent
carcasses. These animals are specialized for entering burrows to predate on small
mammals. It is likely that polecats will go down burrows to prey on voles (Nowak 1991),

and may consume carcasses they find below ground.

59

In general, it was difficult to score behaviors of animals camera trapped by
Cuddeback cameras because, for the most part, they recorded still color photographs of
visiting species. It was easier to determine whether a particular animal was scavenging
when viewing video clips that recorded their behavior for some seconds, than from a still
photograph that only captured a moment of their visit. Some data on actual scavenging
behavior may have been lost because of this. For example, a badger (Meles meles)
visited a Cuddeback CTS on one occasion, but it was close to the camera in the
photograph that was taken, and I could not tell whether it investigated the vole carcass.
This same problem occurred with images taken of steppe polecats at night.

Numerous birds of prey have been observed in the study site near Altanbulag, but
were never recorded at CTSs. Ross (2005) observed 49 upland buzzards (Buteo
hemilasius), 40 steppe eagles (Aquila nipalensis), 30 European kestrels (Falco
tinnunculus) 9 cinereous vultures (Aegypius monachus) and 2 golden eagles (Aquila
chrysaetos) during 3 hours of observation in the fall of 2005. Saker falcons (Falco
cherrug) and black kites (Milvus migrans) have also been observed in the study area.
However, none of these species were recorded at the CTSs. Raptors most likely prey on
moving targets and use visual cues when hunting, therefore, it is doubtful that they would
scavenge on carcasses tethered in front of cameras. However, these species would likely
predate on poisoned voles that were moving around in open areas during the final stages
of death from SGA toxicosis.

Predators such as manul cats, gray foxes (Vulpes vulpes), and wolves (Canis
lupus) were not recorded at camera traps. Manul are a rare species in the study area,

extremely cautious, and easily alarmed, so it is unlikely that this species would approach

60

a camera trap. In a zoo setting, manul cats are often fed dead chicks and mice as part of
their weekly diet, and they readily consume these dead animals (Pelletier pers. com.
2006), whether this is at all indicative of natural behavior remains uncertain. Gray foxes
and wolves are prevalent in the study area, but may be more cautious than other predators
about approaching a camera trap. Human scent on the cameras or carcasses may have
deterred these canids from scavenging. I observed that the longer cameras were left
undisturbed, the more likely it was for an animal to visit the CTSs. Also, even though we
placed cameras in a diversity of habitats, they may not have been placed in areas
frequented by certain species of predators and scavengers.
Risk of secondary poisoning

For those species that readily scavenge carcasses, little risk of secondary
poisoning exists when the recommended concentration (0.005%) of bromadiolone is
used. Corsac foxes were observed scavenging vole carcasses during the camera trapping

study. A similar species, the domestic dog, would potentially need to consume more than
8,500 loaded voles (Table 3-8) from a high-dose area to reach a single lethal dose (LD50;
the lethal dose to kill 50% of the population). Even when considering the long half-life
of bromadiolone in tissues (170 — 318 days), it is unlikely that a dog (or fox) could

consume enough of these voles to be at risk of secondary poisoning, in part because most

carcasses would have disappeared within just a few days of the poisoning operation. The

risk of reaching an LD50 for a domestic cat is even lower; a cat would have to ingest over

9,500 loaded voles (Table 3-8) to reach lethal levels of bromadiolone. However, if a dog
or cat were to prey on voles loaded with maximum levels for a long period of time (i.e.

more than 2 months), the risk of secondary poisoning increases twelve-fold (Table 3-8).

61

If voles were to consume a great deal more of the treated grain, and their total

body bromadiolone levels approached 10 mg, then a canid (fox) would need only to

consume 10 voles to reach a potential LD50 value. However, the dosage of treated grains

would have to be extremely high in this case. Since bromadiolone is brought to
Mongolia in bottles of 0.5% liquid solution, this scenario is highly unlikely.

Table 3-8. Comparison of estimated maximum tolerated dose (MTD) values for two
species, the total amount of bromadiolone required for an MTD in each species given an
average weight, and the estimated number of voles these species would have to consume
to reach their MTD. Two estimates are given, one using the mean (11.7 ug i3.8 SE)
bromadiolone content of voles from high-dose (A = 0.05%) plots, and one using the
maximum bromadiolone content of a vole encountered during the study (151.4 pg).

 

 

T I D A Lethal Dose of No. of Voles for a
Species in"? "(0:9 We:e:%: ) Bromadiolone Lethal Dose
9 9 g 9 (mg) (Mean amt. A) (Max amt.)
Dog 10.0 101703 10.0 100.0 8532 661
Cat 25.0 mo" 4.5 112.5 9599 743

 

a MTD = maximum tolerated dose (Osweiler et a1. 1985). An MTD approaches an LD50.

These comparisons require many assumptions. Lethal dose values likely differ
between species; thus, the lethal dose to kill 50% of a domestic dog population will be
different than that for a population of foxes. Similarly, lethal dose values vary among
individual animals. Furthermore, these calculations assume 50% population mortality;
but this would clearly be an unacceptable level of mortality when considering populations
of threatened and endangered species such as manul cats or saker falcons. A few
mortalities from secondary poisoning may be enough to cause serious declines in the

populations of these rare species. However, information on the lethal dose to kill a

minimum number (e.g., an LD1) of species such as these does not exist.

