National Center tor Environmental Assessment
Office of Research and Development, Cincinnati, OH
Analysis of the Causes of a Decline
in the San Joaquin Kit Fox Population
on the Elk Hills, Naval Petroleum
Reserve #1, California
EPA/600/R-08/130 I November 2008 I www.epa.gov/ncea
&EPA
United States
Environmental Protection
Agency
-------
EPA/600/R-08/130
November 2008
Analysis of the Causes of a
Decline in the San Joaquin Kit Fox
Population on the Elk Hills, Naval
Petroleum Reserve #1, California
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
NOTICE
The document was produced by the U.S. Environmental Protection Agency's
Office of Research and Development. Dr. OFarrell's contributions were funded under
contract no. EP05C00179 with WATASH, L.L.C. The document has been subjected to
the Agency's peer and administrative review and has been approved for publication as
an EPA document. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
ABSTRACT
This report demonstrates the utility of the Causal Analysis/Diagnosis Decision
Information System (CADDIS) for determining the cause of biological impairments on
contaminated sites. The case is a decline in the abundance of a population of the
endangered San Joaquin kit fox on the Elk Hills Naval Petroleum Reserve, California,
between 1981 and 1986. This precipitous decline was a cause for concern at the time
because of its magnitude and because it was associated with an increase in oil
production on the site. Although multiple potential causes were investigated at the time,
the cause of the decline was not determined. This investigation proposed and analyzed
six candidate causes: prey abundance, habitat quality, predation, toxicants, accidents
and diseases. Evidence for each was analyzed using CADDIS's scoring system and 15
types of evidence. The conclusion is that predation by coyotes was the proximate
cause of the decline. Road kills contributed to the high mortality of foxes, but were
much less common. The decline in prey probably contributed to mortality by making the
foxes more susceptible to predation. As a model for causal analysis at contaminated
sites, this study was successful. Contaminants were eliminated as the cause and an
alternative was strongly supported by the evidence. In addition, this study
demonstrated the great utility of some types of evidence that had not previously been
used in CADDIS: mathematical modeling (a kit fox demographic model) and the
analysis of tissues (i.e., fur and blood analyses to eliminate toxicants and diseases,
respectively).
Preferred Citation:
U.S. Environmental Protection Agency (U.S. EPA). (2008) Analysis of the Causes of a Decline in the San
Joaquin Kit Fox Population on the Elk Hills, Naval Petroleum Reserve #1, California. U.S. Environmental
Protection Agency, National Center for Environmental Assessment, Cincinnati, OH. EPA/600/R-08/130.
Cover Photo:
San Joaquin kit fox, Elk Hills, California, T.P. O'Farrell.
ii
-------
TABLE OF CONTENTS
Page
LIST OF TABLES viii
LIST OF FIGURES ix
LIST OF ABBREVIATIONS x
PREFACE xi
AUTHORS, CONTRIBUTORS AND REVIEWERS xiii
ACKNOWLEDGMENTS xiv
EXECUTIVE SUMMARY xv
1. STEP 1. DEFINITION OF THE CASE 1
1.1. BACKGROUND 1
1.2. DESCRIPTION OF IMPAIRMENT 2
1.3. SPECIFIC EFFECTS 2
1.4. DESCRIPTION OF THE GEOGRAPHIC AREA UNDER
INVESTIGATION 6
1.5. DESCRIPTION OF THE REFERENCE SITES 8
1.6. OBJECTIVES OF THE INVESTIGATION 10
2. STEP 2. LIST CANDIDATE CAUSES 11
2.1. INITIAL LIST OF CANDIDATE CAUSES 11
2.2. STAKEHOLDERS 11
2.3. INFORMATION ON POTENTIAL SOURCES 11
2.4. CONCEPTUAL MODELS 13
2.5. FINAL LIST OF CANDIDATE CAUSES 13
2.5.1. Candidate Cause 1. Prey Abundance 18
2.5.2. Candidate Cause 2. Habitat Alteration 18
2.5.3. Candidate Cause 3. Predators 18
2.5.4. Candidate Cause 4. Toxic Chemicals 18
2.5.5. Candidate Cause 5. Vehicular activity 18
2.5.6. Candidate Cause 6. Disease 19
3. STEP 3. EVALUATE DATA FROM THE CASE 20
3.1. PREY ABUNDANCE 21
3.1.1. Spatial/Temporal Co-occurrence 21
3.1.1.1. Developed Versus Undeveloped 21
3.1.1.2. NPR-1 Versus NPR-2 21
iii
-------
TABLE OF CONTENTS cont.
Page
3.1.2. Temporal Sequence 26
3.1.3. Evidence of Exposure or Biological Mechanism 26
3.1.3.1. Prey Utilization 26
3.1.4. Stressor-Response Relationship in the Field 26
3.1.5. Causal Pathway—Disturbance 28
3.1.6. Causal Pathway—Climate 28
3.1.7. Causal Pathway—Competition 30
3.1.8. Manipulation of Exposure 30
3.1.9. Symptoms 30
3.1.9.1. Starvation 30
3.1.9.2. Reproductive 30
3.1.10. Mechanistic Sufficiency 31
3.2. HABITAT ALTERATION 31
3.2.1. Spatial/Temporal Co-occurrence—Disturbance 31
3.2.1.1. Disturbed Versus Undisturbed 31
3.2.1.2. Temporal Co-occurrence—Disturbance 31
3.2.2. Spatial/Temporal Co-occurrence—Climate 31
3.2.3. Temporal Sequence 32
3.2.4. Stressor-Response Relationships in the Field—Disturbance 32
3.2.5. Stressor-Response Relationships in the Field—Climate 32
3.2.6. Evidence of Exposure or Biological Mechanism 32
3.2.7. Causal Pathway—Disturbance 32
3.2.8. Causal Pathway—Climate 32
3.2.9. Mechanistic Sufficiency 33
3.3. PREDATORS 33
3.3.1. Spatial/Temporal Co-occurrence—Developed versus
Undeveloped 33
3.3.2. Spatial/Temporal Co-occurrence—Temporal on NPR-1 33
3.3.3. Spatial/Temporal Co-occurrence—NPR-1 versus NPR-2 33
3.3.4. Temporal Sequence 35
3.3.5. Stressor-Response in the Field 35
3.3.6. Causal Pathway 35
iv
-------
TABLE OF CONTENTS cont.
Page
3.3.7. Evidence of Exposure or Biological Mechanism 35
3.3.8. Manipulation of Exposure 36
3.3.9. Mechanistic Sufficiency 36
3.4. TOXIC CHEMICALS 36
3.4.1. Spatial/Temporal Co-occurrence 36
3.4.2. Temporal Sequence 38
3.4.2.1. Arsenic 38
3.4.2.2. Barium 38
3.4.2.3. Chromium 38
3.4.3. Stressor-Response Relationships in the Field 38
3.4.4. Evidence of Exposure or Biological Mechanism 38
3.4.4.1. Arsenic 39
3.4.4.2. Barium 39
3.4.4.3. Chromium 39
3.4.4.4. Sodium 43
3.4.4.5. Vanadium 43
3.4.5. Causal Pathway 43
3.4.5.1. Soil Concentrations 43
3.4.5.2. Soil Intake 44
3.4.5.3. Local Soil Contamination (Wastes) 44
3.4.5.4. Wastewater 44
3.4.5.5. Petroleum 44
3.4.6. Mechanistic Sufficiency 44
3.5. VEHICULAR ACTIVITIES 44
3.5.1. Spatial/Temporal Co-occurrence 44
3.5.2. Temporal Sequence 45
3.5.3. Evidence of Exposure or Biological Mechanism 45
3.5.4. Causal Pathway 45
3.5.5. Mechanistic Sufficiency 45
v
-------
TABLE OF CONTENTS cont.
Page
3.6. DISEASE 45
3.6.1. Spatial Co-occurrence 45
3.6.2. Temporal Sequence 45
3.6.3. Causal Pathway 46
3.6.4. Evidence of Exposure or Biological Mechanism 46
4. STEP 4. EVALUATE DATA FROM ELSEWHERE 47
4.1. PREY ABUNDANCE 47
4.1.1. Stressor-Response from Other Field Studies 47
4.2. HABITAT ALTERATION 47
4.3. PREDATORS 47
4.3.1. Stressor-Response from Other Field Studies 47
4.4. TOXIC CHEMICALS 47
4.4.1. Stressor-Response from Other Field Studies 47
4.4.2. Stressor-Response from Laboratory Studies 48
4.4.3. Stressor-Response from Other Studies—Fur 48
4.4.3.1. Arsenic 48
4.4.3.2. Chromium 48
4.4.3.3. Selenium 48
4.5. VEHICULAR ACTIVITIES 49
4.5.1. Stressor-Response from Other Field Studies 49
4.6. INCREASED DISEASE 49
4.6.1. Stressor-Response from Other Field Studies 49
5. STEP 5. IDENTIFY THE PROBABLE CAUSE 50
5.1. PROXIMATE CAUSES 50
vi
-------
TABLE OF CONTENTS cont.
Page
6. SOURCES 67
7 LATER STUDIES AND OTHER ATTRIBUTIONS OF CAUSE 70
8. CONCLUSIONS 71
9. LESSONS LEARNED 72
10. REFERENCES 75
vii
-------
LIST OF TABLES
No. Title Page
1 Age-Specific Risk of Death for Developed and Undeveloped Habitat Based
on Location at the Time of Death 4
2 Sites Considered in the Causal Analysis of the Kit Fox Decline 9
3 NPR-1 Developmental and Production Statistics, Fiscal Years 1976-1990 14
4 Candidate Causes 17
5 Relative Abundance of Lagomorphs (number observed) and Kangaroo
Rats (trapping success) in Two Habitats on Elk Hills, California, Between
1980-1984 23
6 Frequency of Occurrence (%) of Lagomorphs and Kangaroo Rats in the
Scats of San Joaquin Kit Foxes Collected in Three Habitats and Two Time
Periods Between 1980-1984, Elk Hills, California 27
7 Percentage of Radiocollared San Joaquin Kit Foxes Dying from Various
Causes on NPR-1 from 1980-1988 37
8 Ranges of Metal Concentrations (ppm) in Hair of Individual San Joaquin
Kit Foxes Sampled on the Elk Hills (NPR-1), Adjacent Buena Vista Hills
(NPR-2), Carizo Plain and Camp Roberts (Other Sites), California,
Compared with Concentrations in Hair from Other Mammals 40
9 Evidence for Prey Abundance (Candidate Cause 1) Caused by Disturbance
1 a, Climate 1 b, or Competition (1c) 51
10 Evidence for Habitat Degradation (Candidate Cause 2) Caused by
Disturbance 1 a or Climate 1b 54
11 Evidence for Predators (Candidate Cause 3) 56
12 Evidence for Toxic Chemicals (Candidate Cause 4) 58
13 Evidence for Vehicular Accidents (Candidate Cause 5) 60
14 Evidence for Disease (Candidate Cause 6) 62
15 Comparison of the Strength of Evidence for the Candidate Causes 64
viii
-------
LIST OF FIGURES
1 Minimum Numbers of Kit Foxes on NPR-1 and NPR-2 from Summer (S) and
Winter (W) Surveys 3
2 The Ratio of the Percentage of Female Kit Foxes Successfully Raising Pups
in Developed Habitat to the Percentage of Females Successfully Raising
Pups in Undeveloped Habitats as a Function of Time from 1980-1985 on
NPR-1 5
3 The Ratio of the Number of Litters Per Square Mile in Developed Habitat
to the Number of Litters Per Square Mile in Undeveloped Habitats as a
Function of Time from 1980-1985 on NPR-1 5
4 A Map Showing the Location of NPR-1 on the Elk Hills and its Context
Including the Buena Vista Oil Field (NPR-2), a Developed Reference Area,
and the Carrizo Plain, an Undeveloped Reference Area 7
5 Undisturbed Habitat on NPR-1 12
6 Disturbed Habitat on NPR-1 Showing Waste Water Sumps 12
7 A Conceptual Model of Three Potential Causes of the Decline in Kit Fox
Abundance That are Related to Natural History: (1) Prey Abundance,
(2) Habitat Alteration, and (3) Predation 15
8 A Conceptual Model of Toxic Chemicals as a Cause of the Decline in Kit
Fox Abundance 16
9 A Conceptual Model of Accidents as a Cause of the Decline in Kit Fox
Abundance 16
10 A Conceptual Model of Disease as a Cause of Decline in Kit Fox Abundance ..17
11 Number of Lagomorphs Observed on NPR-1 and Their Percentage in the
Diets of San Joaquin Kit Foxes From 1980-1984 22
12 Estimated Density of Lagomorphs on NPR-1 from 1983-1991 24
13 Lagomorph Density Estimates on NPR-2 25
14 Average Growing Season Precipitation for Bakersfield, California 29
15 Winter Visitation Indices for Coyotes on the Elk Hills (NPR-1) and Adjacent
Buena Vista Hills (NPR-2), California, 1985-1992 34
16 The Final Conceptual Model for the Cause of the Kit Fox Decline 69
ix
-------
LIST OF ABBREVIATIONS
As
Arsenic
BOPD
Barrels of oil per day
CADDIS
Causal Analysis/Diagnosis Decision Information System
EG&G
EG&G Energy Measurements Inc.
MER
Maximum efficient rate
NE
No Evidence
NPR
Naval Petroleum Reserve
U.S. DOE
U.S. Department of Energy
U.S. EPA
United States Environmental Protection Agency
U.S. FWS
U.S. Fish and Wildlife Service
x
-------
PREFACE
The U.S. Environmental Protection Agency's (2000) method for identifying the
causes of biological impairments of the Nation's waters was adapted and developed
from methods used in epidemiology for human health. Although it is conceptually
generic, the examples and guidance published in its web implementation, CADDIS
(http://www.epa.gov/caddis/). and elsewhere are aquatic. In addition, all of the case
studies have been aquatic. This report presents one of two case studies that
demonstrate the application of the method to contaminated terrestrial sites. This case
also extends the range of case studies in other ways. It is focused on a single
population of an endangered species, the San Joaquin kit fox, on the Elk Hills Naval
Petroleum Reserve, California. It focuses on a discrete temporal event, the precipitous
decline in the kit fox population between 1980 and 1986. It is relatively rich in the
amount of data available, the diverse types of data and the extended period over which
it was generated. Finally, the case relies heavily on mathematical modeling.
This is a cold case. The decline was documented by demographic studies
performed for the U.S. Department of Energy in the early 1980s, a period of increased
oil production. The range of studies and management activities increased in the late
1980s in response to the decline and concerns expressed by the U.S. Fish and Wildlife
Service. However, by the time a study of the factors controlling kit fox abundance was
performed, the original decline on the developed portions of the Elk Hills was no longer
the focus. Because it used large temporal and spatial scales, that study did not explain
the anomalous decline. Hence, it is appropriate to go back and investigate the cause of
the original event of concern.
Although the study was conducted to comply with the National Environmental
Policy Act (NEPA) and the Endangered Species Act, it is a model for studies of valued
populations on Superfund sites. As at Superfund sites, contaminants of concern were
identified by reviewing operational records and analyzing wastes and soils. Then
potential routes of exposure were determined. Finally, exposure was determined by
analysis of fox fur from the contaminated site and comparison to fur from four reference
areas and the literature.
The study is a success in that contaminants could be eliminated as a cause and
the actual proximate cause, predation by coyotes, was identified. It also confirmed that
the inferential approach and types of evidence developed for studies of impaired aquatic
communities were applicable to a terrestrial population. Finally, it provided some
lessons for future causal analyses. They include:
• Obtain data from multiple reference sites.
• Add new types of evidence to the methodology as needed.
• Use body burdens and (potentially) biomarkers when available.
• Determine the mandate for the causal assessment.
xi
-------
• Avoid spatially or temporally diluting causal relationships.
• Be clear about the spatial and temporal extent of the impairment.
In sum, this case study demonstrates that the same causal inference
methodology applies to terrestrial wildlife populations as to aquatic communities. This
result suggests that CADDIS can be usefully applied to biological impairments observed
on contaminated lands and to any documented decline in a population. It also suggests
that causal assessments, like other environmental assessments, are most successful
when data are abundant and data quality is high.
xii
-------
AUTHORS, CONTRIBUTORS AND REVIEWERS
AUTHORS
Glenn W. Suter II
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
Thomas P. O'Farrell
WATASH, L L C.
Boulder City, NV 89005
REVIEWERS
Susan Cormier
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
Michael Kravitz
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
EXTERNAL REVIEWERS
Rebecca A. Efroymson, Ph.D.
Oak Ridge National Laboratory
Environmental Sciences Division
Oak Ridge, TN 37831
Margaret McVey, Ph.D.
ICF International
Blue Hill, ME 04614
Gary Roemer, Ph.D.
New Mexico State University
Department of Fish, Wildlife & Conversation Ecology
Las Cruces, NM 88003
xiii
-------
ACKNOWLEDGMENTS
This study would not have been possible without the efforts of many other
individuals. The Elk Hills staff of EG&G Energy Measurements, Inc. generated the
demographic and natural history data. Particular thanks to Tom Kato for his help and
hospitality for Glenn Suter during his visits to the site. Kitty Gustin did most of the work
of sample collection, preparation and analysis for the toxicological study. Data analysis
and modeling for the original toxicological study were performed at Oak Ridge National
Laboratory by Aaron Rosen, John Beauchamp, Larry Barnthouse, and Sarah Floit.
xiv
-------
EXECUTIVE SUMMARY
The purpose of this report is to test the utility of the Causal Analysis/Diagnosis
Decision Information System (CADDIS; http://www.epa.gov/caddis/) to determine the
cause of effects on a population inhabiting a contaminated terrestrial site. CADDIS is a
web-based tool to implement the U.S. Environmental Protection Agency's (U.S. EPA)
Stressor Identification process (U.S. EPA, 2000). It was developed for determining the
causes of biological impairments in aquatic ecosystems under the Clean Water Act, and
the prior case studies have focused on effects on community metrics in streams.
