Animal Conservation (1998) 1, 179–184 © 1998 The Zoological Society of London Printed in the United Kingdom

Variation in annual estimates of effective population size for

San Joaquin kit foxes

Mark R. M. Otten1 and Brian L. Cypher2

1
Endangered Species Research Program, Naval Petroleum Reserves in California, P.O. Box 178, Tupman, CA 93276, USA
2
P.O. Box 9622, Bakersfield, CA 93389, USA
(Received 27 September 1997; accepted 8 March 1998)

Abstract
We used a demographic model to calculate a series of annual Ne/N estimates ((Ne/N)S) for a popula-
tion of the endangered San Joaquin kit fox (Vulpes macrotis mutica) and to identify the demographic
parameters explaining significant portions of the variation in (Ne/N)S estimates. Demographic data
were collected between 1981 and 1995 from adult kit foxes located on the Naval Petroleum Reserves
in California. Demographic parameters were estimated by determining the sex of captured adult foxes
and monitoring the survival and reproduction of adult foxes fitted with radiocollars. Individual demo-
graphic parameters explaining significant portions (P < 0.10) of the variation in (Ne/N)S estimates was
determined using forward stepwise multiple regression. Estimates of (Ne/N)S were highly variable,
ranging from 0.22 to 1.15, (x̄ = 0.67, SEM = 0.07, n = 14), with a harmonic mean of 0.55. Forward
stepwise multiple regression analysis indicated that mean annual female survival (P = 0.002) and
mean annual female reproductive success (P = 0.004) explained significant portions of the variation
in estimates of (Ne/N)S, while mean annual sex ratio (P = 0.159), mean annual litter size (P = 0.406)
and mean annual male survival (P = 0.922) did not. Long-term kit fox conservation strategies will
probably have to account for wide variation in (Ne/N)S ratios and focus on maintaining or expanding
the size and connectivity of populations.