62

When considering the long-term effects of bromadiolone consumption, it is likely
that additional risk exists. The long half-life of bromadiolone would allow predators and
scavengers to accumulate this toxicant over time if they were foraging in areas treated for
an extended period of time. Even at sub-lethal levels, the probability of mortality
increases (Brakes and Smith 2005). Animals may become more susceptible to the effects
of bromadiolone if they are stressed or encounter periods of increased activity
(Mendenhall and Pank 1980), leading to lower survival. Also, multiple small doses

ingested over time are likely more lethal than a larger initial dose. A study in India

showed that the chronic dose (0.17 pg) of bromadiolone is much smaller than a single
dose (0.37 pg), to achieve an LD50 in field mice (Mus booduga; Balasubramanyam and

Purushotham 1987).
Conclusions

Secondary poisoning of non-target wildlife species has been documented during
rodent control operations using bromadiolone in France and Swizerland (Beguin 1983
and Pedroli 1983 cited in Shawyer 1987, Bemy et al. 1997, Fournier-Chambrillon et a1.
2004). It is difficult to ignore these reports when they suggest that bromadiolone does
indeed pose a risk to predator and scavenger species. Therefore, more research is needed
to determine the chronic and sub-lethal effects of bromadiolone in situations where non-
target wildlife is secondarily exposed to this potent rodenticide.

In cases where higher than recommended dosages of bromadiolone baits are used,
and extensive areas of agricultural land are treated, a risk of secondary poisoning is
highly likely. However, many different factors must be considered when attempting to

assess the risk that bromadiolone use poses. Managers must be aware of the influences

63

that result in outbreaks, and produce clear plans to deal with outbreaks using methods
that will benefit stakeholders while conserving wildlife populations and the environment

to the best of their ability.

64

CHAPTER 4

OVERALL DISCUSSION AND CONCLUSIONS

Introduction

This chapter reviews the initial reports of non-target poisonings in Mongolia,
implications of results gathered from field trials, factors that lead to exposure and further
research that is needed to address the problem of non-target poisoning. This thesis
concludes with overall recommendations on anticipating and managing outbreaks,
risk/benefit analyses of control programs, and in-use monitoring and pasture
management.
Concern over non-target poisoning

Reports of dead wildlife in areas treated with bromadiolone baits (i.e.
impregnated wheat grain) have led to much speculation about the potential threat these
campaigns pose to non-target wildlife species in Mongolia. The death of one child and
poisoning of three others after they ate bromadiolone-laced wheat grain was reported in
numerous newspapers in Mongolia and abroad, bringing heightened concern over the
proper storage and use of this anticoagulant poison. During 2002 and 2003, letters urging
the cessation of bromadiolone use were written by the World Wildlife Fund, UNDP
Eastern Steppe Biodiversity Project, and Wildlife Conservation Society, to the Ministries
of Food and Agriculture, Nature and Environment, and Health.

The most obvious threat to non-target wildlife from large-scale applications of
bromadiolone baits is that of primary poisoning through direct ingestion of treated baits.

Many cranes were found dead over the course of the first two treatment campaigns.

65

Birds are attracted to the red, brightly-colored wheat grains, and quickly ingest lethal
doses of the baits. During May of 2002, when the first non-target deaths occurred, cranes
were migrating to Mongolia from their southern wintering grounds. Roughly 150 cranes
were found dead at seven different locations throughout central Mongolia fiom primary
poisoning resulting from consumption of bromadiolone-impregnated grain.
Hemorrhaging was observed in the neck and thoracic cavity of these birds, typical
clinical signs of anticoagulant poisoning.

The Pallas’ Cat Conservation Project became concerned about the negative effect
poisoning operations may have on threatened Pallas’ cat (manul) populations afier two
manul cats were found dead in Darkhaan soum, Khentii aimag following bromadiolone
treatments in the vicinity. These concerns prompted my investigation of the risks these
large-scale applications might pose to the natural predators and scavengers of Brandt’s
voles.

Implications of the field trials

First, there is clear evidence that lethally-poisoned voles die above ground and are
available to scavengers after bromadiolone treatments. However, these carcasses may
not be available to scavengers for more than 2 or 3 days because carrion beetles bury or
eat the vole carcasses in a short period of time. Live, lethally-poisoned voles have slower
reaction times and often sit out in the open, making them more vulnerable to predation.
These voles may also continue feeding on baits, offering a significant load of toxicant to
predators that prey on them.

Second, various wildlife species will scavenge rodent carcasses when they are

available above ground. Indeed, a study in New Zealand found that predators and

66

scavengers congregate in areas where lethally-poisoned rodents are easy to catch, and
their carcasses are abundant (Alterio and Moller 2000). These factors, combined with
repeated bromadiolone treatments, would result in situations offering a high risk of
secondary exposure to the toxicant.