However, the principles and methods of causal analysis should be applicable to all
environmental effects.
The case is the observed decline in abundance of a population of San Joaquin kit
foxes on the Naval Petroleum Reserve Number 1 (NPR-1) during the period 1980-1985.
NPR-1 is located on the Elk Hills, on the western edge of the San Joaquin Valley, west
of Bakersfield, California. It is an oil field that was held in reserve for the Navy until
1976 when Congress ordered that it be developed to produce at the maximum efficient
rate. A transect survey of wildlife was conducted in 1979 and then in 1980 a program of
demographic monitoring of the San Joaquin kit foxes began on the site. Because
hundreds of radio-collared foxes were monitored, the time and cause of death,
emigration rates, and the fecundity and breeding success of individual foxes could be
determined.
The San Joaquin kit fox (Vulpes macrotis mutica) is an endangered subspecies.
The minimum estimate of their abundance in the NPR-1 study area in summer, based
on capture-recapture estimates, declined from a high of 153 in 1981 to a low of 10 in
1991 (Harris et al., 1987; U.S. DOE, 1993). The kit fox decline occurred in the 1980-
1986 period. It appeared that the population was being negatively affected by
petroleum development activities. This 1981-1986 reduction in abundance, which
prompted the concerns in the U.S. Fish and Wildlife Service's 1987 biological opinion,
will be termed the decline. The population was stable from 1986-1989, but it declined
again from 1989-1991. No formal analysis of the cause of the 1980-1986 decline has
been conducted until now.
CADDIS is based on the comparison of evidence for alternative candidate
causes. The candidate causes for the decline are (1) prey abundance, (2) habitat
alteration, (3) predation, (4) toxic chemicals, (5) vehicular activity and (6) disease. In
addition, for each of the first two candidate causes, two causal pathways are
considered: from disturbance and from climate.
The analytical phase analyzes and scores evidence from the case and from
elsewhere for each candidate cause. Co-occurrence of the candidate causes and
effects were determined by comparing developed and undeveloped NPR-1 and by
comparing NPR-1 to NPR-2, a nearby oil field which apparently did not experience a kit
fox decline during the early 1980s. In addition, temporal co-occurrence at the site was
xv
-------
determined by considering when kit foxes and the candidate causes were either
increasing or decreasing. For causes that co-occurred with the decline or elements of
the causal pathways, regression analysis was used to try to develop exposure-response
relationships. In addition, the analysis included manipulations of exposures, symptoms
of exposure, and evidence of mechanistic sufficiency for some candidate causes.
Prey for kit foxes were primarily lagomorphs (black-tailed jackrabbits and desert
cottontails) and secondarily small mammals (kangaroo rats and pocket mice).
Lagomorphs declined at the same time as kit foxes and their decline was greatest in
developed areas. The abundances of kit foxes and lagomorphs were linearly
correlated. Foxes ate fewer lagomorphs and switched to small mammals. They
produced fewer pups and the litters were male biased, which is consistent with poor
nutrition. Lagomorph abundances were not correlated or were weakly correlated with
precipitation in both the current and previous year. Supplemental feeding increased
young-of-the-year survival.
The kit fox decline co-occurred with habitat disturbance by oil development, but
not with lack of rainfall. However, active oil development declined as the kit fox
population declined so there were no exposure-response relationships for habitat
alteration. There were good relationships between precipitation, plant production and
fox abundance during a later period of drought.
Coyotes increased in abundance on NPR-1 during the kit fox decline, particularly
in the developed areas. Increased mortality, particularly of young-of-the-year foxes,
was the cause of the decline and approximately 80% of the known mortalities were due
to coyotes. After a coyote control program began, coyote abundance declined and the
kit fox population stabilized.
Several toxic chemicals and petroleum leaks occurred in developed NPR-1
during the decline. However, elemental analyses of kit fox fur found that foxes from
developed NPR-1 were not highly exposed on average, and there was no correlation of
longevity with elemental concentrations. However, a few foxes on developed NPR-1
were relatively highly exposed to a few elements that were associated with oil
development. Three kit foxes had arsenic levels equivalent to humans with arsenic
poisoning, but they appeared to be healthy. Soils outside spills and sumps did not have
elevated levels of elements related to petroleum development.
Increased vehicle traffic inevitably accompanied increased development and
approximately 15% of identified kit fox mortalities were road kills. This contributed to
the increase in mortality that was shown by demographic modeling to be the probable
cause of the decline.
Observations of trapped foxes, necropsies and hematological and serological
studies showed no signs of an epizootic that would account for the decline.
-------
The synthesis phase of CADDIS determines the probable cause. The proximate
cause of the kit fox decline was found to be predation by coyotes. The evidence for
coyotes was strong and consistently positive, and evidence for all other candidate
causes was inconsistent.
The CADDIS methodology does not address sources, but the source must be
identified to inform management actions. However, the cause of the increase in coyote
abundance is unclear.
Other analyses have attributed variation in kit fox abundance to climate, but
precipitation was not particularly low during the decline. In particular, two very good
precipitation years occurred in the midst of the decline without influencing the decline in
fox abundance. In addition, climatic differences cannot account for the differences
between sites. This analysis focuses on a particular localized decline rather than larger
scale and longer term dynamics addressed by other analyses. In causal analysis,
spatial and temporal scales are critical.
The CADDIS methodology proved to be applicable to this terrestrial vertebrate on
a contaminated site. However, some modifications were made to accommodate the
case. The most important are the separate analysis of different causal chains to some
candidate causes and consideration of the source of the likely cause. Another is the
identification of a new type of evidence for use in causal analyses in CADDIS,
mechanistic sufficiency, to accommodate evidence from the demographic model that
linked the modes of action of candidate causes to the effect, reduced kit fox abundance.
Third, this case includes an additional synthesis step in which the inconsistencies in
evidence concerning some candidate causes was explained by their roles as links in the
causal pathway rather than as proximate causes. These changes were not made
because the case was terrestrial or because it focused on a population. Rather, they
relate to the peculiarities of the available evidence, particularly the abundance of data
and the opportunity to use demographic modeling.
xvii
-------
1. STEP 1. DEFINITION OF THE CASE
1.1. BACKGROUND
The San Joaquin kit fox (Vulpes macrotis mutica) was a relatively common
carnivore of the semi-arid habitats of California's San Joaquin Valley from San Joaquin
and Stanislaus counties south to Kern County (Grinnell et al., 1937). Starting in the
early 1900s agricultural, industrial, and urban developments brought about habitat loss
that led to population declines. In 1965 the California Fish and Game Commission
classified the San Joaquin kit fox as a protected furbearer. Following passage of the
Endangered Species Preservation Act of 1966, the Secretary of Interior listed the San
Joaquin kit fox as an endangered subspecies. In 1971 the San Joaquin kit fox was
classified as a "rare" species under the California Endangered Species Act of 1970. It
received Federal protection under the Endangered Species Act of 1973 (Public Law
93-205) as an endangered species.
The Elk Hills, Naval Petroleum Reserve #1 (NPR-1), Kern County, California,
was established in 1912. Production is believed to have begun in 1919 and peaked in
1921 at approximately 60,000 barrels of oil per day (BOPD). Production was steadily
reduced to an authorized rate of 2500 BOPD. Under the Naval Petroleum Reserves
Production Act of 1976 (Public Law 94-258) Congress directed the Secretary of the
Navy, and subsequently the Secretary of Energy, to produce petroleum products from
NPR-1 at the maximum efficient rate (MER) consistent with sound engineering
practices. An increase in development activities began in 1974 with the Total Capacity
Development Program, and expanded in 1976 to comply with the law (U.S. DOE, 1979).
In 1979 the U.S. Fish and Wildlife Service (U.S. FWS) notified the U.S. Department of
Energy (U.S. DOE) that the development activities on Elk Hills threatened the continued
existence of the San Joaquin kit fox and that formal consultation was required.
An integrated, multi-phased field program was designed to gather, synthesize,
and interpret ecological information necessary to determine whether U.S. DOE activities
on Elk Hills were compatible with the continued existence of the subspecies. During
July through September, 1979, transects totaling 522 miles were walked through all
sections of Elk Hills. San Joaquin kit fox dens were observed at a relative density of
approximately 9.2 per square mile. Kit fox dens were widely distributed, even in areas
of high relief and intense oil field activity. The prey base, indicated by relative densities
of jackrabbits, cottontails, and quail, was judged to be excellent. The Reserve provided
good habitat for a large proportion of the known, extant population (O'Farrell, 1980).
In 1980 U.S. DOE initiated a 15-year study to document the population dynamics
of the San Joaquin kit fox on Elk Hills conducted by EG&G Energy Measurements Inc.
It estimated abundance, reproduction, mortality, dispersal, and prey abundance, in both
developed and undeveloped habitats. That study documented a severe decline in kit
fox abundance apparently associated with the increase in petroleum development on
NPR-1 (O'Farrell et al., 1986). A 1987 Biological Opinion by the U.S. FWS
1
-------
recommended a study of potential toxic exposures and effects on Elk Hills, which was
conducted from 1988 to 1992 (Suter et al., 1992).
1.2. DESCRIPTION OF IMPAIRMENT
Following increased petroleum development activities on the Elk Hills, the
minimum number of San Joaquin kit foxes in the NPR-1 study area in summer, based
on capture-recapture estimates, declined from a high of 153 in 1981 to a low of 10 in
1991 (Harris et al., 1987; U.S. DOE, 1993) (Figure 1). Winter estimates declined from
165 to 19. Most of the decline occurred in the 1981 -1986 period. It appeared that the
population was being negatively affected by petroleum development activities. This
1981-1986 reduction in abundance, which prompted the concerns in the U.S. FWS's
1987 biological opinion, will be termed the decline. The population was slowly
increasing from 1986-1989, but it declined again from 1989-1991 (Figure 1). The period
1987-1991 is of interest primarily in terms of helping to understand the 1981-1986
decline.
1.3. SPECIFIC EFFECTS
Population dynamics of mammalian species are affected by four processes:
reproduction, mortality (or survival), emigration, and immigration. The NPR-1 study was
designed to study changes in these processes in both developed and undeveloped
habitats, and to assess the causes for any changes observed, especially if the causes
originated from petroleum development activities. However, the effect of concern was
the decline in kit fox abundance and the demographic parameters are explanatory.
From 1980-1986, juvenile kit foxes experienced much higher mortality rates on
developed than undeveloped NPR-1 due to predation and accidents (Table 1). Adults
experienced higher mortality in undeveloped areas, but the differences were smaller.
Zoellick et al. (1987) identified differences in reproduction between developed
and undeveloped areas of NPR-1 during 1980-1985. The proportion of radio-collared
adult vixens that successfully raised pups was 51% in developed habitats and 69% in
undeveloped habitats. The proportion of yearlings that successfully raised pups was
8% in developed habitats and 25% in undeveloped habitats. There were no trends on
the undeveloped area, but on developed NPR-1, the percent of vixens successfully
raising pups declined from 100% in 1980 to 33% in 1985, and the number of litters per
unit area also declined. Average litter size did not differ significantly between
undeveloped (4.1) and developed (4.4) habitats and there were no temporal trends in
litter size between 1980 and 1985. As a result, the reproductive success on developed
areas declined relative to undeveloped areas on both per female and per unit area
bases (Figures 2 and 3). Finally, the sex ratio of pups born on developed, but not
undeveloped, NPR-1 was biased toward males (M:F = 1.33).
The proportion of adults that dispersed during 1980-1986 was 18/78 (0.23) in
undeveloped habitats, and 23/99 (0.23) in developed habitats (Scrivner et al., 1987).
2
-------
200
"I 1 1 i 1 1 1 1 1 1 I I I I I I I I I I I T
swswswswswswswswswswsw
81 82 83 84 85 86 87 88 89 90 91
FIGURE 1
Minimum Numbers of Kit Foxes on NPR-1 and NPR-2 (see Section 1.5) from Summer
(S) and Winter (W) Surveys (Source: U.S. DOE 1993). The minimum population is the
sum of the individuals trapped during each trapping session, plus the number of
untrapped foxes that were known to be alive because they were trapped in a previous
and a subsequent session.
3
-------
TABLE 1
Age-Specific Risk of Death for Developed and Undeveloped Habitat Based on Location
at the Time of Death
Age at Death (years)
Cause
Developed
Undeveloped
0
Predation
0.58
0.36
Vehicle
0.25
0.00
Other
0.03
0.04
Unknown
0.41
0.36
1
Predation
0.25
0.39
Vehicle
0.00
0.08
Other
0.00
0.00
Unknown
0.16
0.21
2
Predation
0.27
0.36
Vehicle
0.00
0.04
Other
0.00
0.00
Unknown
0.06
0.16
3
Predation
0.22
0.37
Vehicle
0.15
0.00
Other
0.15
0.00
Unknown
0.00
0.11
4
Predation
0.27
0.39
Vehicle
0.00
0.39
Other
0.00
0.00
Unknown
0.27
0.22
Source: Floit and Barnthouse (1991).
4
-------
4.0
L0010 V -0.SI3-0.129 X .
** • 0.05
P - 0.001
s.o
2.0
Hi 1.0
0.5
K
<
0.1
FIGURE 2
The Ratio of the Percentage of Female Kit Foxes Successfully Raising Pups in
Developed Habitat to the Percentage of Females Successfully Raising Pups in
Undeveloped Habitats as a Function of Time from 1980-1985 on NPR-1 (Source:
Zoellick et al., 1987)
4.0
at
-J 3.0
LOG1fl Y • 0.307+0.407 X - O.OB3 X
R2 - 0.98
P - 0.003
X
U
m
<
2.0
9
O
«
m
s
m
H
K 0.5
w
IL
o
o
YEAR
FIGURE 3
The Ratio of the Number of Litters Per Square Mile in Developed Habitat to the Number
of Litters Per Square Mile in Undeveloped Habitats as a Function of Time from 1980-
1985 on NPR-1 (Source: Zoellick et al., 1987)
5
-------
Pup dispersal was 16/62 (0.26) from undeveloped habitats, and 32/67 (0.48) from
developed habitats (Scrivner et al., 1987). There appears to be net emigration of pups,
but not adults, from developed habitats to undeveloped habitats. However, data on
immigration were lacking.
In sum, the greater decline in abundance on the developed portion was
associated with greater mortality and greater emigration of young-of-the-year foxes as
well as lower fecundity of one year old females and lower reproductive success.
1.4. DESCRIPTION OF THE GEOGRAPHIC AREA UNDER INVESTIGATION
NPR-1 is located about 30 miles southwest of Bakersfield, Kern County,
California (Figure 4). It encompasses 47,245 acres, including most of the low foothills
of the Temblor Range known as the Elk Hills that extend southeastward into the San
Joaquin Valley. Elevations range between 290 feet above sea level on the valley floor
at the northeastern boundary, to 1551 feet along the main ridge in the western part of
the Reserve. The topography consists of gently rounded slopes with narrow divides, an
intricate system of highly dissected draws and dry stream channels in the higher
elevations, and gently rolling hills and flat valley land along the perimeter. Almost all of
the petroleum developments on NPR-1 are located in the central uplands.
Undisturbed surface soils are predominately sandy loams that support relatively
good vegetative cover. These deep, well drained, sandy loams belong to either the Elk
Hills series or the Cajon and Kimberlina series. Soils over large areas consist of highly
variable, undeveloped stratified Torriorthents that have many saline-alkaline areas
which support little or no vegetation.
NPR-1 and the San Joaquin Valley have a Mediterranean climate characterized
by relatively cool, wet winters and hot, dry summers. About 89% of the mean annual
precipitation (5.72 inches) falls during the growing season (November-April). Low
rainfall is supplemented by a high incidence of fog and high humidity during the winter.
The annual mean maximum temperature is 78°F; the mean minimum temperature is
52°F.
Vegetation on the Reserve is part of a very broad type called Valley Grassland
that surrounds the agricultural land of the Central Valley. In its pristine condition much
of the Valley Grassland probably consisted of perennial bunchgrasses and an overstory
of shrubs, but European plant introductions and livestock grazing converted almost all of
it to an annual grassland in the mid to late 19th century. Vegetation patterns were
further altered by heavy sheep grazing between the late 1860s and 1965 when the
Navy, which operated the Reserve prior to the U.S. DOE, eliminated grazing leases.
NPR-1's dominant ground cover is red brome, Bromus rubens, an introduced annual
grass. The dominant shrub is desert saltbush, Atriplex polycarpa, which is especially
dense in disturbed areas such as along roadsides and edges of well pads.
6
-------
Midway-Sunset
\ Oil Field
Caliente Range
Buttonwillow
State Highway 58
McKittrick
Naval Petroleum
t- Reserve 1
St. Luis Obispo Co;
Buena Vista
Lake Bed
x. Buena Vista v
n\OH Field
Midway-Sunset |
\ Oil Field :
Caliente Range
Maricopa
FIGURE 4
A Map Showing the Location of NPR-1 on the Elk Hills and its Context Including the
Buena Vista Oil Field (NPR-2), a Developed Reference Area, and the Carrizo Plain, an
Undeveloped Reference Area (Source: U.S. DOE, 1993)
7
-------
The Reserve supports 23 species of mammals, 83 species of birds, 8 species of
reptiles, and 2 species of amphibians.