(if any) actual populations operate as ideal, Ne is typi-

INTRODUCTION
Maintenance of genetic variation (heterozygosity) can be
problematic for endangered species, which often persist
as small populations, in fragmented habitats, or both.
Over time, finite populations suffer decreased heterozy-
gosity from the cumulative effects of inbreeding and ran-
dom genetic drift (Kimura & Crow, 1963). Small or
genetically isolated populations tend to lose heterozy-
gosity more rapidly owing to small breeding stock and
limited dispersal. If not minimized, this loss of genetic
variability can quickly lead to short-term inbreeding
depression and, ultimately, to long-term loss of genetic
adaptive potential or extinction (Franklin, 1980). As a
result, the amount of heterozygosity transmitted between
generations is an important consideration in the conser-
vation of small or endangered populations (Lande &
Barrowclough, 1987; Nunney & Elam, 1994).
One means of assessing the amount of heterozygosity
maintained in a population is through the estimation of
effective population size, denoted Ne (Wright, 1931,
1938). Effective population size is defined as the ‘. . . size
of the ideal population that would undergo the same
amount of random genetic drift . . . as the actual popu-
lation’ (Lande & Barrowclough, 1987: 89). Because few
cally somewhat smaller than the census population size
(N). Measuring Ne in wildlife populations, however, is
often problematic, particularly for long-lived species with
overlapping generations (Vucetich, Waite & Nunney,
1997). Consequently, various techniques using both
genetic and ecological methods have been developed to
estimate loss of genetic diversity (reviewed by Caballero,
1994), often as the theoretical ratio of effective popula-
tion size to census (usually adult-only) population size
(Ne/N). While genetic techniques typically estimate Ne/N
by measuring changes in allele frequency over time or
successive generations, ecological methods utilize demo-
graphic characteristics to estimate Ne/N (Nunney & Elam,
1994). Ecological methods for estimating Ne/N may be
particularly useful for populations of long-lived endan-
gered vertebrates. Because they often have extensively
overlapping generations and long generation times,
demographic data for these species are often more read-
ily available than sufficient genetic data.
A species for which conservation of genetic variabil-
ity may be problematic is the endangered San Joaquin
kit fox (Vulpes macrotis mutica). Although it once
roamed over much of central California, widespread
conversion of wildland to agricultural and urban use has
180 M. R. M. OTTEN & B. L. CYPHER
recently restricted the distribution of this species to rem-
nant valley grassland/shrub habitats (US Department of
the Interior, 1983). Much of the remaining habitat is
severely fragmented with kit foxes persisting in numer-
ous disjunct population patches of variable, and largely
unknown, size (Williams et al., 1997). Although poorly
understood, the amount of connectivity between popu-
lations is probably minimal and inconsistent (Williams
et al., 1997), resulting in a high degree of patch isola-
tion and limited potential for genetic exchange. This type
of patch structure may potentially increase population
susceptibility to detrimental inbreeding effects and to
extinction from stochastic events and chance catastro-
phes (Mills & Allendorf, 1996). Therefore, some mea-
sure of the rate at which genetic diversity is being lost
from kit fox populations may be critical for the effec-
tive conservation of the species.
In this study, we estimated the annual Ne/N ratio, over
a 15-year period, for a San Joaquin kit fox population
on the Naval Petroleum Reserves in California (NPRC).
Estimates of annual Ne/N (hereafter denoted (Ne/N)S)
were derived from estimates of annual demographic
parameters. This information is potentially useful in the
crafting, refinement and prioritization of species recov-
ery strategies. In addition, we determined which of the
measured demographic parameters most significantly
influenced variation in estimates of (Ne/N)S. Identi-
fication of the parameters most significantly influencing
variation in (Ne/N)S estimates is potentially valuable for
devising species management strategies by which effec-
tive population size may be increased.
STUDY AREA AND METHODS
The NPRC extends over approximately 315 km2 of semi-
arid grassland/shrub habitat and is located 40 km
west/south-west of Bakersfield, California. Land uses are
a mixture of well-developed oil fields interspersed with
large areas of relatively undeveloped habitat. The vege-
tative community on NPRC is broadly described as
belonging to the Allscale Series and is composed pri-
marily of annual grasses, annual forbs and xerophytic
shrubs (Sawyer & Keeler-Wolf, 1995). The dominant
grasses are red brome (Bromus madritensis) and mouse-
tail fescue (Festuca myuros), the dominant forbs are red-
stemmed filaree (Erodium cicutarium) and fiddleneck
(Amsinkia menziesii), and the dominant shrubs are desert
saltbush (Atriplex polycarpa) and bladderpod (Isomeris
arborea). The climate of the NPRC region is
Mediterranean and is characterized by hot, dry summers
and mild, wet winters (Major, 1977). Annual precipita-
tion varies widely with a mean of 15.1 cm, most of which
falls between November and April (National Oceanic
and Atmospheric Administration, 1996). Over the dura-
tion of this study, annual precipitation ranged between
7.3 and 27.6 cm (x̄ = 16.1 cm, SEM = 1.6 cm, n = 15).
We used the minimal demographic model of Nunney
& Elam (1994) to derive estimates of (Ne/N)S for all years
between 1981 and 1995, except for 1993 when insuffi-
cient data were collected. Cumulative and annual esti-
mates of Ne were not calculated due to the lack of adult-
only estimates of N. The minimal demographic model
approximates the effective population size equation of
Hill (1972, 1979) through the assumption that variation
in female fecundity and male mating success are due to
seasonal random factors (Nunney & Elam, 1994).
Estimation of (Ne/N)S required the prior estimation of
eight model parameters, and is given by:
 Ne  =
 N S
4r (1 − r)T
, (1)
rAf (1 + IAf ) + (1 − r) Am (1 + IAm ) + (1 − r)Ibm + rIbf
where, r is the male proportion of the adult sex ratio, T
is the generation time, Af and Am are the mean female
and male adult life spans, respectively, IAf and IAm are
the standardized variances in female and male adult life
span, respectively, and Ibf and Ibm are the standardized
variances in female and male reproductive success per
season, respectively. Estimation of (Ne/N)S assumed age-
independent (Type II) adult survival, age-independent
adult fecundity, that all adult females attempt to breed,
that all adult males attempt to breed, and that juvenile
survival to maturity is distributed in a Poisson manner
(Nunney & Elam, 1994).
Minimal demographic model parameters were derived
from estimates of seven San Joaquin kit fox demographic
parameters. Male proportion of the adult sex ratio (r) was
estimated annually by determining the sex of kit foxes
known to be adults captured during population monitor-
ing conducted on NPRC each winter. Captured foxes that
could not be aged with certainty were excluded from esti-
mation of r. Mean annual female survival rate (vf), mean
annual male survival rate (vm), observed mean annual lit-
ter size (x), and mean annual female reproductive success
rate (pb) were estimated by monitoring the survival and
reproductive success of radiocollared adult foxes. An indi-
vidual fox was considered to have survived through a par-
ticular year if it wore an operable radiocollar and was
known to be alive on February 15 of the following year.
Mean annual female reproductive success (the percentage
of females successfully rearing ≥ one offspring) and x were
assessed through observations of natal dens containing
≥ one radiocollared adult fox. Litter size was evaluated
based on the number of live pups observed at den sites
after weaning. Expected mean litter size (bf) was also cal-
culated as a function of x in order to estimate variance in
female reproductive success per season (Nunney & Elam,
1994). Female and male maturation times to adulthood (Mf
and Mm, respectively) were estimated from existing liter-
ature and assumed to be equal and constant for all years.
We calculated the arithmetic mean and the harmonic
mean of (Ne/N)S estimates as measures of central ten-
dency. A harmonic mean (x̄h) may be employed when
averaging ratios (Zar, 1984) and was calculated as:
xh = 1 , (2)
1 
t  ∑ ( Ne / N )S (i) 
1
 Effective population size in kit foxes 181