Despite the above two observations -— which confirm that non-target species are
likely to encounter toxic carcasses and that some will then scavenge them - our field
trials suggest that there is little risk of secondary poisoning of non-target species when
recommended dosages (0.005%) of bromadiolone are used. However, the risk of
secondary poisoning becomes greater if the concentration of toxicant in the bait exceeds
the recommended level. This is supported by our findings that the concentration of
bromadiolone in tissues of target animals increased significantly when the dosage was
increased 10-fold.

If areas are frequently treated with bromadiolone, this compound may become
persistent in animal tissues, allowing it to accumulate in tissues of predators and
scavengers ingesting poisoned voles. If these non-target animals were able to feed on
toxic voles for several days, they could potentially accumulate levels of bromadiolone

that would have negative, sub-lethal effects, or lead to death from hemorrhaging.

We used LD50 values for our calculations; however, it would be more appropriate

to asses the risk to threatened and endangered species such as Pallas’ cats and saker

falcons using LD1 values. Unfortunately, this information is unavailable, but we can

assume that these species would have to ingest far fewer voles than estimated in our

studies to reach the minimum lethal dose value.

67

Exposure factors

Anticoagulants are not recommended for field use (Stone et al. 1999, Erickson
and Urban 2004); rather they are manufactured for use in buildings and enclosed spaces
where the secretive nature of rodents and other target species reduces their availability to
predators (Godfrey 1985). When SGAs are used for pest control in agricultural settings,
baits are scattered in open area, and are thus available to many non-target animals. Both
carcasses and live, intoxicated pests are more accessible to scavengers and predators in
these settings. The persistent nature of SGAs, coupled with increased exposure of non-
target species in open areas, greatly increases the potential for non-target secondary
poisoning (Mendenhall and Pank 1980, Godfrey 1985). Thus, the risk of secondary
poisoning to non-target animals depends on exposure factors such as the behavior of live,
lethally-poisoned target species during the final stages before death, the availability of
carcasses above ground, the toxicant loading of the target species, and the behavior and
diet of non-target species (Kaukeinen 1982, Howland et al. 1999).

Treatment campaigns using bromadiolone have been proven very effective in the
short-term. Bromadiolone treatments were reported to be 90% effective against rodents in
initial field trials in Mongolia, 2001. However, vole populations recover rapidly due to
increased availability of resources and increased reproductive rates. Voles can become
relatively abundant within a short time (2 4 months) afier treatments, eventually leading
to additional outbreaks (Zhong et al. 1999). A great deal of money and time has been
invested into bromadiolone campaigns in Mongolia, yet this management practice must

continue annually or bi-annually to keep vole populations at low levels in outbreak areas.

68

Further Research

Clearly, primary poisoning poses the greatest risk to wildlife species in Mongolia.
The impacts of this phenomenon should be studied in greater detail if bromadiolone
applications continue. Several species of vulnerable and endangered cranes migrate to
breeding grounds in Mongolia annually. These species include white-naped cranes (Grus
vipio), red-crowned cranes (Grus japonensis) and Siberian cranes (Grus leucogeranus).
The impact of primary poisoning may be greatest during the spring and fall when cranes
are migrating to and from Mongolia — these seasons are also the preferred time to spread
rodenticides for vole control.

Regarding secondary poisoning, it is important to determine the effect of repeated
exposure to bromadiolone over time in non-target species. Even though low levels of
bromadiolone were detected in vole tissues during this study, the risk that repeated
exposure to these levels poses is not known. There is evidence that several small doses
of SGAs are more toxic than the same amount given as a single dose (Lund and
Rasmussen 1986). Studies of first-generation anticoagulants document that small doses
administered over a period of time are more toxic than a single large dose (Jackson and
Ashton 1992). Chronic exposure may reduce the ability of predators to regulate prey
populations (Zhong et al. 1999). Therefore, it is important to study the effects of
repeated exposure to bromadiolone to ascertain its lethality under these circumstances.

Any compromise to the health of predator populations, whether it is sub-lethal
effects or secondary deaths, will lead to a reduction in the number of natural predators of

Brandt’s voles. These predator populations have a much lower reproductive rate than

69

Brandt’s voles, and, therefore, may not be able to control vole populations during years
following bromadiolone treatments.

Overall Recommendations

Overgrazing and vole outbreaks

Researchers have documented a link between overgrazing and rodent outbreaks in
areas of Tibet and Inner Mongolia (Fan et a1. 1999, Smith and Foggin 1999, Zhong et al.
1999). Recent studies have also recorded the relationship between rodent outbreaks and
climatic indices (Zhang et al. 2003). Brandt’s voles invade areas and become abundant
when conditions become favorable to them. Overgrazing, or drier climatic conditions,
result in shorter grasses, creating ideal habitat for Brandt’s voles. Zhong et al. (1999)
found that when degraded grassland is fenced to exclude livestock, these areas recover
and Brandt’s vole numbers decline. Heavily grazed areas are in danger of Brandt’s vole
outbreaks (Pech et a1. 2002, Zhang et al. 2002). Thus outbreaks may be avoided, or at
least reduced, through improved rangeland management.