NPR-1 has been operated by the federal government as an oil field since 1912.
In 1976 at the beginning of MER there were already 1286 wells onsite (U.S. DOE,
1993). Subsequently an additional 1029 wells were drilled and support facilities were
expanded and upgraded.
Analysis of causal relationships on NPR-1 is limited by the confounding of
topography (uplands and lowlands) and degree of development (developed and
undeveloped). As described above, the majority of petroleum developments were in the
uplands, and the majority of the lowlands were undeveloped. There were insufficient
areas of either developed lowlands or undeveloped uplands to distinguish those factors.
Also, the area was not pristine prior to recent disturbances. Oil development affected
the site to some degree since the early 20th century. Measurements of conditions prior
to petroleum developments or prior to 1974 were unavailable. The analysis is also
limited because kit foxes are highly mobile and can have home ranges that include both
developed and undeveloped habitats. Hence, it was necessary to consider spatially
disjoined reference sites.
The descriptions in this section of the natural features of NPR-1 were adapted
from O'Farrell et al. (1986).
1.5. DESCRIPTION OF THE REFERENCE SITES
This causal analysis uses three reference sites. One is a nearby oil field and the
other two have no oil development (Table 2).
Naval Petroleum Reserve No. 2 (NPR-2) occupies the Buena Vista Hills, south of
NPR-1 (Buena Vista Oil Field in Figure 4). It has lower elevations but is ecologically
similar. However, its oil resources were developed earlier then those on NPR-1, and,
although some oil development activities occurred there in the 1980s, there was no
increase in production as on NPR-1. Kit fox demographic studies on NPR-2 began in
1983. The population was apparently stable until 1988, but declined thereafter
(Figure 1). It is likely that some movement of foxes between the two reserves occurs,
but the fact that the decline on NPR-1 in the early-to-mid 1980s was not mirrored on
NPR-2 suggests that they are not a single population. NPR-2 was a reference area for
the study of chemical exposures and for activities and physical disturbances associated
with increased oil production.
The Carrizo Plain (now a National Monument) lies south of NPR-1 and NPR-2,
beyond the Temblor Range in San Louis Obispo County (Figure 4). It is primarily
grassland, supporting some cattle grazing. It has no oil production.
Camp Roberts is a California Army National Guard training site in San Louis
Obispo and Monterey Counties between the Salinas River floodplain and the Santa
8
-------
TABLE 2
Sites Considered in the Causal Analysis of the Kit Fox Decline
Site
Development
Kit Foxes
Status
Elk Hills/
NPR-1
developed
Oil field with active drilling, facility
construction and oil production
during the decline
Primary site of
the decline
The site of
the case
Elk Hills/
NPR-1
undeveloped
Low density of oil development
with pipe lines and other support
facilities
The decline was
later and less
intense
Near field
negative
reference
Buena Vista
Hills/ NPR-2
Heavily developed prior to the
period of concern. Little oil
production and little active
development
No apparent
decline in the
period of concern
Positive
reference
Carrizo Plain
No oil development
Cattle grazing
No apparent
decline in the
period of concern
Far field
negative
reference
Camp Roberts
No oil development
Military training activities
No apparent
decline in the
period of concern
Far field
negative
reference
9
-------
Lucia Mountains. It encompasses 172 km2 of primarily rolling hills with grassland, oak
woodland and chaparral. It has no oil production.
1.6. OBJECTIVES OF THE INVESTIGATION
Although data were collected and analyzed to support a determination of the
cause of the apparent decline of the San Joaquin kit fox population on Elk Hills in the
1980s, a complete causal analysis was not performed at that time. This investigation
applies the Causal Analysis/Diagnosis Decision Information System (CADDIS;
http://www.epa.gov/caddis/) to the problem. CADDIS is a web-based tool to implement
the U.S. EPA's Stressor Identification process (U.S. EPA, 2000). This investigation
reanalyzes the data that were available in the 1980s to determine the cause of a decline
in the abundance of San Joaquin kit foxes on NPR-1. Data were available from
EG&G's kit fox demographic study which included a research program by Oak Ridge
National Laboratory, Tennessee, to determine whether toxic chemicals from oil
development were a plausible cause of the decline. The results of those studies
satisfied concerns of the California Fish and Game Commission and U.S. FWS, but no
formal analysis was conducted at that time to identify the cause of the decline. This
reanalysis serves as a test of the applicability of CADDIS to an impairment of a wildlife
population on a contaminated terrestrial site. Although CADDIUS was developed
primarily to address impairments of aquatic biotic communities, there is no reason that it
would not serve to address populations and terrestrial cases as well. The report ends
with a consideration of data collected after 1990 and of subsequent studies of the
factors controlling kit fox demographics at larger scales as a check of the
reasonableness of the CADDIS results.
10
-------
2. STEP 2. LIST CANDIDATE CAUSES
2.1. INITIAL LIST OF CANDIDATE CAUSES
Six potential candidate causes were identified in the 1980s: altered food
supplies, habitat alteration, predators, toxic chemicals, vehicular activity, and disease.
2.2. STAKEHOLDERS
Two types of stakeholders provided the impetus and funding for the studies of kit
fox demographics and the causes of decline. The natural resource trustees, the U.S.
FWS and the California Department of Fish and Game, were concerned that habitat
disturbance and toxic chemicals associated with increased oil production were harmful
to kit foxes. The concerns of the U.S. FWS were expressed in biological opinions
released in 1980 and 1987. The U.S. DOE and Chevron U.S.A. were also concerned
with environmental protection and they funded the studies, but their primary mission
was increasing oil production. They emphasized predation and the influence of climate
on habitat and prey availability. All causes advocated by stakeholders were considered.
2.3. INFORMATION ON POTENTIAL SOURCES
Petroleum development involved removal of vegetation which potentially altered
food supplies (candidate cause 1) and kit fox habitat (candidate cause 2). In addition,
metals and other petroleum-associated contaminants (candidate cause 4) would occur
in the developed areas. Accidents (candidate cause 5) were primarily road kills; the
length of roads and the amount of traffic was believed to be increased by development.
It was hypothesized that development might attract coyotes (candidate cause 3) and
coyotes or workers might carry diseases (candidate cause 6). Hence, the candidate
causes were spatially confounded, because they were all hypothesized to be primarily
associated with developed areas.
Oil production is a source of disturbance in the form of devegetated areas.
Sections, and in some cases half sections, of the study area were classified as
undeveloped or developed based on the area of land disturbed by oil field development
(well pads, sumps, roads, pipelines, pipe storage yards, facilities). The areal percent
disturbance was calculated by overlaying transparent dot grids (Bryant, 1943; Mosby,
1980) on 1:10,000 scale aerial photographs taken in 1983, and counting the proportion
of dots that overlapped disturbances. When the proportions were graphed the
distribution of numbers was bimodal. There was a clear demarcation between the two
bell curves at about 15%. For purposes of the study, sections or half sections with
greater than 15% of the land area disturbed were considered to be developed habitat
(Figures 5 and 6).
The study area occupied 44.5 square miles of NPR-1 and a small adjacent
buffer: 18.5 square miles were undeveloped and 26 square miles were developed.
11
-------
FIGURE 5
Undisturbed Habitat on NPR-1 Photo by T.P. O'Farrell, February 1984.
FIGURE 6
Disturbed Habitat on NPR-1 Showing Waste Water Sumps. Photo by T.P. O'Farrell,
June 1980.
12
-------
approximately 3.6 acres disturbed per well, 3334 acres were disturbed for MER through
1983 and 3704 acres through 1988 or approximately 8% of NPR-1.
In addition to this passive measure of disturbance, land area disturbed, active
sources of disturbance must be considered. These include active well drilling and
associated construction of roads, pipelines and other facilities, which produce noise,
dust, chemical spills, vehicle traffic, and human presence as well as removal of
vegetation. These development activities may be represented by the drilling rate
(Table 3). Between 1974-1983, 93 wells were drilled per year on average, after which
the average rate dropped to 26 per year (U.S. DOE, 1993). Operation and maintenance
activities may be indicated by rates of oil production which climbed rapidly from
1976-1982 and then declined (Table 3).
Oil development is a source of various potentially toxic materials. Therkelsen
(1972) studied wildlife conservation problems in the petroleum fields of Kern County,
California, and reported that dissolved solids, salts, and other minerals caused deaths,
nervous disorders, diarrhea, and decreased reproduction in livestock and wildlife. Suter
(1988) reviewed the literature and records at NPR-1 and determined that potentially
toxic materials were used or produced on the site, that livestock and wildlife effects had
been documented in the past, and that potential routes of exposure for kit foxes were
present on the site. Sources were widely distributed in the developed portions of the
site.
2.4. CONCEPTUAL MODELS
Conceptual models represent the relationships among potential sources, routes
of transport and exposure, proximate causes and effects. Sources are represented by
parallelograms, intermediate steps in the causal process are represented by rounded
rectangles, proximate causes are represented by rectangles, mechanisms are
represented by hexagrams and the endpoint effect in all cases is "kit fox abundance."
The first conceptual model (Figure 7) presents the consequences of changes in
habitat, prey or predators resulting from either anthropogenic disturbances or reduced
precipitation. The second conceptual model (Figure 8) represents the induction of toxic
effects due to exposure to chemicals associated with oil production. The third
conceptual model (Figure 9) represents acute lethality due to either traffic accidents or
accidents during oil development. The last conceptual model (Figure 10) represents
disease-induced death or infertility due to a combination of exposure to a pathogen and
susceptibility. Potential sources of exposure include coyotes and humans carrying
pathogens from their pets.
2.5. FINAL LIST OF CANDIDATE CAUSES
The final candidate causes, including subcauses associated with different
sources, are listed in Table 4 and described in the following sections.
13
-------
TABLE 3
NPR-1 Developmental and Production Statistics, Fiscal Years 1976-1990
Fiscal Year
Crude Oil (106 bbl)
Developmental Wells
1974-1976
NA
258
1976
3.8
NA
1977
36.9
168
1978
43.5
120
1979
52.6
82
1980
58.3
64
1981
62.6
46
1982
60.7
101
1983
57.4
87
1984
50.5
30
1985
47.7
22
1986
42.2
22
1987
39.8
29
1988
39.2
NA
1989
35.5
NA
1990
29.5
NA
Total
660.5
1029
Source: U.S. DOE (1993).
NA = Not available
14
-------
j Disturbance/ j j
III Precipitation/
2.
Kit Fox
Habitat Quality
«
Vegetation
Cover and
Quality
Lagomorph
Habitat Quality
1.
Prey
Abundance
Coyote
Habitat Quality
/Control I
Program /
<
Food
Consumption
Time Hunting
Coyote
Abundance
,3
Immigration/
Emigration
><
Reproduction > < Starvation
Predation
/ ^ s
f Kit Fox A
Abundance J
Mortality
J
FIGURE 7
A Conceptual Model of Three Potential Causes of the Decline in Kit Fox Abundance
That are Related to Natural History: (1) Prey Abundance, (2) Habitat Alteration, and (3)
Predation
15
-------
Spills
Land
Applicatio
Waste
Ponds
li_E
5.
Exposure to
Chemicals
7
<
>
<
>
Reproductive
Decrement
Mortality
Kit Fox
Abundance
FIGURE 8
A Conceptual Model of Toxic Chemicals as a Cause of the Decline in Kit Fox
Abundance
/ Traffic r /Devebpmen/
1
Accidents
<
>
Mortality
C
)
Kit Fox
Abundance
FIGURE 9
A Conceptual Model of Accidents as a Cause of the Decline in Kit Fox Abundance
16
-------
/coyotes 7 L Humans//
Unknown
Reproductive
Decrement
Mortality
f Kit Fox A
^Abundance J
Disease
Susceptibility
Exposure to
Patogens
FIGURE 10
A Conceptual Model of Disease as a Cause of Decline in Kit Fox Abundance
TABLE 4
Candidate Causes
Number
Description
1a
Reduced prey abundance due to habitat disturbance
1b
Reduced prey abundance due to climate
1c
Reduced prey abundance due to predation
2a
Kit fox habitat alteration due to disturbance
2b
Kit fox habitat alteration due to climate
3
Predation on kit foxes
4
Toxic effects on kit foxes
5
Accidents involving kit foxes
6
Diseases of kit foxes
17
-------
2.5.1. Candidate Cause 1. Prey Abundance
This candidate cause refers to a reduction in prey abundance due to reduced
vegetation cover or quality which in turn may be due to disturbance during oil
development (Cause 1a), to climatic effects, particularly reduced precipitation
(Cause 1 b), or to coyote competition for kit fox prey species (Cause 1 c) (Figure 7). This
cause includes changes in the relative abundance of prey species, particularly declines
in lagomorphs (black-tailed jackrabbit and desert cottontail) relative to small rodents
(primarily kangaroo rats and pocket mice).
2.5.2. Candidate Cause 2. Habitat Alteration
This candidate cause refers to direct effects of habitat quality on kit foxes
including the abandonment of the site by foxes seeking more acceptable habitat or to
reduced reproductive success due to fewer adequate denning sites. In addition to
physical disturbance of the soil and vegetation, human activities may cause stress,
disruption of hunting and increased energy expenditure. Activities close to whelping
and pupping dens might disturb vixens and cause them to neglect or even abandon
their litters. Like Candidate Cause 1, habitat alteration may be ultimately caused by
disturbance during oil development (Cause 2a) or to climatic effects, particularly
reduced precipitation (Cause 2b) (Figure 7). Also, habitat alteration may be cumulative
(e.g., the total area devegetated by development) or immediate (e.g., the effects of
active construction and drilling activities on the willingness of foxes to use an area).
2.5.3. Candidate Cause 3. Predators
This candidate cause refers to increased abundance of coyotes resulting in
increased competition or mortality of foxes by coyotes (bobcats are also potential
predators but were much less abundant). Oil development may make the site more
attractive to coyotes by increasing road kills and food waste to be scavenged and, until
the control program began, by protecting coyotes from hunters.
2.5.4. Candidate Cause 4. Toxic Chemicals
This candidate cause refers to toxic effects on the foxes due to exposure to
chemicals associated with oil development. The two principal sources were spills of oil
or chemicals used in production activities or waste ponds that contained produced water
(water pumped up with the petroleum) (Figure 8).
2.5.5. Candidate Cause 5. Vehicular activity
This candidate cause refers to kit fox mortality due to being struck by vehicles or
injured by equipment during oil production. Increased oil production increased vehicle
traffic and construction activities that may bury foxes in their dens (Figure 9).
18
-------
2.5.6. Candidate Cause 6. Disease
This candidate cause refers to any of various diseases of canids that may have
been endemic or may have been brought to the site by coyotes or by humans from their
pets (Figure 10).
19
-------
3. STEP 3. EVALUATE DATA FROM THE CASE
The case consists of the kit fox population on the Elk Hills (NPR-1) in the first half
of the 1980s. Three spatial comparisons are possible (Table 2). (1) The NPR-1 site
was divided into developed and undeveloped areas, which allows the comparison of
areas in which foxes were directly exposed to drilling, construction, and other
development activities during the surge in oil production and areas where there was
very little development activity. (2) Comparisons could be made between NPR-1 as a
whole and a reference site, the Buena Vista Hills (NPR-2). This is a comparison of an
actively developing oil field and one that is in production but where little new
development was occurring. (3) Comparisons can be made between the developed
and undeveloped portions of both oil fields combined (NPR-1 and NPR-2). This
comparison incorporates the loss of habitat due to oil development, but not the effects
of active development.
Although no baseline period is available to allow comparison of the development
period with a pre-development period, temporal comparisons are possible. In the period
1981-1986, the NPR-1 kit fox population declined rather precipitously but the NPR-2
population was relatively stable at a high level (Figure 1). The NPR-2 population
decreased precipitously from 1987-1991 while the NPR-1 population increased slightly
until 1989 and then resumed its decline until 1991. Hence, we are interested in what
happened in the early-to-mid 1980s on NPR-1 that did not occur on NPR-2.
Each line of evidence is given a score, as follows:
+++
convincingly supports
convincingly weakens
++
strongly supports
—
strongly weakens
+
somewhat supports
—
somewhat weakens
0
neither supports nor weakens
NE
no evidence
For each of the first two candidate causes, (1) prey abundance and (2) habitat
alteration, two sources were considered (a) disturbance and (b) climate. Where
evidence permits, the candidate cause was scored, and each causal pathway from a
source to a cause was scored. Note that, for a particular piece of evidence, the score is
never higher and is usually lower for the full pathways (e.g., 1b. from climate to prey
abundance) than for the cause (e.g., 1. prey abundance), because a more complex
hypothesis requires stronger evidence to achieve the same degree of belief. The logic
is the same as the logic that requires that joint probabilities must be smaller than simple
probabilities.
20
-------
This case uses a type of evidence not found in prior implementations of Stressor
Identification or CADDIS, demographic modeling of the affected kit fox population. Floit
and Barnthouse (1991) created a projection matrix model of the NPR-1 population for
the period 1981-1986, the period of decline. A similar model created by EG&G for the
supplemental impact statement extended to 1989, so it included the recovery period
(U.S. DOE, 1993). These models gave qualitatively similar results, but some rates were
quite different. Because these models show which mechanisms were sufficient to
cause the decline, this type of evidence from the site is called mechanistic sufficiency.
The evidence from the site for each candidate cause is presented in this section.
Evidence from the site addresses the issue, did the candidate cause, in fact, cause the
effect in this case. In the following section (Section 4) the evidence from elsewhere is
presented for each candidate cause. That evidence addresses the issue-is the
candidate cause capable of causing effects of this type? In Section 5, the evidence for
each candidate cause is summarized and compared. As you read the evidence, you
may wish to look ahead to the summaries of the evidence in Tables 9-14.
3.1. PREY ABUNDANCE
3.1.1. Spatial/Temporal Co-occurrence
Prey abundance is judged to co-occur with the kit fox decline if prey abundance
was low where and when the kit fox decline occurred. Two comparisons were possible.
3.1.1.1. Developed Versus Undeveloped
Lagomorphs, initially the primary prey of kit foxes on Elk Hills, declined in both
developed and undeveloped habitats (1980-1984 based on road surveys), but the
decline was much greater (5.3x) in the developed area where they were more abundant
than in the undeveloped area (1,9x) (Figure 11, Table 5). Kangaroo rat abundances did
not show a trend, but were much higher in the undeveloped area (Table 5). Hence, a
decline in the principal prey co-occurred in space with disturbance and with the most
rapidly declining component of the kit fox population, which supports prey abundance as
a cause (1 = ++). This evidence supports prey abundance through the disturbance
pathway as a cause (1a = +). It does not support the climatic pathway because the
climate did not differ between areas of NPR-1. However, that evidence is weak
because local weather or soil moisture data were not available (1b = -).
3.1.1.2. NPR-1 Versus NPR-2
Transect surveys from 1983-1991 showed a consistent decline in jack rabbits for
NPR-1 as a whole (Figure 12). On NPR-2, lagomorph densities did not decline until
1987 but then declined until 1991 (U.S. DOE, 1993) (Figure 13). That is consistent with
the delay in onset of kit fox decline on NPR-2 relative to NPR-1 (Figure 1). Hence, the
declines in kit foxes on both NPR sites co-occurred with declines in lagomorph prey,
which supports prey abundance as a cause (1 = ++), but the declines were not
21
-------
Observed
FIGURE 11
Number of Lagomorphs Observed on NPR-1 and Their Percentage in the Diets of San
Joaquin Kit Foxes From 1980-1984 (Source: U.S. DOE, 1993)
22
-------
TABLE 5
Relative Abundance of Lagomorphs (number observed) and Kangaroo Rats (trapping success) in Two Habitats on Elk
Hills, California, Between 1980-1984
Months
Year
Undeveloped
Developed
Number of
Lagomorphs
Observed
Kangaroo Rats
Number of
Lagomorphs
Observed
Kangaroo Rats
Trapping Effort
(trap-nights)
Trapping
Success(%)
Trapping Effort
(trap-nights)
Trapping
Success(%)
Jun-Nov
1980
139
250
41.6
850
675
5.0
Dec-May
1981
133
900
54.3
286
1425
10.2
Jun-Nov
1981
103
900
41.7
630
1598
2.1
Dec-May
1982
112
900
45.9
182
1700
3.3
Jun-Nov
1982
115
900
34.3
246
1800
2.4
Dec-May
1983
133
900
66.6
142
1800
5.7
Jun-Nov
1983
89
899
38.9
282
2399
2.0
Dec-May
1984
84
450
64.4
224
1350
5.9
Jun-Nov
1984
71
300
57.3
160
900
7.6
Source: Scrivner et al. (1987).
-------
8 600
Jackrabbits
Cottontails
O 400
£ 300
O 200
83 84 85 86 87 88 89 90 91
Year
FIGURE 12
Estimated Density of Lagomorphs on NPR-1 from 1983-1991 (Source: U.S. DOE, 1993)
24
-------
400-|
SPRING
SUMMER
500-
400-
300-
800"
100-
: ALL
0-
)-
1905
1986
1987
FIGURE 13
Lagomorph Density Estimates on NPR-2 (Source: U.S. DOE, 1993)
25
-------
contemporaneous. This evidence supports prey abundance through the disturbance
pathway (1a = +). It does not support the climatic pathway, but without local weather or
soil moisture data the evidence is weak (1 b = -).
3.1.2. Temporal Sequence
Since the decline in both lagomorphs and foxes appears to have been underway
at the beginning of the time series, it is not possible to determine whether a decline in
prey began before the decline in foxes. Temporal sequence might also be derived from
a time series, if there were a consistent lag between a decline in abundance of prey and
a decline in kit foxes. However, the steady decline in both predators and prey during
the period of concern precludes the identification of a clear temporal sequence
(Figures 1,11 and 12). As a result, the correlations of lagomorph and fox abundance
are not consistently better with a one year time lag than without (see Stressor-
Response Relationship in the Field, below). The temporal sequence is undefined (0).
3.1.3. Evidence of Exposure or Biological Mechanism
3.1.3.1. Prey Utilization
During the period of decline, the proportion of fecal samples from NPR-1
containing fur of lagomorphs decreased and kangaroo rats, usually the secondary prey,
increased in developed and undeveloped areas (Table 6). This indicates changes in
prey utilization that are consistent with a decline in preferred prey and switching to
secondary prey in all areas. This evidence is clear and consistent with declines in prey
abundance as a cause (1 = ++). Since it occurred in developed and undeveloped areas
it is consistent with the climate pathway, but not disturbance (1a = - & 1b = +).
3.1.4. Stressor-Response Relationship in the Field
Kit fox abundance was linearly related to lagomorph abundance in the previous
year on developed NPR-1 during 1981-1985 based on road surveys (r2 = 0.68 for y =
-230 + 6.Ox) and less well related to lagomorph abundance in the same year (r2 = 0.31).
Kit fox abundance was even better related to lagomorph abundance in the same year
on undeveloped NPR-1 during 1981-1985 based on road surveys (r2 = 0.98 for y = 8.2 +
0.69x) and less well related to lagomorph abundance in the previous year (r2 = 0.66).
Kit fox abundance was highly linearly correlated with jack rabbit abundance in the
same year on NPR-1 during 1983-1991, based on transect surveys (r2 = 0.89 for y =
19 + 0.15x) and less well correlated with jack rabbit abundance in the previous year
(r2 = 0.56). In both cases, the relationship is due to the first and last two years of the
series.
In sum, the decline of foxes and of lagomorphs on both developed and
undeveloped NPR-1 in the 1980s results in multiple good stressor-response
relationships from two different lagomorph surveys, with or without a time lag. This
26
-------
TABLE 6
Frequency of Occurrence (%) of Lagomorphs and Kangaroo Rats in the Scats of San Joaquin Kit Foxes Collected in
Three Habitats and Two Time Periods Between 1980-1984, Elk Hills, California
Habitat
Year
Frequency of Occurrence (%)
Dec-May
Jun-Nov
Sample
Size
Lagomorphs
Kangaroo
Rats
Sample
Size
Lagomorphs
Kangaroo
Rats
Undeveloped Flat
1980
5
100.0
0.0
4
100.0
0.0
1981
26
84.6
3.8
60
78.3
13.3
1982
76
77.6
19.7
31
45.2
32.3
1983
64
39.1
45.3
33
27.3
42.4
1984
22
36.4
50.0
32
43.8
37.5
Undeveloped Hilly
1980
17
100.0
5.9
5
60.0
0.0
1981
49
87.8
14.3
21
81.0
0.0
1982
51
78.4
9.8
15
33.3
20.0
1983
49
61.2
20.4
13
76.9
15.4
1984
12
58.3
33.3
29
51.7
20.7
Developed Hilly
1980
5
100.0
0.0
24
91.7
4.2
1981
122
88.5
4.9
108
85.2
0.9
1982
176
82.7
2.8
78
67.9
5.1
1983
69
81.2
7.2
22
40.9
18.2
1984
17
47.1
29.4
28
57.1
17.9
Source: Scrivner et al. (1987).
-------
result is consistent with loss of prey as a cause (1 = ++) and, because the correlations
occurred in both developed and undeveloped areas, with the climate pathway (1b = +)
but not disturbance (1a = -).
3.1.5. Causal Pathway—Disturbance
The primary pathway for disturbance is from oil development activities to reduced
vegetation, reduced prey and reduced kit foxes (Figure 7). The creation of well pads
and other construction activity inevitably destroyed vegetation thereby reducing food
and cover for prey organisms. The lagomorph and kit fox declines were greatest in the
developed areas of NPR-1. Hence, all steps in the causal pathway were present which
qualitatively supports the disturbance pathway (1a = +).
3.1.6. Causal Pathway—Climate
The primary causal pathway for climate is from reduced precipitation to reduced
vegetation, reduced prey and reduced kit foxes (Figure 7). During the 1981-1986 period
of kit fox decline and the three preceding years, effective precipitation measured in
Bakersfield was above average in five years and below average in three (Figure 14). In
particular, the good and extremely good precipitation in 1982 and 1983 had no apparent
effect on the ongoing kit fox decline (Figure 1). This is contrary to other studies that
found a relationship when they included both NPR sites and a longer time period
(Cypher et al., 2000). That suggests that something was negating the expected
precipitation effects on NPR-1 in the period of concern. This evidence weakens climate
(1b = -).
Kit fox abundance on NPR-1 was not correlated with precipitation in the same
year, the prior year, two years previously, or three years previously. This was true for
both the period of decline (1981-1986) and for the entire study period (1981-1990).
(The time lags account for potential causal lags due to the time required for vegetation
and prey to respond to precipitation.) This evidence weakens climate (1b = -).
In addition, if precipitation was the source of the kit fox decline, one would expect
to see the same pattern of decline on NPR-2. However, kit fox abundance on NPR-2
was stable during 1983-1987 and declined thereafter (Figure 1). (Note that the
apparent fluctuations in Figure 1 are seasonal rather than annual.) This evidence
weakens climate (1b = -).
The relationship of total lagomorph counts (mostly jack rabbits) from road
surveys on NPR-1 in 1980-1984 to precipitation in the same year and the previous year
was analyzed by linear regression. With one exception, correlations were extremely low
for both developed and undeveloped areas. In undeveloped areas, lagomorph
abundance was negatively correlated with precipitation in the previous year, which is
contrary to expectations. Jack rabbit abundance by transect survey was weakly
positively correlated with precipitation in the same year (r2 = 0.30) or in the previous
year (r2 = 0.32) during 1983-1990. This evidence weakens climate (1b = -).
28
-------
AVERAGE GROWING SEASON PRECIPITATION BASED ON 1861-90 DATA
0
60 61 82 63 M 66 88 67 SB 69 70 71 72 73 74 76 78 77 78 78 80 81 82 83 84 86 88 87 88 89 90
GROWING SEASON
GROWING SEASON DEFINED AS PRECIPITATION FOR JAN.-MAR OF CURRENT YEAR PLUS OCT.-DEC. OF PREVIOUS YEAR
FIGURE 14
Average Growing Season Precipitation for Bakersfield, California (Source: U.S. DOE,
1993)
29
-------
Based on visual inspection, vegetation production on NPR-1 declined between
1988 and 1991 (U.S. DOE, 1993). Also, at an undisturbed 32 acre site on NPR-1,
annual production declined from 1596 pounds/acre in 1988, to 644 pounds/acre in 1989
and to 85 pounds/acre in 1990. This corresponds to a period of steady decline in
precipitation (Figure 14). This evidence is consistent with precipitation as a cause of
reduced plant production, but it does not relate to the principal period of kit fox decline
when precipitation was higher (1981-1986). This evidence is ambiguous (1b = 0).
In sum, the evidence for the causal pathway from climate to vegetation,
lagomorph prey, and kit fox is negative (overall 1b = -).
3.1.7. Causal Pathway—Competition
Coyotes are primarily predators of lagomorphs and, to a much lesser extent,
small rodents (Cypher and Spencer, 1998). Their increase between 1979 and 1984
coincided with declines in lagomorphs and kit foxes. However, regular quantitative
monitoring of coyote abundance and analysis of coyote diets did not begin until 1985
and after that time coyote abundance declined. By then the principal decline of foxes
and lagomorphs was complete and kit foxes had switched primarily to kangaroo rats.
Hence Cypher and Spencer's (1998) conclusion that there was little competition for food
may not be relevant to the period of concern. The evidence is consistent with coyote
competition during the period of kit fox decline, but the evidence for the following period
is not (1c = 0).
3.1.8. Manipulation of Exposure
To determine the influence of food availability of kit foxes, a supplemental
feeding study was conducted in 1988 and 1989. Supplemental feeding at individual
occupied dens in 1988 increased survival of pups relative to controls from 10-50% and
increased survival of adults from 30-70% (U.S. DOE, 1993). Results were positive in
1989 as well, but the differences were smaller due to increased survival of unfed foxes.
That may be due to heavier coyote control activities in 1989 (U.S. DOE, 1993). This
evidence supports prey abundance, but is not strong because the studies occurred after
the decline and the manipulation was not of the prey (1 = +).
3.1.9. Symptoms
3.1.9.1. Starvation
Starvation was not reported to be a cause of death in kit fox necropsies. That is
negative evidence for a shortage of prey as a cause of mortality (-).
3.1.9.2. Reproductive
Male-biased sex ratios of pups, as observed on developed NPR-1, are
characteristic of female canids that are in poor condition due to poor nutrition (Zoellick
30
-------
et al., 1987). This symptom supports prey abundance but may occur with other causes
(1 = +). Because this symptom occurred on developed NPR-1, it supports prey
abundance through the disturbance pathway (1a = +). It does not support the climatic
pathway because the climate did not differ between areas of NPR-1 (1 b = -).
3.1.10. Mechanistic Sufficiency
The most likely demographic mechanism of low prey abundance is poor nutrition
and reduced fecundity, but the observed reduction in fecundity was only a minor
contributor to the decline (Floit and Barnthouse, 1991). This evidence weakens the
case for prey abundance (1 = -).
3.2. HABITAT ALTERATION
On NPR-1, habitat alteration has been thought to result from disturbance
associated with oil development (2a - Disturbance) or climatic effects (2b - Climate).
The climatic effects are assumed to be reduced plant biomass and production, resulting
in reduced habitat quality. In contrast, oil development may act through loss of
vegetation, noise, human presence, or other disturbances. Although habitat
preferences in terms of vegetation types are known, no detailed habitat model is
available for kit foxes that would allow quantification of the effects of disturbance on
habitat quality. The area developed, number of wells drilled and volume of oil produced
are used as surrogates for habitat disturbance. Growing season precipitation and plant
production were used as surrogates for habitat alteration due to climate.
3.2.1. Spatial/Temporal Co-occurrence—Disturbance
3.2.1.1. Disturbed Versus Undisturbed
The NPR-1 kit fox decline was most severe in the disturbed areas. By 1990, very
few foxes in the NPR-1 study area occurred in the developed upland areas; the
remaining foxes were found primarily in the flatter undeveloped areas (U.S. DOE,
1993). Hence, the decline spatially co-occurred with cumulative habitat disturbance.
This evidence supports disturbance of habitat (2a = +).
3.2.1.2. Temporal Co-occurrence—Disturbance
During the period of decline (1981-1986), oil development continued with a peak
in 1982-1983 followed by a relatively low level of drilling (Table 3). Given the possibility
of time lags and cumulative effects, temporal co-occurrence is ambiguous (2a = 0).
3.2.2. Spatial/Temporal Co-occurrence—Climate
Precipitation was believed to be similar on both developed and undeveloped
areas of NPR-1 and on NPR-2, so the differences in the rates and timing of kit fox
declines is not accounted for by climatic effects on habitat (2b = -).
31
-------
3.2.3. Temporal Sequence
The period of increased development began in 1974 and drilling appeared to
peak in 1976-1978 (Table 3). The beginning of the kit fox decline is uncertain, but it
began no later than the first monitored interval (1981-1982). Hence, the temporal
sequence is ambiguous (2 = 0).
3.2.4. Stressor-Response Relationships in the Field—Disturbance
The 1981 -1986 period of kit fox decline and the full 1981 -1990 study period were
also periods of decline in well drilling and oil production (except for a blip in 1982-1983,
Table 3). Hence, stressor-response models for active disturbance (i.e., number of wells
completed) and kit fox abundance have the wrong sign for the causal hypothesis.
Another approach is to relate the proportional change in kit fox abundance to the
number of wells completed in the same year or the previous year, but that yielded no
apparent relationships. Hence, the stressor-response relationships weaken the causal
hypothesis (2a = —).
3.2.5. Stressor-Response Relationships in the Field—Climate
Few data quantify changes in habitat quality or quantity that might result from
climate and that could be related to fox abundances. However, plant production may be
a surrogate for climate-mediated habitat quality. At an undisturbed 32 acre site on
NPR-1, annual production declined from 1596 pounds/acre in 1988, to 644 pounds/acre
in 1989 to 85 pounds/acre in 1990 (U.S. DOE, 1993). These data do not correlate well
with kit fox abundance in the same year, but they do correlate perfectly (r2 = 0.999) with
kit fox abundance in the following year (y = 7.4 + 0.032X). Although suggestive, models
based on three points inspire little confidence, and the time series is outside the period
of concern, so the evidence is ambiguous with respect to the decline (2b = 0).
3.2.6. Evidence of Exposure or Biological Mechanism
No evidence.
3.2.7. Causal Pathway—Disturbance
Oil development involves the destruction of vegetation, which diminishes habitat.
Noise and human activity also diminish habitat during the period of construction and
drilling activity. All steps in this causal pathway were present (2a = ++).
3.2.8. Causal Pathway—Climate
The climate was not consistently poor and vegetation data are lacking in the
period of decline. In particular, while kit foxes steadily declined in the period of concern,
precipitation was above average, then below, then above again and below again (Figure
32
-------
14, Section 2.1.6). This lack of a relation between precipitation and kit fox abundance
weakens the case for climate-induced habitat alteration as a cause (2 = —).