When outbreaks do occur, rodenticide treatments should be followed by intense
rangeland management to restore the natural vegetation height, creating sub-optimal
habitat for Brandt’s voles. Under these conditions outbreaks are less likely to occur
(Zhong et al. 1999). Climatic conditions can be predicted so that during years of low
precipitation, grazing pressures can be fiirther reduced to avoid vole outbreaks. Using
these predictions will allow managers to anticipate outbreaks and take the necessary

precautions.

70

Risk/benefit analyses and alternatives

Managers need to conduct risk/benefit analyses before deciding to use persistent
toxicants to control rodent outbreaks. Rodenticide treatments must be repeated
frequently to keep vole numbers low. Therefore, this management tool may not be the
most economical, since poisoning campaigns are expensive and labor-intensive. Also,
poisoning campaigns do not address the underlying cause of vole outbreaks — pasture
degradation from overgrazing. In the long-term, rangeland management strategies that
restore vegetation to its normal height and create sub-optimal habitat for Brandt’s voles
may be the most cost effective strategy for rodent control in Mongolia.

Less invasive methods for rodent control are currently being studied. One
alternative is the future possibility of fertility control of pests through immuno-
contraceptive vaccines that are delivered orally through baiting (Hinds 2005). However,
these ‘biological’ tools are still in development and have significant technical and
political hurdles to overcome before they could have any field application in Mongolia.
In—use monitoring

Brown et al. (1998) suggest a stepwise approach to assessing the environmental
impact of rodenticides. If bromadiolone applications continue in Mongolia, it is
imperative to implement an assessment procedure such as this. This approach would
include 1) laboratory testing, 2) field testing, and 3) in-use monitoring of bromadiolone.
To study the environmental hazard of treatments, the procedure should include a protocol
for collecting and freezing carcasses and samples from both target and non-target animals
found dead in treated areas. If it is not feasible to freeze the whole carcass, the liver from

dead animals can be frozen separately. These samples can later be analyzed to ascertain

71

the toxicant-loading of tissues, and assess the risk to non-target animals, including
livestock and humans.

Ideally, bromadiolone use would be avoided and instead pastures would be
properly managed, thus avoiding the risk of secondary poisoning. If bromadiolone
treatments are used to stem rodent outbreaks, pasture management, focusing on
sustainable stocking rates and pasture rotation, should be a central focus of the
management program. If pastures are better managed, the probability of frequent

Brandt’s vole outbreaks can be expected to greatly decrease.

72

 

 

 

 

 

 

 

 

 

 

5:: 000.00 - 40.: 00.3 - 00.: 00.0 - 00.: 50.40 032 No
00.: - 00.0 - 00.: 00.0 0:0 00.: 00.0 - 5.0 8.5 N
00.0 - 00.0 00,0 00.0 00.3 0: :0 00.0 00.0 00.0 31.2 N
03 000.0 - 00.0 01. - 00.0 03 - 3.0 8.0 032 S
00.0 000.0 - 00.0 00.0 - 40.: 00.: - 00.0 0: 032 3
00.0: 02.0 - - 00.0 - - 00.: - - 00.0 022 :
:3 «04.0 - 00.0 E0 - ~00 $.20. - 00.0 00.0 5.2 5
00.3 00.0: - :0 0:0 - 20 0:0. - 0:0 2.0 352 E
00.00 - 00.0 00.0 00.0: 00.0: 30 Sa :0 00.0 $0 032 :
00.3 - t: - 00.0: 2.: 00.0 00.0 00.0 - 00.0 032 F
00.0 - 00.0 00.0 00.0 00.0 00.0 0: 00.0 00.0 00.0 03: F
3.0 - 00.0 00.0 30 00:0 3.0 40.0 00.0 00.0 0:0 032 :
3.00.0. < 000.5020
.25.. mm >000 wwwpcoww .02.. 000 wwwfiowm .9: >000 “WMHMW 9: 0.0500 00:00
«.525 30.2.3595 “a .0an 2.3:. .25“. 0:22—00:85

 

 

.33 58:50 waist azomaoz .MEEe >8. $025532 Laos no.8 €30 0 E 33:80 was
39905 80% 86> .Eogaob :08 £53, 958085-008 0.9% v H N can and N H C woven mfinmmb 3 98 I 908 U use @03qu
m A0089 < 10580080 8 wEEoooa «080: 80 Sam .23 :08 now 00:00: :0 Sow $588.0 magnum .3 30030—3 33500 20>
:03 com one—290805 mo 5an been :38 05 Ed dab 0:00: some mo 03mm?» 2: 52m 08303200 H5550 ~89 E26033 HEB
.Emfiwofifiofio 2:34 ooggofiom :93 a? 35088 02980.0 Econ 98 £528 £008on ao>= com maouwbaooqoo 0:0—06088.0.—

< Mam—755.3.