Vegetation data were available for a later period. Based on visual inspection,
vegetation production on NPR-1 declined in undisturbed areas between 1988 and 1991
(U.S. DOE, 1993). At an undisturbed 32 acre site on NPR-1, annual production
declined from 1596 pounds/acre in 1988, to 644 pounds/acre in 1989 to 85 pounds/acre
in 1990. This corresponds to a period of steady decline in precipitation (Figure 14).
These data are well correlated with precipitation in the same year (r2 = 0.987) (y =
1011x - 2395). However, models based on three points inspire little confidence, the
time series is outside the period of concern, and the precipitation was lower than in the
period of concern, so the evidence is ambiguous (2 = 0).
The combined score for the climate to habitat pathway is weakly negative (2 = -).
3.2.9. Mechanistic Sufficiency
Because habitat could affect mortality, fecundity and emigration, the
demographic models cannot be used to determine the mechanistic sufficiency of habitat
modification as a cause (2 = 0).
3.3. PREDATORS
3.3.1. Spatial/Temporal Co-occurrence—Developed versus Undeveloped
Coyotes were more abundant on developed than undeveloped NPR-1 in the
period of decline and the decline was greater on developed NPR-1 (U.S. DOE, 1993).
That spatial co-occurrence supports predation (+).
3.3.2. Spatial/Temporal Co-occurrence—Temporal on NPR-1
Coyote numbers were lowest when the first survey was conducted on NPR-1 in
1979 (8 observed on 522 miles of transect), but 5 years later 108 were observed over
those transects (U.S. DOE, 1993). Hence, an increase in coyote numbers occurred
within the same time interval as the observed decline in kit fox abundance, but the
pattern of abundance between those dates in unknown. Hence, the decline co-occurred
with the candidate cause (+).
3.3.3. Spatial/Temporal Co-occurrence—NPR-1 versus NPR-2
Coyote abundance on NPR-2 was not known for the period of concern (i.e., prior
to 1985) (Figure 15). After that, it was irregular and did not correspond to kit fox
abundance patterns except that both dropped in the late 1980s, after the kit fox
decline (0).
33
-------
X
-------
3.3.4. Temporal Sequence
The low abundance of coyotes in 1979 suggests that an increase in coyote
abundance did not precede the decline in kit foxes, but the timing of the coyote increase
and the beginning of the kit fox decline are unclear. This evidence is ambiguous (0).
3.3.5. Stressor-Response in the Field
Between 1979 and 1985, the coyote population on NPR-1 greatly increased and
the kit fox population greatly declined. Then the coyote population declined from 1985
(when coyote control and regular monitoring, using scent stations, began) until 1991
(Figure 15) (U.S. DOE, 1993). For three years following the onset of coyote control
(1986-1989) the kit fox population stopped declining. Then from 1989-1991, both
declined. Because of the switch from transect surveys to scent stations, correlations
with kit fox abundance cannot be calculated for the period of decline or the entire period
of interest. However, the stressor-response relationship is qualitatively correct until
1989. When coyotes increased, kit foxes declined, and when coyotes declined, kit
foxes stopped declining. Hence, the stressor-response relationship could not be
quantified (NE) and the qualitative association is scored as spatial/temporal co-
occurrence, above.
3.3.6. Causal Pathway
Multiple causal pathways that may associate coyote abundance with oil
development were not documented (Figure 7). It is speculated that the absence of
shooting and trapping prior to the control program may have allowed the increase
coyote abundance, but it does not explain the initially low numbers. Coyotes may have
also benefited from increased road kills to be scavenged or from food discarded by
workers (Cypher and Spencer, 1998). Those resources inevitably increased with
increased oil production activities in the late 1970s and would have been associated
with developed areas.
There is some evidence for the causes of the coyote decline. Coyote abundance
declined during the period of the control program beginning in 1985. The decline also
corresponded to the decline in lagomorph prey and, after 1988, to below average
precipitation.
Evidence exists for complete causal pathways to coyote abundance and
predation on kit foxes (+).
3.3.7. Evidence of Exposure or Biological Mechanism
Because coyotes do not consume the foxes that they kill, predation by coyotes
was well documented by necropsy of foxes from NPR-1. Coyote-killed foxes were
identified by characteristic puncture wounds and associated muscle and bone injuries
35
-------
(Cypher and Spencer, 1998). This evidence for the predation mechanism is clear and
consistent (++).
3.3.8. Manipulation of Exposure
A coyote control program was conducted for 6 years on and around NPR-1
beginning in 1985. The decline of the kit fox population ended in the second year
of this period. Overall, 591 coyotes were killed. This evidence supports predation
as the cause, but is ambiguous because there is no reference and other factors may
confound the effects of coyote control (+).
3.3.9. Mechanistic Sufficiency
A demographic model of kit foxes on NPR-1 for the period 1980-1986 found that
the decline was caused by high mortality, particularly of young-of-the-year foxes (Floit
and Barnthouse, 1991). The mortality due to predation alone was more than sufficient
to cause a decline. Although fecundity was depressed in developed areas relative to
undeveloped areas, the population abundance was insensitive to variance in fecundity.
Net emigration from the developed areas did not significantly contribute to the decline.
A less detailed analysis of an equivalent model for the U.S. DOE (1993) that extended
to 1989 gave qualitatively similar results, but different rates because of inclusion of a
period after the decline (1986-1989). In sum, mortality was the mechanism of the
decline and predation was the overwhelming cause of mortality (80%; Table 7). This
line of evidence strongly supports predation as the proximate cause (+++).
3.4. TOXIC CHEMICALS
The data concerning kit fox exposures and data analyses used for this candidate
cause are presented in Suter et al. (1992). That report presents more results in more
detail.
3.4.1. Spatial/Temporal Co-occurrence
Chemicals related to oil production occurred on the developed areas at much
greater concentrations than on undeveloped areas during the period of population
decline. Sources included produced water sumps, oil spills, drilling fluids in sumps or
deposited on land, and spills of chemicals used in oil production (Suter, 1988; U.S.
DOE, 1993). The arsenical anti-corrosion compound W-41 and the hexavalent
chromium added to drilling fluids were particular concerns. Arsenic-contaminated water
was deposited in six unlined sumps and arsenic-contaminated wastes were deposited in
unlined trenches. Hexavalent chromium was spilled on at least 65 sites. The less toxic
trivalent chromium in drilling fluids is widely distributed on the site. This evidence
supports toxicants (+).
36
-------
TABLE 7
Percentage of Radiocollared San Joaquin Kit Foxes Dying from Various Causes on
NPR-1 from 1980-1988
Class
Number
Cause of Death
(as % of deaths of identified cause)
Predation
Vehicle
Other
Sex
Male
91
78.0
16.5
5.5
Female
106
83.0
13.2
3.8
Age
Pup
82
80.5
14.6
4.9
Adult
103
81.6
14.6
3.9
Year
1980-82
45
80.0
20.0
0.0
1983-85
113
79.6
12.4
8.0
1986-88
39
84.6
15.4
0.0
Total
197
80.7
14.7
4.6
Source: U.S. DOE (1993).
37
-------
3.4.2. Temporal Sequence
It is hypothesized that increased development after 1976 increased exposures,
but there are no data to support that assumption. Until 1986, all wastes were deposited
on site, and waste waters and drilling fluids continued to be deposited in sumps and on
land, respectively (U.S. DOE, 1993).
3.4.2.1. Arsenic
The arsenical water treatment chemical W-41 was used on NPR-1 from 1922-
1970. Although arsenic residues persisted at the site, arsenic use did not increase
immediately before the decline (-).
3.4.2.2. Barium
Barium was used throughout the period of concern and the years before.
Increased drilling before the decline inevitably meant increased use of barite and
presumably an accumulation of barite on developed NPR-1 (+).
3.4.2.3. Chromium
Lignochromates and hexavalent chromium salts were used in drilling fluids from
1954-1983. Hence, it is plausible that chromium exposures increased as drilling
increased in the mid 1970s to early 1980s and chromium contamination increased on
developed NPR-1 (+).
Overall, this evidence is ambiguous because there are no data from the late
1970s and no good temporal data (0).
3.4.3. Stressor-Response Relationships in the Field
There were no large or statistically significant correlations of longevity with fur
concentrations of any element among the 21 foxes for which both time of death and fur
concentration data were available. This evidence weakens toxicants (-).
3.4.4. Evidence of Exposure or Biological Mechanism
Chemical exposures were investigated by analyzing the elemental composition of
kit fox fur samples. Elements that were not detected by neutron activation analysis in at
least half of the samples were excluded, leaving 35 elements. Fox pups were excluded
because of relatively low concentrations and the sexes were combined because they
did not differ. Samples came from NPR-1 (49), NPR-2 (12), Camp Roberts (20) and
Carrizo Plain (6). Data analysis focused on typical (median concentration) foxes at
each site and level of development and on foxes with exceptionally high (top decile)
concentrations for each element.
38
-------
Analysis of data for all 35 elements served to indicate the degree of systematic
variance among sites in exposure to metals and metalloids. Statistically significant
differences among sites were found for all but three elements (chlorine, cobalt, and
vanadium (Table 8). However, most elemental concentrations were not highest on oil
fields. Of the 35 elements, Camp Roberts fur had the highest concentrations for 21 and
second highest for 6, Elkhorn Plain fur was highest for 7 and second highest for 17,
developed NPR-1 fur was highest for 1 and second highest for 6, NPR-2 fur was highest
for 4 and second highest for 4 (all NPR-2 foxes were from developed areas), and
undeveloped NPR-1 was not highest or second highest for any element, but was lowest
for 23 and second lowest for 8. In sum, fur from the undeveloped remote reference
sites had the highest concentrations of most detected elements, fur from undeveloped
areas on NPR-1 had low concentrations and sites with extensive oil development had
intermediate levels. Hence, although there was a statistically significant positive
correlation of fur concentration and percent disturbance of the fox's home range on
NPR-1 and NPR-2 combined for 23 elements, it was attributable to the exceptionally low
concentrations for undeveloped NPR-1, not high concentrations in developed locations.
Some elements in fur were associated with oil development and identified as
particular hazards.
3.4.4.1. Arsenic
Median arsenic levels were higher in fur from developed NPR-1 and NPR-2 than
from other sites. Arsenic in fur was significantly positively correlated with percent
disturbance, total wells, and new wells in the fox's home ranges. Arsenic
concentrations were highly variable among individuals (>2600x) but only moderately
variable among site medians (4.3x).
3.4.4.2. Barium
Barite is a major constituent of drilling fluids. The median barium concentration
in fur from developed NPR-1 was higher than from any other site. The highest
individual concentration and seven of the top ten concentrations were from foxes from
developed NPR-1, but the second and fourth highest were from Camp Roberts. Barium
concentrations in fur from both oil fields were significantly positively correlated with
percent disturbance and the number of wells.
3.4.4.3. Chromium
The median chromium concentration in fur from developed NPR-1 was lower
than for any other site. Although the median fur concentration and soil concentrations
were low, the highest fur concentration was from developed NPR-2 and four of the top
ten concentrations were from developed areas. This suggests that some individual
foxes had been exposed to chromium-containing wastes. Chromium concentrations in
fur were significantly positively correlated with the number of wells in the home range
but not the percent disturbance.
39
-------
TABLE 8
Ranges of Metal Concentrations (ppm) in Hair of Individual San Joaquin Kit Foxes Sampled on the Elk Hills (NPR-1),
Adjacent Buena Vista Hills (NPR-2), Carizo Plain and Camp Roberts (Other Sites), California, Compared with
Concentrations in Hair from Other Mammals
Metal
NPR-1 Kit
Foxes
NPR-2 Kit
Foxes
Other Site
Kit Foxes
Wildlife
Teton
Coyotes
(Huckabee,
1972)
High
Exposure
Areas
Human
Normal
Human
Toxic
Aluminum
66.8-1710
110.0-881
68.6-2830
5a
Antimony
0.008-1.4
0.017-0.44
<0.005-0.60
<0.2-12b
0.09-1.8
0.03-24c
Arsenic
0.03-4.7
0.15-5.4
<0.01-2.6
0.3-8.9d,e,f
0.0-2.0c
3C
Bromine
3.6-66
8.4-23
1.9-26
30a
Calcium
<67-1000
<67-400
<67-2800
497g
Cerium
<0.3-2.3
0.4-1.5
<0.3-3.0
<1-20b
1.9-2.6
Chromium
<0.1-3.9
0.7-7.7
<0.1-5.8
<0.3-640b
0.7-5.8
3.9-4.8h,i
0.0-40c
Cobalt
0.15-2.40
0.21-1.15
0.14-1.10
0.1a
Copper
0.015-54
11-23
12-48
6.9-8.3CJ
7.8-120c
Iron
151-1430
282-4150
270-5500
<21-6400b
23-160
26.79
Magnesium
<40-640
<40-360
<40-660
56.7g
-------
TABLE 8 cont.
Metal
NPR-1 Kit
Foxes
NPR-2 Kit
Foxes
Other Site
Kit Foxes
Wildlife
Teton
Coyotes
(Huckabee,
1972)
High
Exposure
Areas
Human
Normal
Human
Toxic
Manganese
0.95-31.70
2.13-27.70
1.74-50.60
0.3a
Mercury
0.21-1.2
0.28-3.9
0.25-10
<0.008-10.7b
<0.008-2.8
9.8-117.5kJ
0.01-30c
50-200c
Nickel
<1-7
<1-10
<1-8
0.18-1. T
0.0-11c
Gold
0.0007-
0.065
0.0015-
0.011
0.0013-
0.135
<0.04-0.6b
0.002-0.04
Potassium
<28-360
41-250
<28-1300
67.69
Rubidium
<0.3-2.9
<0.3-1.5
<0.3-3.7
5.8-8.3bm
Scandium
0.04-0.46
0.06-0.21
0.05-0.54
<0.05-2b
0.005-0.009
Selenium
0.60-1.8
0.90-3.0
0.50-4.2
0.71-27n
0.08-17b
0.8-7.83
3.8-12°'p
0.89-13q
0.97-18r
0.3-13C
8-30c
Silver
<0.1-0.2
<0.1
<0.1-0.3
<0.4-110b
0.06-12
Sodium
4.1-212.0
6.6-98.0
14.0-208.0
309g
Titanium
<14-120.0
<14-58.0
<14-114.0
4a
-------
TABLE 8 cont.
Metal
NPR-1 Kit
Foxes
NPR-2 Kit
Foxes
Other Site
Kit Foxes
Wildlife
Teton
Coyotes
(Huckabee,
1972)
High
Exposure
Areas
Human
Normal
Human
Toxic
Vanadium
0.3-11.5
0.6-3.2
<0.1-4.4
0.006-2.7C
Zinc
93-220
118-178
87-180
13-6300b
91-620
65-2003
aLenihan (1978).
bHuckabee etal. (1972).
cJenkins (1979).
dLewis (1972).
^ 8Orheim et al. (1974).
ho livestock grazing near smelters; reference animals had 0-0.46 ppm.
sBarker et al. (1976).
hTaylor et al. (1975).
'Cotton rats from near cooling towers using chromate corrosion inhibitors; reference rats had 0.39 ppm.
livestock grazing near smelters; reference animals had 6.8-7.8 ppm.
kDoi (1973).
'Cats from the vicinity of Minamata, Japan.
mRodents from areas of heavily mineralized soils in Idaho.
"Kit foxes from Bakersfield.
°Schroeder et al. (1970).
pRats fed nominally toxic levels of Se; control rats had 0.6 ppm.
qKit foxes from the Kesterson Reservoir.
rCoyotes from the Kesterson Reservoir.
sPetering et al. (1971).
Source: Suteretal. (1992).
-------
3.4.4.4. Sodium
Sodium was used as a marker for produced water which is primarily a sodium
chloride solution. However, neither median nor extreme fur concentrations of sodium
were high for developed NPR-1 relative to other sites.
3.4.4.5. Vanadium
Vanadium occurs in relatively high concentrations in petroleum and is used as a
marker for petroleum in the environment. The median fur concentration was highest for
the Carizo Plain, but six of the top ten individuals were from developed NPR-1 and the
other four of the top ten were from undeveloped NPR-1 even though undeveloped
NPR-1 had the lowest median concentration. Vanadium concentrations were
significantly positively correlated with percent development on both NPR-1 and NPR-2.
This suggests that some foxes were exposed to petroleum.
To summarize, the median concentrations of arsenic and barium were higher on
developed NPR-1 than on other sites and some foxes appeared to be relatively highly
exposed. However, there was considerable overlap of the distribution of concentrations
with the other sites. Median chromium and vanadium concentrations from NPR-1 were
not higher than other sites, but some foxes were relatively highly exposed. This
evidence is taken as positive in that it showed that some foxes were exposed to
petroleum or metals in the area where the decline occurred (++).
3.4.5. Causal Pathway
Individual pathways of exposure and lines of evidence are scored separately.
3.4.5.1. Soil Concentrations
Soil may be a pathway of exposure through direct ingestion or through the food
web. Direct ingestion includes grooming and soil ingested incidentally with prey.
However, there were no large or statistically significant correlations between elemental
concentrations in random soil samples and percent disturbance in the quarter section
from which the sample was taken. Similarly, the differences in fur concentrations
among sites were not attributable to those soil concentrations. There were no large or
statistically significant positive correlations of soil and fur concentrations at NPR-1 or
Camp Roberts. Differences in soil concentrations among sites were small relative to
differences in fur concentrations. Hence neither soil contamination nor natural soil
concentrations account for differences in exposure among foxes. However, this
conclusion addresses only soil contamination that is sufficiently wide-spread to be
detected by random soil sampling (-).
43
-------
3.4.5.2. Soil Intake
Differences in exposure to metals in soils may be due to differences in rates of
intake rather than differences in concentration. Differences in disturbance between
developed and undeveloped NPR-1 may result in increased exposure to soil due to
dust, but cannot account for differences among other sites. Differences in prey
composition may explain the differences among sites (Suter et al., 1992), but that
explanation does not account for the decline of foxes on NPR-1 (-).
3.4.5.3. Local Soil Contamination (Wastes)
Local spills and deposits of contaminants were abundant on developed NPR-1.
Hence, the evidence for soil as a pathway is positive on the basis of local soil
contamination (+).
3.4.5.4. Waste Water
Produced waters in open sumps may have been a route of exposure to toxicants
due to drinking. Kit foxes are desert animals that do not require drinking water and do
not normally drink, but they may consume water from produced water sumps. There is
no evidence for waste waters as a route of exposure (0).
3.4.5.5. Petroleum
Foxes were potentially exposed to petroleum in spills and oil recovery sumps.
One kit fox died in spilled oil during the period of study. The evidence for contact with
oil as an exposure route is weakly positive (+).
3.4.6. Mechanistic Sufficiency
The demographic model indicated that the decline was caused by mortality,
primarily due to predation. There was no evidence that toxicity caused mortality of
foxes so it was not the proximate cause (-).
3.5. VEHICULAR ACTIVITIES
3.5.1. Spatial/Temporal Co-occurrence
The increase in development inevitably increased vehicle traffic, and it seems
likely that traffic was greatest in developed areas. Most vehicle deaths involved young-
of-the-year foxes and occurred in developed areas (Table 7). This evidence supports
vehicular activity as a cause (+).
Other types of accidents were minor. Among all known kit fox mortalities on
NPR-1 (1980-1990), 1 was buried during construction, 1 was trapped in a pipe, and
44
-------
4 died in live traps during the demographic studies (U.S. DOE, 1993). Hence, other
accidents are not considered.
3.5.2. Temporal Sequence
No data are available to determine the sequential relationship between vehicular
activity and kit fox mortality (NE).
3.5.3. Evidence of Exposure or Biological Mechanism
Fifteen percent of identified mortalities of radio-collared kit fox on NPR-1 during
1980-1988 were due to vehicle collisions based on location and necropsy results
(U.S. DOE, 1993). This evidence supports vehicular activity as a cause (++).
3.5.4. Causal Pathway
Vehicular activity was not quantified, but it inevitably increased on the site due to
increased oil development activities. The 15% of total mortality on NPR-1 due to
vehicle strikes was higher than in most other studies where vehicular strikes rarely
exceed 10% of mortalities (Bjurlin and Cypher, 2003). This suggests that oil
development was responsible for elevated kit fox mortality and supports the causal
pathway (+).
3.5.5. Mechanistic Sufficiency
A demographic model of kit foxes on NPR-1, between 1981 and 1986, found that
the decline was caused by high mortality, particularly of young-of-the-year foxes (Floit
and Barnthouse, 1991). A less detailed analysis of an equivalent model, but for the
period 1981-1989, gave qualitatively similar results (U.S. DOE, 1993). Hence, mortality
was the cause of the decline and vehicular strikes were responsible for approximately
15% of identified mortality (Table 7). This line of evidence supports accidents as a
contributing proximate cause (+).
3.6. DISEASE
3.6.1. Spatial Co-occurrence
Necropsies provided little evidence of possible disease-induced mortality on
either NPR-1 or NPR-2 between 1980 and 1995 (Cypher et al., 2000). However, it is
possible that an increased frequency of nonlethal disease may have weakened foxes
and thereby caused increased predation on developed areas. The absence of evidence
of co-occurrence weakens disease as a cause (-).
3.6.2. Temporal Sequence
There is no evidence concerning changes in disease rates (NE).
45
-------
3.6.3. Causal Pathway
The elements of the hypothesized causal pathway (humans with pets and
coyotes) were present, but transport of pathogens onto NPR-1 was not documented so
the pathways remain hypothetical (0).
3.6.4. Evidence of Exposure or Biological Mechanism
A serological survey for pathogens was conducted in 1981-1982 and 1984
(McCue and O'Farrell, 1986, 1988), and serum chemistry was analyzed (McCue and
O'Farrell, 1992). Canine parvovirus antibodies were found in nearly all foxes,
regardless of development. Antibodies for other pathogens were rare and data were
insufficient to make comparisons between levels of development. The investigators
presumed that if foxes were highly exposed to pathogens it would be reflected in
changes in hematological parameters. Sufficient data on hematology were gathered in
1981-1982 to make comparisons between levels of development, but no differences in
either mean or extreme values were found (McCue and O'Farrell, 1987). This evidence
greatly weakens disease as a cause (—).
46
-------
4. STEP 4. EVALUATE DATA FROM ELSEWHERE
All candidate causes are mechanistically plausible (+). That type of evidence is
not discussed, because it would not influence the relative strength of evidence for the
candidate causes.
4.1. PREY ABUNDANCE
4.1.1. Stressor-Response from Other Field Studies
Numerous studies have demonstrated a positive correlation between the
abundances of mammalian predators and their prey. In particular, Egoscue (1975)
showed that the abundance of kit foxes (V. m. nevadensis) in Utah followed the
abundances of black-tailed jack rabbits. That population also showed an elevated
male:female ratio of pups. This relationship agrees qualitatively with the relationship at
the site (+).
4.2. HABITAT ALTERATION
No evidence from elsewhere (NE).
4.3. PREDATORS
4.3.1. Stressor-Response from Other Field Studies
Coyotes were the cause of 65% of kit fox mortalities on the nearby Carrizo Plain
(Ralls and White, 1995). Coyotes may also be a significant cause of mortality in
populations of swift foxes (Scott-Brown et al., 1987) and gray foxes (Cypher, 1993).
Field studies have documented decreases in red fox abundance in apparent response
to increased coyote abundance (Harrison et al., 1989; Major and Sherburne, 1987;
Sargent et al., 1987). This evidence qualitatively supports predation (+).
4.4. TOXIC CHEMICALS
4.4.1. Stressor-Response from Other Field Studies
Livestock have died from drinking produced waters at other oil fields, primarily
due to osmotic burden (McCoy and Edwards, 1980). Sump waters on NPR-1 were
highly saline; samples contained 1720-14,400 mg/L of sodium and four other metals
were found at >1000 mg/L, which is consistent with the McCoy and Edwards (1980)
study. However, kit foxes do not require drinking water, and, as desert animals, they
may not be as sensitive as livestock to osmotic stress. This evidence is ambiguous (0).
47
-------
4.4.2. Stressor-Response from Laboratory Studies
Four metals (cadmium, copper, molybdenum, and strontium) were found in
produced waters from open sumps at concentrations above drinking water criteria, so
they are potentially toxic in chronic exposures. However, there is no evidence of
exposure. This evidence is ambiguous (0).
Although soils concentrations were available for the site, it was not possible to
estimate exposures to these materials for comparison to toxic doses. Soil consumption
is inevitable, but unquantifiable. This evidence is ambiguous (0).
4.4.3. Stressor-Response from Other Studies—Fur
Elemental concentrations in fur that are related to toxic effects are rare.
Concentrations in the fur of wildlife from undeveloped areas were taken to be no-effect
levels and concentrations in coyotes from the Grand Tetons National Park, Wyoming,
were considered particularly relevant. These no-effect concentrations were available for
12 elements, and none of them were exceeded by NPR-1 kit foxes. Concentrations in
fur from various contaminated sites were considered to represent potentially toxic
levels. Finally, concentrations associated with toxic effects were available for a few
elements. Comparisons are presented here for the three elements of concern for which
effects or no-effects data were found (Table 8).
4.4.3.1. Arsenic
One fox associated with NPR-1 had 26 ppm arsenic (As) in its fur. This is much
higher than concentrations in the hair of humans who died of As poisoning (3 ppm;
Table 8). However, that fox lived north of NPR-1 along the California aqueduct and may
have been exposed to residues of arsenical agrochemicals. That fox was alive and
apparently healthy at the time that fur was collected and lived for more than a year after
capture. One fox from developed NPR-1 and one from developed NPR-2 also
exceeded the 3 ppm level. This suggests that human hair concentrations are not good
indicators of toxic exposures to arsenic in kit foxes. No data were found for other
wildlife.
4.4.3.2. Chromium
Chromium concentrations in NPR-1 kit foxes were within the range of
concentrations in Teton coyotes and other wildlife from uncontaminated areas, so they
are assumed to be nontoxic.
4.4.3.3. Selenium
Selenium concentrations in NPR-1 kit foxes were low relative to concentrations in
rats fed toxic doses of Se, relative to humans experiencing Se toxicity, and relative to kit
foxes and coyotes at Kesterson reservoir where birds experienced severe Se toxicity.
They were also lower than concentrations in Teton coyotes and other wildlife.
48
-------
This evidence weakens toxic chemicals as a cause, because there was no
indication that the observed concentrations were related to toxicity (-).
4.5. VEHICULAR ACTIVITIES
4.5.1. Stressor-Response from Other Field Studies
The mortality rate from this cause during the period of concern (~15%) is
somewhat higher than of kit fox mortalities from vehicles observed in other studies
(~10%), as summarized by Cypher et al. (2000). This evidence strengthens vehicles as
a cause of the decline (+).
4.6. INCREASED DISEASE
4.6.1. Stressor-Response from Other Field Studies
Mortality due to disease is hard to detect, and Cypher et al. (2000) found no
documentation of epizootics in kit foxes. However, field studies have documented
decreases in the abundance of other fox species associated with diseases (Nicholson
and Hill, 1984). This evidence qualitatively supports disease (+).
49
-------
5. STEP 5. IDENTIFY THE PROBABLE CAUSE
5.1. PROXIMATE CAUSES
Having analyzed the evidence for each candidate cause in the prior two sections,
the next step is to summarize the results, determine the consistency of the evidence for
each cause and determine whether there is a reasonable explanation for any
inconsistencies. The individual types of evidence and the analysis across types of
evidence for each cause are summarized in Tables 9-14. The consistency of the
evidence is evaluated across types of evidence for each candidate cause. That is, is
the evidence all positive, all negative or mixed? All candidate causes except predation
had inconsistent data. The second criterion is the existence of an explanation for the
inconsistencies. Explanations were developed for the inconsistencies in three
candidate causes (habitat modification, prey abundance and vehicular activity) that
involved converting them from candidate causes to contributors to the most likely cause.
After the evidence for each candidate cause is summarized, the evidence is
compared across candidate causes to determine which one is best supported
(Table 15). First, the candidate causes that can be eliminated are identified, then the
most probable cause from among those that remain is identified, and finally the other
candidate causes are discussed.
Although the evidence was inconsistent, disease (Candidate Cause 6) can be
clearly eliminated, because the evidence from the site was negative. Very few of the
trapped foxes were observed to be diseased, little evidence of disease was found
during necropsies, and neither serological nor hematological analyses showed evidence
of an epizootic that would account for the decline. Disease has caused population
declines in other places, but this supporting evidence would be relevant only if there is
some positive evidence from the site.
In contrast, evidence for predation (Candidate Cause 3) as the principal
proximate cause is consistent and strong. Predation by coyotes is the major cause of
death in kit foxes, and a demographic analysis showed that the decline was due to high
mortality, with little influence from low fecundity or high emigration (Floit and
Barnthouse, 1991).
Evidence for vehicular accidents (Candidate Cause 5) is also positive, but the
mortality rate due to accidents is much lower than for predation and not sufficient to
account for the decline. Hence, it was a contributing cause.
The evidence for environmental contaminants (Candidate Cause 4) was
inconsistent and complex. Contaminants from oil development were present and
potential routes of exposure were identified, but only two chemicals, arsenic and barium
were elevated in fur from most foxes from developed NPR-1 relative to reference sites.
Arsenic levels in three foxes reached levels that indicate acute toxicity in humans, but
50
-------
TABLE 9
Evidence for Prey Abundance (Candidate Cause 1) Caused by Disturbance 1a,
Climate 1b, or Competition (1c)
Type of Evidence
Finding
Score
Types of Evidence that Use Data from the Case
Spatial/Temporal
Co-occurrence:
Developed vs.
Undeveloped
1. Prey declined where foxes declined.
++
1a. Prey decline was greatest in developed NPR-1.
+
1b. The difference between developed and
undeveloped areas is unlikely to be climatic.
-
Spatial/Temporal
Co-occurrence:
NPR-1 vs. NPR-2
1. Declines in foxes at both sites correspond to
declines in prey.
++
1a. Prey decline on NPR-1 in the early 1980s
coincided with development.
+
1b. The difference in timing is unlikely to be climatic.
-
Temporal Sequence
The beginning of the decline is undocumented and
the data do not permit analysis of a time lag.
0
Stressor-Response
Relationship in the
Field
1. Kit fox abundance is highly linearly correlated with
lagomorph abundance on NPR-1.
++
1a. The correlation occurred in developed and
undeveloped areas.
-
1b. The correlation may be related to climate.
+
Causal Pathway:
Prey utilization
1a. All steps in this causal pathway are present, but
the relationships are qualitative.
+
1 b. Multiple lines of evidence for the climate pathway
are negative or ambiguous.
-
1c. Coyote abundance increased during the
NPR-1 kit fox decline, but the evidence for
competition is weak to ambiguous.
0
51
-------
TABLE 9 cont.
Type of Evidence
Finding
Score
Evidence of
Exposure or
Biological
Mechanism
1. The frequency of lagomorph remains in feces
declined during the kit fox decline.
++
1a. The decline occurred in developed and
undeveloped areas.
-
1b. The decline could have been related to climate.
+
Manipulation of
Exposure
Feeding studies weakly support food limitations as a
cause of kit fox mortality.
+
Laboratory Tests of
Site Media
None.
NE
Verified Predictions
None.
NE
Symptoms,
Starvation
Necropsies did not report starvation in kit foxes.
-
Symptoms,
Reproductive
A male-biased sex ratio of pups is characteristic of
malnourished females, but other stressors may also
cause it.
+
1a. The bias occurred in developed areas only.
+
1 b. The decline is unlikely to have been related to
climate.
-
Mechanistic
sufficiency
Reduced fecundity, the likely first effect of reduced
food, was not sufficient to cause or significantly
contribute to the decline.
—
Types of Evidence that Use Data from Elsewhere
Mechanistically
Plausible
Low prey abundance, by either the disturbance or
climatic pathway, is mechanistically plausible.
+
Stressor-Response
from Laboratory
Studies
None.
NE
Stressor-Response
from Other Field
Studies
A study of kit foxes in Nevada showed that their
abundance tracked jack rabbit abundance.
+
52
-------
TABLE 9 cont.
Type of Evidence
Finding
Score
Manipulation of
Exposure at Other
Sites
None.
NE
Analogous Stressors
None.
NE
Evaluating Multiple Types of Evidence as a Form of Evidence
Consistency of
Evidence
1a. The evidence is mixed.
-
1b. The evidence is mixed.
-
Reasonable
Explanation of the
Evidence
The loss of lagomorph prey may have resulted in
increased hunting effort and reduced the nutritional
status of females which reduced fecundity but not
mortality due to starvation. Hence, it may have been
detrimental but not the cause of the decline.
C
C = the explanation makes the candidate cause a contributing factor for another cause.
53
-------
TABLE 10
Evidence for Habitat Degradation (Candidate Cause 2) Caused by Disturbance 1a or
Climate 1b
Type of Evidence
Finding
Score
Types of Evidence that Use Data from the Case
Spatial/Temporal
Co-occurrence
2a. Spatial—Disturbance and the kit fox decline
were greatest in developed NPR-1.
+
2a. Temporal—The temporal pattern of production
does not match fox numbers, but time lags are
unclear.
0
2b. Climatic differences are unlikely to account for
differences in kit fox declines between disturbed and
undisturbed or between NPR-1 and NPR-2.
Temporal Sequence
Data from the beginning of the production surge is
lacking and the relationship at the end is ambiguous.
0
Stressor-Response
Relationship in the
Field
2a. Models relating disturbance to kit fox abundance
had the wrong sign or no relationship.
—
2b. Kit fox abundance is related to plant production,
but the data are for a different time period.
0
Causal Pathway
2a. All steps in the causal pathway are present, but
unquantified.
++
2b. Kit fox numbers were not related to precipitation
during the period of decline. Evidence for the full
causal pathway is limited to a drought after the
period of concern.
Evidence of
Exposure or
Biological
Mechanism
No evidence.
NE
Manipulation of
Exposure
No evidence.
NE
Laboratory Tests of
Site Media
No evidence.
NE
Verified Predictions
No evidence.
NE
54
-------
TABLE 10 cont.
Type of Evidence
Finding
Score
Symptoms
No evidence.
NE
Mechanistic
sufficiency
Because habitat could affect mortality, fecundity and
emigration, the demographic models cannot be used
to determine the mechanistic sufficiency of habitat
modification.
0
Types of Evidence that Use Data from Elsewhere
Mechanistically
Plausible
The mechanism is plausible.
+
Stressor-Response
from Laboratory
Studies
No evidence.
NE
Stressor-Response
from Other Field
Studies
No evidence.
NE
Manipulation of
Exposure at Other
Sites
No evidence.
NE
Analogous Stressors
No evidence.
NE
Evaluating Multiple Types of Evidence as a Form of Evidence
Consistency of
Evidence
1a. The evidence is inconsistent.