73

 

 

 

 

 

 

 

 

 

 

00.0 1 1 1 00.0 0V.0N 1 00.N 1 1 00.0 05400 0
00.0 1 00.0 00.0 00.0 1 00... 0...N 00.0 00.0 00.0 051.00 0
00.0 1 00.0 1 00.0 ..0.N.. 1 00.N 00.0 1 00.0 051.00 0
00.0 000 .0 1 00.0 00.0 1 00.0 00.N 1 00.0 00.0 001:0 F
00.0 1 00.0 00.0 00.0 N0. 00 00.0 00.0 00.0 00.0 00.0 N01. r0 _.
- 3.8.0. o 20532.0
0 .. .v 0NN .0 1 1 00.0 1 0N.0 00.0 1 1 00... V0..Nm N
00.0 000.0 1 1 00.0 1 00.0 05.0 1 1 00.0 00.50 N
00.N 1 00.0 00.0 00.N 00.V0 00.0 00.0 00.0 00.0 00.0 N0..Nm N
00.0 N05 ... 1 1 0N.0 1 0...0 00.N 1 1 FN._. v0.50 N
00.0 1 00.0 00.0 00.0 00.0N FV.0 N5.N 00.0 00.0 00.0 504Nm N
00.0 000.0 1 1 00.0 1 0N0 00.N 1 1 V0.0 00..Nm N
V0.5V 000 .V 1 N...00 50.5 1 00... 5V... 1 V0.00 0...0 NV1. v0 00
00 .0 000.0 1 00.0 00 .0 1 00.0 00.0 1 00.0 N00 ..V.. 90 00
0N.0 000 .0 1 1 0N0 1 1 v0.0 1 1 NV.0 00.. rm PG
00.0 000... 1 00.0 00.0 1 00.0 00.N 1 00.0 .5... 001:5 F
00.0 000.0 1 1 00.0 1 0...0 00.N 1 1 00.0 3.. ..m ..
0...0 1 00.0 1 00.0 5V0? 5N0 00.N 00.0 1 00.0 001:0 ..
5V0 005 ... 1 1 00.N 1 NN.0 0V. V 1 1 00... 0N1. rm ..
0V... 550.0 1 1 NF... 1 00.0 00.0 1 1 V0... VN.=.m ..
3.0 1 5w... 1 V0.0 0V.0N 0...0 00.N 00.0 1 V0... ”NJ—.m _.
0N0 550.0 1 1 000 1 0.20 0V.w 1 1 00.0 NN1. rm ..
0...0 000.0 1 1 0...0 1 N00 .00 1 1 0N0 FN... rm ..
05.0 000 .0 1 1 00.0 1 No.0 00.0 1 1 0.... 0N1. rm ..
00.0 1 00.0 00.0 00.0 0V.5N 0N0 No.0 00.0 00.0 00.0 0.... ..m ..
V0.0 V: .0 1 00.0 05.0 1 00.0 N..... 1 00.0 00.0 0r... Fm F
05.0 500 .0 1 1 00.0 1 00.0 00.0 1 1 0N... 5W. rm —.
0...N 1 00.0 00.0 3... 00.0.. 00.0 00... 00.0 V0.0 05.0 0....vm ..
0N0 00N.N 1 00.0 00.V 1 00.0 50... 1 00.0 00.N 0p... ..m F
00.V 000... 1 1 No.0 1 0N0 N0... 1 1 006 V: ..m ..
No.0 N05 .0 1 00.0 VN.0 1 05.0 N0.F .. 00.0 05... 2... pm ..
.0300. m 32502...
.59 “Home.“ .000 “WMWMW .95 .000 “@900“ .02.. .000 wwwhfi 9: 0.028 8000
.0... «caoEa 30.23520 .0. 3a.: 2.3.... .Enfidcoo 0:22—02:80 _

 

 

74

APPENDIX B

Select images of wildlife species that visited Camera Trapping Stations (CTSs) from
June 12 to July 17, 2005, in the study area near Altanbulag, Tov aimag, Mongolia.

Little Owl 18 June 2005

 

Snow Finch 6 July 2005 Ground Squirrel 15 July 2005

7S

   

Steppe Polecat 12 June 2005

‘1‘

1.1.1

_,‘_¢f
.‘aJ.

Daurian Hedgehog 13 July 2005 Daurian Hedgehog 17 July 2005

 

 

Jerboa 26 June 2005 Eurasian Badger 17 July 2005

76

           

I"

Corsac Fox 15 July 2005

   

Corsac Fox 13 June 2005 Corsac Fox 5 July 2005

 

Corsac Fox 16 June 2005 Corsac Fox 13 July 2005

77

 

Corsac Fox 12 June 2005 Juvenile Corsac Foxs 28 June 2005

78

LITERATURE CITED

Alterio, N., and H. Moller 2000. Secondary poisoning of stoats (Mustela erminea) in a
South Island podocarp forest, New Zealand: implications for conservation.
Wildlife Research 27(5):501-508.

Analytical Sofiware. 2003. Statistix 8.0. Tallahasse.

Balasubramanyam, M., and KR. Purushotham. 1987. Comparative Effect of 3
Rodenticides Warfarin, Bromadiolone and Brodifacoum on the Indian Field-
Mouse, Mus-Booduga Gray. International Biodeterioration 23:307-314.

Batdelger, D., and E. Potapov. 2002. Mass mortality of birds in Mongolia. Falco:4-5.

Battur, B. Personal Interview. 3 June 2004, Plant Protection and Research Institute,
Ulaanbaatar.

Bemy P.J., T. Buronfosse, F. Buronfosse, F. Larnarque, and GB. Lorgue. 1997. Field
evidence of secondary poisoning of foxes (Vulpes vulpes) and buzzards (Buteo
buteo) by bromadiolone, a 4-year survey. Chemosphere: chemistry, biology and
toxicology as related to environmental problems 35(8): 1 817-1 829.

Bold. Personal Interview. 12 May2004, Institute of Biology, Mongolian Academy of
Sciences, Ulaanbaatar.

Brakes, CR, and RH. Smith. 2005. Exposure of non-target small mammals to
rodenticides: short-term effects, recovery and implications for secondary
poisoning. Journal of Applied Ecology 42(1):118-128.

Brown, R.A., A.R. Hardy, P.W. Greig-Smith, and P.J. Edwards. 1988. Assessing the
impact of rodenticides on the environment. European Plant Protection
Organization (EPPO) Bulletin 18:283-292.

Buckle, A.P. 1994. Rodent Control Methods: Chemical. Pages 127-160 in A.P. Buckle,
and RH. Smith, editors. Rodent Pests and Their Control. CAB International,
Wallingford.

Chalermchaikit, T., L.J. Felice, and M.J. Murphy. 1993. Simultaneous determination of
eight anticoagulant rodenticides in blood serum and liver. Journal of Analytical
Toxicology l7(1):56-61.

Cox, P., and R. H. Smith. 1992. Rodenticide Ecotoxicology: Pre-lethal effects of
anticoagulants on rat behaviour. Pages 165-170 in J. E. Borrecco, and R. E.
Marsh, editors. Proceedings of the 15th Vertebrate Pest Conference. University of
Davis, California.

79

Cox, RR, and RH. Smith. 1990. Rodenticide Ecotoxicology - Assessing Nontarget
Population Effects. Functional Ecology 4(3):315-320.

Damdin, Ts., and Sh. Dagva. 2004. Chemical Substance Consumption and the
Environment. Pages 49-54 in L. Dolgormaa, editor. Proceedings of the
International Seminar on Brandt's Vole Management. Orbis, Ulaanbaatar.

Davaasuren, J ., P. Shirbazar, and S. Buyanjargal. 2002. Inspection Report on the Process
and Results of Measures to Control Field Voles by Manual Methods in Kentii
Aimag. Ministry of Food and Agriculture, Ministry of Nature and Environment,
and Bureau of Food Security and Agricultural Protection, Ulaanbaatar. 4 pp.

Dolgormaa, L. Personal Interview. 15 May 2004, World Wildlife Fund Mongolia
Programme Office, Ulaanbaatar.

DuVall, M.D., M.J. Murphy, A.C. Ray, and J .C. Reagor. 1989. Case studies on second-
generation anticoagulant rodenticide toxicities in nontarget species. J oumal of
Veterinary Diagnostic Investigation l(l):66-8.

Eason, C.T., E.C. Murphy, G.R.G. Wright, and EB. Spurr. 2002. Assessment of risks of
brodifacoum to non-target birds and mammals in New Zealand. Ecotoxicology
1 1(1 ):3 5-48.

Eason, C.T., and EB. Spurr. 1995. Review of the toxicity and impacts of brodifacoum on
non-target wildlife in New Zealand. New Zealand Journal of Zoology 22(4):371-
379.

Erickson, W., and D. Urban. 2002. Potential risks of nine rodenticides to birds and
nontarget mammals: A comparative approach. US Environmental Protection
Agency, Washington, DC. 198 pp.

Fan, N., W. Zhou, W. Wei, Q. Wang, and Y. J iang. 1999. Rodent Pest Management in
the Qinghai-Tibet Alpine Meadow Ecosystem. Pages 285-304 in GR. Singleton,
L.A. Hinds, H. Leirs, and Z. Zhang, editors. Ecologically-based Management of
Rodent Pests. Australian Centre for International Agricultural Research,
Canberra.

Fisher, P., C.T. Eason, C.E. O'Connor, C.H. Lee, G.B. Smith, and S. Endepols. 2003.
Coumatetralyl residues in rats and hazard to barn owls. Pages 457-460 in GR.
Singleton, L.A. Hinds, C.J. Krebs, and D.M. Spratt, editors. Rats, Mice and

People: Rodent Biology and Management. Australian Centre for International
Agricultural Research, Canberra.

Fournier-Chambrillon, C., P.J. Bemy, O. Coiffrer, P. Barbedienne, B. Dasse, G. Delas, H.
Galineau, A. Mazet, P. Pouzenc, R. Rosoux, and P. Fournier. 2004. Evidence of
secondary poisoning of free-ranging riparian mustelids by anticoagulant
rodenticides in France: Implications for conservation of European mink (Mustela
lutreola). Journal of Wildlife Diseases 40:688-695.

80

Godfrey, M. E. R. 1985. Non-target and secondary poisoning hazards of "second
generation" anticoagulants. Acta Zoologica F ennica 173 2209-212.

Gratz, N.G. 1994. Rodents as Carriers of Disease. Pages 85-108 in A.P. Buckle, and RH.
Smith, editors. Rodent Pests and Their Control. CAB International, Wallingford.