-
1 b. The evidence is negative or ambiguous.
—
Reasonable
Explanation of the
Evidence
The difference in habitat between developed and
undeveloped NPR-1 may have an indirect effect
through prey or predator abundance.
C
C = the explanation makes the candidate cause a contributing factor for another cause.
55
-------
TABLE 11
Evidence for Predators (Candidate Cause 3)
Type of Evidence
Finding
Score
Types of Evidence that Use Data from the Case
Spatial/Temporal
Co-occurrence:
Developed vs.
Undeveloped
Coyote abundance was greater on developed NPR-1
where the decline was greatest.
+
Spatial/Temporal
Co-occurrence:
NPR-1 vs. NPR-2
Coyote abundance on NPR-2 was irregular and did
not correspond to kit fox abundance patterns except
that both dropped in the late 1980s, after the period
of concern.
0
Spatial/Temporal
Co-occurrence:
Temporal co-
occurrence on
NPR-1
Coyote abundance increased between 1979 and
1984, the period of kit fox decline, but the pattern of
abundance between those dates in unknown. From
1985 to 1991 coyotes declined and kit foxes were
stable.
+
Temporal Sequence
The low abundance of coyotes in 1979 suggests that
an increase in coyote abundance did not precede the
decline in kit foxes, but the timing of the coyote
increase and the beginning of the kit fox decline are
unclear.
NE
Stressor-Response
Relationship in the
Field
Relationships could not be developed until after the
period of decline.
NE
Causal Pathway
The elements of a causal pathway from oil
development to coyote predation on kit foxes were
observed.
+
Evidence of
Exposure or
Biological
Mechanism
Necropsies demonstrated that most mortalities were
caused by coyotes.
++
Manipulation of
Exposure
Killing coyotes was associated with decreased kit fox
mortality.
+
56
-------
TABLE 11 cont.
Type of Evidence
Finding
Score
Laboratory Tests of
Site Media
No evidence.
NE
Verified Predictions
No evidence.
NE
Symptoms
No evidence.
NE
Mechanistic
sufficiency
The decline was due to mortality and 80% of
mortality was due to predation.
+++
Types of Evidence that Use Data from Elsewhere
Mechanistically
Plausible
This mechanism is plausible.
+
Stressor-Response
from Laboratory
Studies
No evidence.
NE
Stressor-Response
from Other Field
Studies
Other studies of coyote predation on foxes show
high rates and reduced fox abundances.
+
Manipulation of
Exposure at Other
Sites
No evidence.
NE
Analogous Stressors
No evidence.
NE
Evaluating Multiple Types of Evidence as a Form of Evidence
Consistency of
Evidence
All evidence is positive.
+++
Reasonable
Explanation of the
Evidence
None needed.
NA
57
-------
TABLE 12
Evidence for Toxic Chemicals (Candidate Cause 4)
Type of Evidence
Finding
Score
Types of Evidence that Use Data from the Case
Spatial/Temporal
Co-occurrence
Toxic chemicals were found on the site during the
period of kit fox decline.
+
Temporal Sequence
Increased drilling activity may have increased
exposure to toxicants but the temporal sequence is
uncertain.
0
Stressor-Response
Relationship in the
Field
Kit fox longevity was not negatively correlated with
contaminant concentrations in fur.
—
Causal Pathway
Soil—routes of exposure to soil exist, but
contaminant concentrations were not elevated in
random soil samples from developed NPR-1.
—
Wastes—spills and deposits of waste were present
and available for direct or indirect exposure.
+
Water—waste waters were highly contaminated, but
there was no evidence of drinking by kit foxes.
0
Petroleum—spills and sumps were available to foxes
and at least one died.
+
Evidence of
Exposure or
Biological
Mechanism
Some foxes showed elevated exposure to
contaminants in the developed area of NPR-1 or on
NPR-1 as a whole.
++
Manipulation of
Exposure
None.
NE
Laboratory Tests of
Site Media
None.
NE
Verified Predictions
None.
NE
58
-------
TABLE 12 cont.
Type of Evidence
Finding
Score
Symptoms
None.
NE
Mechanistic
sufficiency
The toxic effects, if any, could not account for the
elevated mortality rate that induced the decline.
-
Types of Evidence that Use Data from Elsewhere
Mechanistically
Plausible
Both lethalities and sublethal debilitation due to
chemicals occurring on the site are mechanistically
plausible.
+
Stressor-Response
from Laboratory
Studies
Except for arsenic in three foxes, concentrations in
fur were not at known toxic levels.
0
Stressor-Response
from Other Field
Studies—Water
Livestock have died from drinking produced waters,
but kit foxes do not require drinking water.
0
Stressor-Response
from Other Field
Studies—Fur
There is no evidence that observed concentrations
were toxic.
—
Manipulation of
Exposure at Other
Sites
No evidence.
NE
Analogous Stressors
No evidence.
NE
Evaluating Multiple Types of Evidence as a Form of Evidence
Consistency of
Evidence
The evidence was inconsistent.
-
Reasonable
Explanation of the
Evidence
Although contaminants were available, few foxes
were exposed.
—
59
-------
TABLE 13
Evidence for Vehicular Accidents (Candidate Cause 5)
Type of Evidence
Finding
Score
Types of Evidence that Use Data from the Case
Spatial/Temporal
Co-occurrence
Most vehicle deaths occurred in developed areas.
+
Temporal Sequence
No evidence.
NE
Stressor-Response
Relationship in the
Field
No evidence.
NE
Causal Pathway
Vehicular activity increased on the site due to
increased oil development activities.
+
Evidence of
Exposure or
Biological
Mechanism
Necropsy of kit foxes established that vehicle
collisions were the cause of death.
++
Manipulation of
Exposure
No evidence.
NE
Laboratory Tests of
Site Media
No evidence.
NE
Verified Predictions
No evidence.
NE
Symptoms
No evidence.
NE
Mechanistic
sufficiency
This source of mortality is not sufficient to cause the
decline, but it does contribute to the demographic
mode of action.
+
Types of Evidence that Use Data from Elsewhere
Mechanistically
Plausible
This mechanism is plausible.
+
60
-------
TABLE 13 cont.
Type of Evidence
Finding
Score
Stressor-Response
from Laboratory
Studies
No evidence.
NE
Stressor-Response
from Other Field
Studies
The proportion of mortalities due to vehicles is a little
higher than most other fox populations.
+
Manipulation of
Exposure at Other
Sites
No evidence.
NE
Analogous Stressors
No evidence.
NE
Evaluating Multiple Types of Evidence as a Form of Evidence
Consistency of
Evidence
The evidence was inconsistent.
-
Reasonable
Explanation of the
Evidence
The evidence is consistent with vehicular accidents
as a contributory cause, but it would not have been
sufficient alone.
C
C = the explanation makes the candidate cause a contributing factor for another cause.
61
-------
TABLE 14
Evidence for Disease (Candidate Cause 6)
Type of Evidence
Finding
Score
Types of Evidence that Use Data from the Case
Spatial/Temporal
Co-occurrence
Necropsies found few disease-induced mortalities.
-
Temporal Sequence
No evidence.
NE
Stressor-Response
Relationship in the
Field
No evidence.
NE
Causal Pathway
Elements of the hypothesized causal pathways were
present but no evidence supported their operation.
0
Evidence of
Exposure or
Biological
Mechanism
No differences were found in serological or
hematological parameters that would support
disease as a cause.
Manipulation of
Exposure
No evidence.
NE
Laboratory Tests of
Site Media
No evidence.
NE
Verified Predictions
No evidence.
NE
Symptoms
Symptoms of nonlethal disease were not recorded.
NE
Mechanistic
sufficiency
Disease was not sufficient to cause or significantly
contribute to the mortality that induced the decline.
—
Types of Evidence that Use Data from Elsewhere
Mechanistically
Plausible
The mechanism is plausible.
+
62
-------
TABLE 14 cont.
Type of Evidence
Finding
Score
Stressor-Response
from Laboratory
Studies
No evidence.
NE
Stressor-Response
from Other Field
Studies
Epizootics causing decreased abundance have been
observed in other fox species.
+
Manipulation of
Exposure at Other
Sites
No evidence.
NE
Analogous
Stressors
No evidence.
NE
Evaluating Multiple Types of Evidence as a Form of Evidence
Consistency of
Evidence
The evidence was inconsistent.
-
Reasonable
Explanation of the
Evidence
The evidence from elsewhere was weakly positive,
but the evidence from the site, which was
consistently negative, was much more relevant.
—
63
-------
TABLE 15
Comparison of the Strength of Evidence for the Candidate Causes. Types of evidence with no evidence for any candidate cause
were excluded.
Types of Evidence
Prey
Habitat
Predation
Toxics
Accidents
Disease
Disturbance
Climate
Disturbance
Climate
Evidence that Uses Data from the Case
Spatial/Temporal Co-occurrence
++
+
—
+
+
+
—
+
-
Temporal Sequence
0
0
0
0
NE
NE
Evidence of Exposure or Biological
Mechanism (Pathway independent)
++
NE
NE
+ +
+ +
+ +
Evidence of Exposure or Biological
Mechanism (By pathway)
-
+
Causal Pathway3
+
-
+ +
-
+
+
+
0
Stressor-Response Relationships
from the Field (pathway independent)
+ + +
0
NE
NE
NE
NE
Stressor-Response Relationships
from the Field (by pathway)
-
+
Manipulation of Exposure
+
NE
NE
+
NE
NE
NE
Symptoms, Starvation15
-
NE
NE
NE
NE
NE
NE
Symptoms, Reproductive15 (pathway
independent)
+
Symptoms, Reproductive15 (by
pathway)
+
-
-------
TABLE 15 cont.
Types of Evidence
Prey
Habitat
Predation
Toxics
Accidents
Disease
Disturbance
Climate
Disturbance
Climate
Mechanistic Sufficiency
-
0
+++
-
+
—
Evidence that Uses Data from Elsewhere
Mechanistically Plausible Cause
+
+
+
+
+
+
+
+
Stressor-Response Relationships
from Other Field Studies
+
NE
NE
0
0
+
+
Stressor-Response Relationships
from Laboratory Studies
NE
NE
NE
NE
-
0
NE
NE
Evaluating Multiple Lines of Evidence
Consistency of Evidence
-
-
-
+++
-
-
-
Explanation of the Evidence
Cc
Cc
NA
-
Cc
-
aAn additional causal pathway for prey abundance, competition for prey by coyotes, was ambiguous.
bThe categories of symptoms apply only to prey abundance.
cThe explanation of the evidence makes the candidate cause a contributor to another cause.
-------
those foxes appeared healthy when captured and their longevity was not apparently
reduced. Barium is much less toxic and, although fur levels on NPR-1 were high, they
significantly overlapped with fur from reference sites. One fox died after becoming
coated in oil. In sum, there was no evidence that toxic exposures could account for the
high mortality rates that caused the decline.
The availability and utilization of lagomorph prey (Candidate Cause 1) were
strongly related to kit fox abundance, but clinical symptoms of poor condition or
starvation were not observed in trapped animals or during necropsies. Prey availability
can affect fecundity and females on developed areas produced fewer pups, but the
demographic analysis indicated that variance in kit fox fecundity did not significantly
contribute to variance in kit fox abundance. Hence, prey availability does not appear to
be a significant proximate cause. However, it may be a contributing factor in other
sources of mortality. That is, fewer large prey and greater use of small prey implies
more time spent hunting and greater exposure to coyotes and vehicles.
The evidence for habitat alteration (Candidate Cause 2) is ambiguous. The area
devegetated is known, but the quality of habitat provided by the vegetated and
devegetated areas and the affects of human activities on habitat utility for kit foxes are
unknown. The fact that emigration from the developed areas exceeded emigration from
the undeveloped areas suggests that habitat quality was lower in developed areas.
66
-------
6. SOURCES
Although causal analysis must begin by identifying the proximate cause, the
source of that cause must be identified in order to plan management actions. Hence,
we must ask why coyote abundance and associated mortality increased in the early
1980s.
Climate is a potential source of habitat alteration and reduced abundance of
lagomorph prey. This region is semi-arid and a few drier than average years can
reduce the fecundity and survival of lagomorph prey. However, the period of concern
was not especially or consistently dry. The year with the second highest precipitation in
the 30 year record occurred during the decline (Figure 14). In addition, climate would
be the same for developed and undeveloped areas and for both NPR-1 and NPR-2.
Hence, just as climate can be eliminated as the cause of kit fox decline via the habitat or
prey causal pathways (1b and 2b), it cannot be the cause of increased coyote numbers
or predation in the early 1980s. The later dry period of 1988-1990 shows that low
precipitation can produce a clear signal: production of plants and abundance of
lagomorphs, coyotes and kit foxes all declined. Therefore, climate can be eliminated as
the source of the decline.
Disturbance due to oil development and production is a source of habitat
alteration and reduced prey abundance. Evidence for the effects of disturbance comes
primarily from comparisons of developed and undeveloped areas of NPR-1. The
decline in both kit foxes and lagomorphs was greater on developed than undeveloped
NPR-1. Although the mechanism is unclear, it seems likely that some aspect of active
oil development contributed to the decline. However, it is possible that the differences
in the demographics of kit foxes and lagomorphs between developed and undeveloped
areas were due to natural differences.
Counterintuitively, disturbance may also be a source of increased coyote
abundance. Coyotes were more abundant on developed than undeveloped NPR-1
during the decline. Prior to the coyote control program, site development may have
improved coyote habitat by keeping hunters off the site and by providing sources of
fresh water, discarded food and road kills to be scavenged. Cypher and Spencer
(1998) suggested that the availability of anthropogenic food resources may have
increased coyote abundance and predation of kit foxes on the Elk Hills.
Diseases in coyotes may also be sources of changes in coyote abundance. The
low observed abundance of coyotes in 1979 may have been due to disease. Between
1972 and 1983 the prevalence of antibodies against canine parvovirus in wild coyotes
captured in three western states coincided with the epizootic of the disease in domestic
dogs (Thomas et al., 1984). It is a significant potential pathogen for wild canids, and it
was believed to be linked with declines in coyote numbers (Cypher et al., 2000). There
is no known evidence of a parvovirus epizootic in coyotes in the San Joaquin Valley, but
kit foxes tested on NPR-1 carried parvovirus antibodies. It is possible that the increase
67
-------
in coyotes was a rebound from parvovirus and that kit foxes are resistant. However,
that hypothesis suggests that the high abundance of coyotes in 1985 reflected the peak
of a population that oscillates over long time periods. That would suggest in turn that kit
foxes are normally rare or absent in the developed areas of NPR-1.
The final conceptual model for the cause of the kit fox decline is presented in
Figure 16. The proximate cause is predation. The mechanism is mortality which is
shared with vehicular accidents, so accidents are a contributor but are not sufficient.
The source of the increased predation is much less clear. However, the availability of
prey and other food are likely contributors. The coyote control program is a likely
source of the decline in coyote abundance that ended the kit fox decline, but reduced
prey abundance may have also contributed.
68
-------
Vegetation
Cover and
Quality
Control
Program
Time
Hunting
Coyote
Abundance
^ Mortality ^
31
(Kit Fox ^
Abundance J
Prey
Abundance
Vehicular
Accidents
Predation
Coyote
Habitat Quality
Lagomorph
Habitat Quality
FIGURE 16
The Final Conceptual Model for the Cause of the Kit Fox Decline. The thickness of the
arrow lines indicates the degree of confidence in the causal connection.
69
-------
7. LATER STUDIES AND OTHER ATTRIBUTIONS OF CAUSE
The kit fox demographic studies continued on the NPR sites until 1995
(Cypher et al., 2000). Kit fox numbers rose from 1991-1994, reaching nearly the same
level as in 1981. They then dropped in 1995 to the same levels as in the mid 1980s.
Prey abundances followed a similar pattern. Kit fox numbers were very low in 1991 due
to a preceding period of consistently dry years with very low vegetation production, and
the recovery was associated with higher precipitation.
Spotlight survey of kit foxes on the Elkhorn Plain (on the northeastern side of the
Carizo Plain; Figure 4) by the California Department of Fish and Game found a decline
of approximately 50% between 1980 and1994 (Ralls and Eberhardt, 1997). In contrast,
1994 was a year of extremely high kit fox abundance on the Naval Petroleum Reserves
(Cypher et al., 2000). This is weak evidence, but it supports the idea that local forces
can override a regional source such as climate.
The U.S. DOE's (1993) supplemental EIS attributed the decline to precipitation,
based on a comparison of the 3 and 5 years before 1981 to those after and to
unspecified effects of development. They confused the nearly constant proportion of
mortality that was due to predation with a nearly constant predation rate. This error
created the mistaken impression that there was not sufficient variation in predation to
cause the variation in kit fox mortality or abundance.
Cypher et al. (2000) thoroughly reviewed available information concerning the
cause of variance in kit fox abundance. They concluded that kit fox abundance was
driven by precipitation in the previous year. However, they lumped data from kit foxes
on both NPRs and some adjoining areas and they included data out to 1995. The
differences between their results and the results of this analysis of the decline on NPR-1
in the early 1980s illustrate the importance of scale in causal analysis.
70
-------
8. CONCLUSIONS
The available evidence indicates that the cause of the kit fox decline in the early-
to-mid 1980s was increased predation by coyotes. The cause of the increased
predation appears to have been increased coyote abundance and, when prey declined,
increased kit fox susceptibility due to increased time out of the dens, hunting. The
cause of the increased coyote abundance is unclear. The kit fox decline ended after a
coyote control program was instituted and coyote numbers declined.