Greaves, J .H. 1994. Resistance to Anticoagulant Rodenticides. Pages 197-217 in A.P.
Buckle, and RH. Smith, editors. Rodent Pests and Their Control. CAB
International, Wallingford.

Harrell, T.W., K.S. Latimer, P.J. Bain, and RM. Krimer. 2003. Anticoagulant
Rodenticide Toxicosis in the Dog and Cat. Department of Pathology, College of
Veterinary Medicine, University of Georgia. Retrieved January 28, 2005 from the
World Wide Web: http://www.vet.uga.eduNPP/clerk/l-Iarrell/indexhtm

Heptner, V.G., and AA. Sludskii. 1992. Mammals of the Soviet Union, 11, part 2:
Carnivora (hyenas and cats). Smithsonian Institution Libraries and National
Science Foundation, Washington DC. 784 pp.

Howald, G.R., P. Mineau, J .E. Elliott, and KM. Cheng. 1999. Brodifacoum poisoning of
avian scavengers during rat control on a seabird colony. Ecotoxicology 8:431-
447.

Jackson, W.B. 1977. Evaluation of Rodent Depredations to Crops and Stored Products.
European Plant Protection Organization (EPPO) Bulletin 7:439—458.

Jackson, W.B., and AD. Ashton. 1992. A review of available anticoagulants and their
use in the United States. Pages 156-160 in J .E. Borrecco, and RE. Marsh, editors.
Proceedings of the 15th Vertebrate Pest Conference. University of Davis,
California.

Kaukeinen, D.K. 1982. A review of the secondary poisoning hazard potential to wildlife
from the use of anticoagulant rodenticides. Pages 151-158 in R.E. Marsh, editor.
Proceedings of the 10th Vertebrate Pest Conference. University of Davis,
California.

Lechevin, J. C., and A. Vigie. 1992. Which useful toxicological information can be
drawn from studies on the hepatic fixation of anticoagulant rodenticides. Pages
204-207 in J .E. Borrecco, and RE. Marsh, editors. Proceedings of the 15th
Vertebrate Pest Conference. University of Davis, Newport Beach, California.

Lund, M. 1994. Commensal Rodents. Pages 23—43 in A.P. Buckle, and RH. Smith,
editors. Rodent Pests and Their Control. CAB International, Wallingford.

Lund, M., and A.M. Rasmussen. 1986. Secondary poisoning hazards to stone martens

(Martesfoina) fed bromadiolone-poisoned mice. Nordisk Veterinaerrnedicin
38:241-243.

81

Macdonald, D.W., and M G.P. Fenn. 1994. The Natural History of Rodents:
Preadaptations to Pestilence. Pages 1-21 in A.P. Buckle, and RH. Smith, editors.
Rodent Pests and Their Control. CAB Intemational, Wallingford.

Marsh, R.E., W.B. Howard, and W.B. Jackson. 1980. Bromadiolone - a New Toxicant for
Rodent Control. Pest Control 48:22-26.

Mcdonald, R.A., and S. Harris. 2000. The use of fumigants and anticoagulant
rodenticides on game estates in Great Britain. Mammal Review 30:57-64.

Mendenhall, V.M., and LP. Pank. 1980. Secondary poisoning of owls by anticoagulant
rodenticides. Wildlife Society Bulletin 8:311-315.

Mendjargal, B. Personal Interview. 13 May 2004, Ministry of Food and Agriculture,
Ulaanbaatar.

Morin, M.F., N. Merlet, G. Naulleau, and M. Dore. 1990. Primary Toxicity of
Bromadiolone on the Coypu. Bulletin of Environmental Contamination and
Toxicology 44:595-601.

Munkhtsog, B., and S. Ross. 2006. Pallas cat (Otocolobus manul) foraging ecology and
conservation in central Mongolia. Unpublished data.

Nahas, K. 1987. Kinetics of bromadiolone, anticoagulant rodenticide, in the Norway rat
(Rattus norvegicus). Pharmacol Res Commun 19:767-775.

Nowak, R. 1991. Carnivora: family Mustelidae. Pages 1104-1105 in Walker's Mammlas
of the World, Vol. 2, 5th Edition. John Hopkins University Press, Baltimore.

Osweiler, G.D., T.L. Carson, W.B. Buck, and GA. Van Gelder, editors. 1985. Clinical

and Diagnostic Veterinary Toxicology, 3rd Edition. Kendall/Hunt, Dubuque,
Iowa. 400 pp.

Parrnar, G., and PL. Batten. 1987. Detection of an Anticoagulant Binding-Site in Rat-
Liver Microsomes and Its Relevance to the Anticoagulant Effect. Human
Toxicology 62432-432.

Pech, R.P., G.M. Hood, G.R. Singleton, E. Salmon, R.I. Forrester, and RR. Brown. 1999.
Models for predicting plagues of house mouse (Mus domesticus) in Australia.
Pages 81-112 in GR. Singleton, L.A. Hinds, H. Leirs, and Z. Zhang, editors.
Ecologically-based Management of Rodent Pests. Australian Centre for
International Agricultural Research, Canberra.