The elimination of toxicants and diseases as causes has practical management
implications. No additional measures need be taken to eliminate exposures to toxicants
or to reduce the introduction of pathogens. The prior assessment of contaminant risks
to kit foxes was sufficient to allay the concerns of stakeholders (Suter et al., 1992).
However, this assessment is superior in two respects. First, the use of a formal causal
analysis method provides greater assurance of the quality of the results. Second,
identification of the likely proximate cause provides increased confidence that the
negative results for contaminants were not a result of inadequate data or analysis.
The implications of coyote predation for management are less clear, because the
cause of the increase in coyotes is unclear. However, the finding that precipitation is
not absolutely or invariably determinate of kit fox abundance should encourage
management actions. These might include revegetation to increase prey abundance,
preservation of kit fox dens which provide cover from predators, coyote control, and, in
extreme situations, supplemental feeding. All of these were practiced on NPR-1 for
some time and to some degree, but it is not clear how successful any of them were.
However, the endangered status of kit foxes could justify adaptive management studies
to determine the most efficacious practices.
71
-------
9. LESSONS LEARNED
Be clear about the difference that defines the impairment. In prior Stressor
Identification and CADDIS case studies, the impairments have been defined as a
condition that was considered impaired by comparison to a reference site or a regional
reference. In this case, the impairment was defined as a decline in abundance over a
defined temporal interval.
Obtain a baseline. When, as in this case, an impairment is defined by a time
series, it is particularly desirable to include a time period prior to the onset of the
impairment. In this case, that would have meant beginning the monitoring activities on
NPR-1 prior to the increased development in 1974, or at least before implementing the
legal mandate for production at the maximum efficient rate in 1976. A semi-quantitative
survey of kit foxes was performed in 1979 and the kit fox demographic surveys began in
1981 (eight years after development increased).
Monitor the potential causes and sources. Lagomorph monitoring began
along with kit fox monitoring, but full prey monitoring began two years later. Vegetation
production was monitored only in a small plot seven years later, and precipitation and
soil moisture were not monitored on site. Regular coyote surveys began in 1985. Site
contamination was measured erratically, but contamination studies aimed at kit fox
exposures did not begin until nine years later. Ideally, a problem formulation should
precede any monitoring program including the development of conceptual models of the
hypothesized causal relationships.
Allow for time lags when analyzing evidence of temporal co-occurrence.
Although this advice occurs in CADDIS, lags have not been demonstrated in prior case
studies. In the long-term multi-site study, prey abundance lagged one year and kit fox
abundance lagged two years behind precipitation (Cypher et al., 2000). Time lags in
this study were less clear, but also ranged from 0-2 years.
Avoid spatially or temporally diluting causal relationships. The impetus for
assessing kit fox abundance on NPR-1 was to determine the effects of the increase in
production as required by NEPA and then to determine the cause of the observed
decline in the early 1980s in response to the U.S. FWS's 1987 Biological Opinion.
However, the final causal analysis commissioned by the U.S. DOE lumped NPR-1 with
NPR-2 and modeled combined data from 1983-1995 (Cypher et al., 2000). Because
the analysis was spatially and temporally extensive, it identified a spatially and
temporally extensive cause, precipitation. However, precipitation does not explain the
biological trends in the early 1980s.
Consider the mandate for the causal assessment. If the mandate is to
assess contaminants as a cause, as in a Superfund assessment, this causal
assessment is complete. Contaminants were not supported as a cause by the
evidence, and other causes were. However, if the mandate were to manage the kit fox
72
-------
population, the assessment narrows the range of concerns and is suggestive of
management options but is not conclusive.
Consider alternative causes. Although the evidence for contaminants was
weak to negative, they might still have been suspect if alternative causes had not been
supported by evidence.
Use internal measures of exposure (i.e., body burdens, biomarkers, and
immunological markers) when possible. If sufficient reference data are available for
comparison, concentrations of contaminants in biological samples can be used to
determine whether exposures are occurring. Specific contaminants that are not
elevated can be eliminated. Biomarkers that are specific to a chemical or class of
chemicals could be used equivalently. However, body burdens and biomarkers of
exposure are, at best, weak positive evidence unless they can be linked to effects in an
exposure-response relationship. Immunological markers can be used equivalently to
determine whether organisms have been infected by specific pathogens.
Obtain data from multiple reference sites. Comparisons of elemental
concentrations in fox fur from developed and undeveloped locations were misleading.
That was not apparent until data were obtained from other reference sites.
Source identification may need to be integrated. Stressor Identification and
CADDIS were designed to determine the most likely proximate cause in impaired water
bodies. Under the Clean Water Act, sources are identified and the pollution load is
apportioned among them in a separate step after the cause is identified. However, in
some contexts of causal analysis, it is expected that the source will be identified along
with the cause. In addition, sources must be distinguished in some cases to identify the
proximate cause. The separation of causal pathways from climate and petroleum
development was important because some evidence clearly supports one causal
pathway and not the other. Finally, source identification suggests which management
actions are likely to be successful. This case study shows how that integration can be
accommodated within the CADDIS methodology, but the approach depends on the
nature of the cause. If the cause had been a chemical contaminant or pathogen, a
separate step would have been required for source identification with a different
inferential approach. However, if the cause is a change in a species such as an
increase in a predator or a decrease in prey, then the same inferential approach may be
used. In addition, source assessment is constrained by the definition of the cause. If
the proximate cause can be clearly defined, as with prey abundance, the integration is
easy, because there are few sources and only the causal pathways are different.
However, habitat quality for kit foxes cannot be clearly defined or quantified, so the
entire analysis is devoted to aspects of the sources of habitat modification. In the case
of predation, the sources of increased coyote abundance were potentially numerous
and unclear (because coyotes were counted but not studied), so a formal analysis of
sources could not be performed.
73
-------
Adapt the SI and CADDIS methodology as needed. Although the essential
features of the methodology (comparison of multiple candidate causes by weighing
multiple types of evidence using a formal scoring system) must be retained, the details
should be adapted to fit the case at hand. Two modifications were used in this case.
(1) The demographic modeling results did not fit any of the standard types of evidence,
so a new type (mechanistic sufficiency) was developed for this case. This type of
evidence is used when causal mechanisms can be identified and quantified and differ
among the candidate causes. (2) In addition, a new explanation of inconsistent
evidence was developed. The SI and CADDIS guidance allows for the "reasonable
explanation of the evidence" to explain how all evidence could be consistent (the cause
is true) or inconsistent (the cause is eliminated) if certain suppositions are true. In this
case we explain how inconsistent evidence could be shown to be consistently positive
or negative if the candidate cause were a contributor to the true cause. For example, it
could be part of the causal pathway to the true cause or it could act additively with the
true cause through a common mode of action.
74
-------
10. REFERENCES
Barker, D.H., A.C. Rencher, B.M. Mittal, S.V. Shanbhag, E.N. Sharma and L.S. Sharma.
1976. Metal concentrations in human hair from India (Pilani, Rajasthan). In: Trace
Substances in Environmental Health, Vol 10. D. Hemphill, Ed. University of Missouri
Press, Columbia, MO. p. 71-81.
Bryant, M.M. 1943. Area determination with modified acreage grid. J. Forestry
41:764-766.
Bjurlin, C.D. and B.L. Cypher. 2003. Effects of roads on San Joaquin kit foxes: A
review and synthesis of existing data. pp. 397-406. In: Irwin, C.L., P. Garrett, and K.P.
McDermott (eds.). International Conference on Ecology and Transportation. North
Carolina State University, Raleigh, NC. Available at
http://repositories. cdlib.org/ca i/viewcontent. cai?article=1125&context=imie/roadeco
Cypher, B.L. 1993. Food item use by three sympatric canids in Southern Illinois.
Trans. Illinois State Acad. Sci. 86:139-144.
Cypher, B.L. and K.A. Spencer. 1998. Competitive interactions between coyotes and
San Joaquin kit foxes. J. Mammal. 79:204-214.
Cypher, B.L., G.D. Warrick, M.R.M. Otten et al. 2000. Population dynamics of San
Joaquin kit foxes at the Naval Petroleum Reserves in California. Wildlife Monographs,
No. 145. 43 pp.
Doi, R. 1973. Environmental mercury pollution and its influence in the cities of Japan.
Annual Report Tokyo Metropolitan Research Institute of Environmental Protection
3:257-261.
EG&G/EM (EG&G Energy Measurements). 1992. Endangered species program, Naval
Petroleum Reserves in California, Annual report FY92. U. S. Department of Energy
Topical Report, EG&G/EM Santa Barbara Operations Report No. EGG-10617-2166.
44 pp.
Egoscue, H.J. 1975. Population dynamics of the kit fox in western Utah. S. Calif.
Acad. Sci. Bull. 74:122-127.
Floit, S.B. and L.W. Barnthouse. 1991. Demographic analysis of a San Joaquin kit fox
population. Oak Ridge National Laboratory, Oak Ridge, TN. ORNL/TM-11679.
Grinnell, J., J.S. Dixon and J.M. Linsdale. 1937. Kit foxes. In: Fur-bearing animals of
California, Vol. 2, J. Grinnell, J.S. Dixon and J.M. Linsdale, Ed. University of California
Press, Berkeley, California, p. 399-420.
75
-------
Harris, C.E., T.P. O'Farrell, P.M. McCue and T.T. Kato. 1987. Capture-recapture
estimation of San Joaquin kit fox population size on Naval Petroleum Reserve #1, Kern
County, California. U. S. Department of Energy Topical Report, EG&G/EM Santa
Barbara Operations Report No. EG&G 10282-2149. 42 pp.
Harrison, D.J., J.A. Bissonette and J.S. Sherburne. 1989. Spatial relationships
between coyotes and red foxes in Eastern Maine. J. Wildl. Manage. 53:181-185.
Huckabee, J.W., F.O. Cartan and G.S. Kinnington. 1972. Environmental influence on
trace elements in hair of 15 species of mammals. U. S. Department of Energy Topical
Report, Oak Ridge National Laboratory Report No. ORNL/TM-3747.
Jenkins, D.W. 1979. Toxic trace metals in mammalian hair and nails. U.S.
Environmental Protection Agency, Las Vegas, NV. EPA-600/4-79-049.
Lenihan, J. 1978. Hair as a mirror of the environment. P. In: Measuring and
Monitoring the Environment. J. Lenihan and W.W. Fletcher, Ed. Blackie, Glasgow, p.
66:86.
Lewis, T.R. 1972. Effects of air pollution on livestock and animal products. In: Helena
Valley, Montana Area Pollution Study, U. S. Environmental Protection Agency, Office of
Air and Water Programs, Research Triangle Park, NC. p. 113-124.
McCue, P.M. and T.P. O'Farrell. 1986. Serologic survey for disease in endangered
San Joaquin kit fox, Vulpes macrotis mutica, inhabiting the Elk Hills Naval Petroleum
Reserve, Kern County, California. U. S. Department of Energy Topical Report,
EG&G/EM Santa Barbara Operations Report No. EGG 10282-2110. 26 pp.
McCue, P.M. and T.P. O'Farrell. 1987. Hematologic values of the endangered San
Joaquin kit fox, Vulpes macrotis mutica. J. Wildl. Dis. 23(1): 144-151.
McCue, P.M. and T.P. O'Farrell. 1988. Serological survey for selected diseases in the
endangered San Joaquin kit fox (Vulpes macrotis mutica). J. Wildl. Dis. 24(2):274-281.
McCue, P.M. and T.P. O'Farrell. 1992. Serum chemistry values of the endangered San
Joaquin kit fox (Vulpes macrotis mutica). J. Wildl. Dis. 28(3):414-418.
Major, J.T. and J.S. Sherburne. 1987. Interspecific relationships of coyotes, bobcats,
and red foxes in Western Maine. J. Wildl. Manag. 51:606-616.
McCoy, C.P. and W.C. Edwards. 1980. Sodium ion poisoning in livestock from oil field
wastes. Bov. Pract. 15:152-154.
Mosby, H.S. 1980. Reconnaissance mapping and map use. In: Wildlife Management
Techniques Manual, 4th ed, S.D. Schemnitz, Ed. Wildlife Society, Bethesda, MD.
p. 277-290.
76
-------
Nicholson, W.S. and E.P. Hill. 1984. Mortality in gray foxes from East-Central
Alabama. J. Wildl. Manage. 48:1429-1432.
O'Farrell, T.P. 1980. Elk Hills endangered and threatened species program, phase 1
progress summary. U. S. Department of Energy Topical Report, EG&G/EM Santa
Barbara Operations Report No. EGG 1183-2403. 19 pp.
O'Farrell, T.P., C. E. Harris, T. T. Kato and P. M. McCue. 1986. Biological assessment
of the effects of petroleum production at maximum efficient rate, Naval Petroleum
Reserve #1 (Elk Hills), Kern County California, on the endangered San Joaquin kit fox,
Vulpes macrotis mutica. U. S. Department of Energy Topical Report, EG&G/EM Santa
Barbara Operations Report No. EGG 10282-2107. 76 pp.
Orheim, R.M., L. Lippman, C.J. Johnson and H.H. Bovee. 1974. Lead and arsenic
levels of dairy bovine in proximity to a copper smelter. Environ. Lett. 7(3):229-236.
Petering, H.G., D.W. Yeager and S.O. Witherup. 1971. Trace metal content of hair. I.
Zinc and copper content of human hair in relation to age and sex. Arch. Environ. Health
23(3):202-207.
Ralls, K. and L.L. Eberhardt. 1997. Assessment of abundance of San Joaquin kit foxes
by spotlight surveys. J. Mammal. 78(1):65-73.
Ralls, K. and P.J. White. 1995. Predation on San Joaquin kit foxes by larger canids. J.
Mammal. 76:723-729.
Sargent, A.B., S.H. Allen and J.O. Hastings. 1987. Spatial relations between sympatric
coyotes and red foxes in North Dakota. J. Wildl. Manage. 51(2):285-293.
Schroeder, H.A., D.V. Frost and J.J. Balassa. 1970. Essential trace elements in man:
Selenium. J. Chronic Dis. 23:227-243.
Scott-Brown, J.M., S. Herrero and J. Reynolds. 1987. Swift fox. In: Wild Furbearer
Management and Conservation in North America, M. Novak, J.A. Baker, M.E. Obband
and B. Malloch, Ed. Ministry of Natural Resources, Toronto, Ontario, p. 432-441.
Scrivner, J.H., T.P. O'Farrell and T.T. Kato. 1987. Diet of the San Joaquin kit fox,
Vulpes macrotis mutica, on Naval Petroleum Reserve #1, Kern County, California,
1980-1984. U. S. Department of Energy Topical Report, EG&G/EM Santa Barbara
Operations. Report No. 10282-2168. 26 pp.
Suter, G.W., II, M.K. Gustin, S.B. Floit et al. 1988. Investigations of relationships
between oil field materials and practices and wildlife: Progress report - May 1988-
October 1988. Oak Ridge National Laboratory, Oak Ridge, TN. ORNL/M-659.
77
-------
Suter, G.W., II, A.E. Rosen, J.J. Beauchamp and T.T. Kato. 1992. Results of analyses
of fur samples from the San Joaquin kit fox and associated soil and water samples from
the Naval Petroleum Reserve No. 1, Tupman, California. U. S. Department of Energy
Topical Report, Oak Ridge National Laboratory Report No. ORNL/TM-12244. 199 pp.
Available at http://www.osti.gov/energvcitations/servlets/purl/6661412-
1 Ue7vh/6661412 PDF.
Taylor, F.G., L.K. Mann, R.C. Dahlman and F.L. Miller. 1975. Environmental effects of
chromium and zinc in cooling-water drift. In: Cooling Tower Environment - 1974. U.S.
Energy Research and Development Administration Symposium, Washington, DC.
Conference Report 740302. p. 408-426.
Therkelsen, R.L. 1972. Wildlife habitat enhancement of oil fields in Kern County,
California. California Division of Oil and Gas, Sacramento, California. 15 pp.
Thomas, N.J., W.J. Foreyt, J.F. Evermann, L.A. Windberg and F.F. Knowlton. 1984.
Seroprevalence of canine parvovirus in wild coyotes from Texas, Utah, and Idaho.
(1972 to 1983). J. Am. Vet. Med. Assoc. 185(11): 1282-1287.
U.S. DOE (U.S. Department of Energy). 1979. Final Environmental Impact Statement,
Petroleum production at maximum efficient rate, Naval Petroleum Reserve No. 1 (Elk
Hills), Kern County, California. U. S. Department of Energy, Tupman, CA.
DOE/EIS-0012.
U.S. DOE (U.S. Department of Energy). 1993. Supplemental Environmental Impact
Statement, Petroleum production at maximum efficient rate, Naval Petroleum Reserve
No. 1 (Elk Hills), Kern County, California. U.S. Department of Energy, Tupman, CA.
DOE/EIS-0158. Available at
http://www.osti.gov/bridge/servlets/purl/10167697-YskxJJ/native/10167697.PDF.
U.S. EPA. 2000. Stressor identification guidance document. U.S. Environmental
Protection Agency, Washington, DC. EPA/822/B-00/025. Available at
http://www.epa.gov/waterscience/biocriteria/stressors/stressorid.pdf.
Zoellick, B.W., T.P. O'Farrell, P.M. McCue, C.E. Harris and T.T. Kato. 1987.
Reproduction of the San Joaquin kit fox on Naval Petroleum Reserve #1, Elk Hills,
California, 1980-1985. U.S. Department of Energy Topical Report, EG&G/EM Santa
Barbara Operations Report No. 10282-2144. 42 pp.
78
------- |