82

Pech, R.P., S.A. Davis, and GR. Singleton. 2002. Outbreaks of rodents in agricultural
systems: pest control problems or symptoms of dysfunctional ecosystems? Pages
311-315 in G. R. Singleton, L.A. Hinds, C.J. Krebs, and D.M. Spratt, editors.
Rats, Mice and People: Rodent Biology and Management. Australian Center for
International Agricultural Research, Canberra.

Pelletier, K. Personal correspondence. 12 March 2006, The Philadelphia Zoo,
Philadelphia, PA.

Pelz, H.-J. 2002. Current approaches towards environmentally benign prevention of vole
damage in Europe. Pages 233-237 in GR. Singleton, L.A. Hinds, C.J. Krebs, and
D.M. Spratt, editors. Rats, Mice and People: Rodent Biology and Management.
Australian Center for International Agricultural Research, Canberra.

Poche, RM. 1988. Rodent Tissue Residue and Secondary Hazard Studies with

Bromadiolone. European Plant Protection Organization (EPPO) Bulletin 7:323-
330.

Proctor, D.L., editor. 1994. Grain storage techniques: Evolution and trends in developing
countries. Food and Agriculture Organization of the United Nations, Rome, Italy.

Pyper, W. 2001. Plain. Ecos 106:19-21.

Rudebeck, G. 1950. The Choice of Prey and Modes of Hunting of Predatory Birds with
Special Reference to Their Selective Effect. Oikos 2:65-88.

Samjaa, R., U. Zophel, and J. Peterson. 2000. The Impact of the Vole Mictorus brandti

on Mongolian Steppe Ecosystems. Marburger Geographische Schrifien 135:346-
360.

Shawyer, DR. 1987. The barn owl in the British Isles. The Hawk Trust, London, UK.
113 pp.

Shore, R.F., J .D.S. Birks, and P. Freestone. 1999. Exposure of non-target vertebrates to
second-generation rodenticides in Britain, with particular reference to the polecat
Mustela putorius. New Zealand Journal of Ecology 23:199-206.

Shore, R.F., J .D.S. Birks, P. Freestone, and A. C. Kitchener. 1996. Second-generation

rodenticides and polecats (Mustela putorius) in Britain. Environmental Pollution
91:279-282.

Smith, A.T., and J .M. Foggin. 1999. The plateau pika (Ochotona curzoniae) is a keystone
species for biodiversity on the Tibetan plateau. Animal Conservation 2:23 5-240.

Stone, W. B., J. C. Okoniewski, and J. R. Stedelin. 1999. Poisoning of wildlife with
anticoagulant rodenticides in New York. Journal of Wildlife Diseases 35:187-193.

83

Sullivan, D. 1988. Determination of the Environmental Fate of Ground Squirrel
Carcasses. Pages 169-173 in AC. Crabb, and RE. Marsh, editors. Proceedings of
the 13th Vertebrate Pest Conference. University of Davis, Lincoln, Nebraska.

Tasheva, M. 1995. Anticoagulant Rodenticides. United Nations Environment
Programme, International Labour Organisation, World Health Organization,
Geneva. 121 pp.

Tseveenmyadag, N., and B. Nayambayar. 2002. The impact of rodenticide used to control
rodents on Demoiselle Crane (Anthropoides Virgo) and other animals in
Mongolia. Unpublished Report. Mongolian Academy of Sciences and Peregrine
Fund. 7 pp.

Townsend, M.G., M.R. Fletcher, E.M. Odam, and PI. Stanley. 1981. An Assessment of
the Secondary Poisoning Hazard of Warfarin to Tawny Owls. Journal of Wildlife
Management 45:242-248.

WHO/FAO. 1996. WHO/FAO Data Sheets on Pesticides No. 88: BROMADIOLONE.
Available at http://www.inchem.orgfilocmnents/pds/pds/pestS8 e.htm.

Wood, B.J. 1994. Rodents in Agriculture and Forestry. Pages 45-83 in A.P. Buckle, and
RH. Smith, editors. Rodent Pests and Their Control. CAB International,
Wallingford.

WWF, WCS and ESBP. 2004. Recommendations on prevention and restriction of
Brandt's Vole outbreaks. Pages 17-23 in L. Dolgormaa, editor. Proceedings of the
International Seminar on Brandt's Vole Management. Orbis, Ulaanbaatar.

Zhang, Z. 2002. Rodent problems and management in the grasslands of China. Pages
316-319 in GR. Singleton, L.A. Hinds, C.J. Krebs, and D.M. Spratt, editors. Rats,
Mice and Peolpe: Rodent Biology and Management. Australian Center for
International Agricultural Research, Canberra.

Zhang, Z. B., R. Pech, S. Davis, D. Z. Shi, X. R. Wan, and W. Q. Zhong. 2003. Extrinsic
and intrinsic factors determine the eruptive dynamics of Brandt's voles Microtus
brandti in Inner Mongolia, China. Oikos 100:299-310.

Zhong, W., M. Wang, and X. Wan. 1999. Ecological Management of Brandt's vole
(Microtus brandtt) in Inner Mongolia, China. Pages 199-214 in GR. Singleton,
L.A. Hinds, H. Leirs, and Z. Zhang, editors. Ecologically-based Management of
Rodent Pests. Australian Centre for Intemation Agricultural Research, Canberra.

84