where t is the number of estimates and (Ne/N)S(i) is the
estimated (Ne/N)S ratio for year i. In effective population
size studies, the harmonic mean estimate has typically
been used to adjust Ne to account for fluctuations in N
(Lande & Barrowclough, 1987; Nunney, 1996). In this
study, however, we employed it to account for annual
fluctuations in demographic parameter estimates and as
a potential means of identifying a ‘typical’ annual Ne/N
ratio for the kit fox population on NPRC.
We used forward stepwise multiple regression analy-
sis (Zar, 1984) to determine the demographic parameters
explaining significant portions of the variation in (Ne/N)S
estimates. Annual estimates of r, vf, vm, x and pb were
independent variables, while estimates of (Ne/N)S was
the dependent variable. Because Mf and Mm were held
constant across years, and bf varied as a function of x,
they were not included as independent variables during
regression analyses. Independent variables were added
to the multiple regression equation using a P-to-enter
value of 0.10. That is, a variable was added to the equa-
tion if its inclusion resulted in a significant (P < 0.10)
increase in r2.
RESULTS
Five of seven demographic parameters (r, vf, vm, x and
pb) were estimated annually for adult San Joaquin kit
foxes on NPRC between 1981 and 1995 (Table 1). The
range and arithmetic mean of estimates were consistent
with published values (Morrell, 1972; Egoscue, 1975;
McGrew, 1979; Ralls & White, 1995). Coefficients of
variation (c.v.) indicted that vf (c.v. = 0.482) and pb (c.v.
= 0.442) varied more widely than vm (c.v. = 0.335),
r (c.v. = 0.268), or x (c.v. = 0.195). Literature indicated
that female and male maturation time to adulthood is 22
months (Morrell, 1972; McGrew, 1979), where adult-
hood was defined as the age at which ‘. . . breeding suc-
cess reaches a recognizable adult level’ (Nunney &
Elam, 1994: 178). Thus, the value of both Mf and Mm
was 1.83 for all 14 years. The high estimate of vf for
1992 (1.00) may be an artifact of low sample size and
possibly represents an outlier. Therefore, we recalculated
the harmonic mean of (Ne/N)S and performed an addi-
tional stepwise regression analysis with data from 1992
excluded.
Estimates of (Ne/N)S for adult San Joaquin kit foxes
on NPRC varied widely, ranging from 0.22 to 1.15
(Table 1), with an arithmetic mean of 0.67 (SEM = 0.07,
n = 14). The harmonic mean estimate was 0.55 (Table
1) and is similar to short-term Ne/N estimates for other
monogamous species with overlapping generations
(Nunney & Elam, 1994). When data from 1992 were
excluded, the minimum estimated (Ne/N)S was 0.31 with
an arithmetic mean of 0.70 (SEM = 0.06, n = 13) and a
harmonic mean of 0.63. The difference in harmonic
means with inclusion and exclusion of 1992 data illus-
trates the tendency of harmonic mean estimates to give
greater weight to low value ratios (Lande &
Barrowclough, 1987).
Forward stepwise multiple regression analysis indi-
cated that vf and pb explained significant portions of the
variation in estimates of (Ne/N)S, while r, x and vm did
not (Table 2). Inclusion of vf in the regression equation
at step 1 and of vf and pb at step 2 explained approxi-
mately 36 and 70% of the variation observed in (Ne/N)S,
respectively (Table 2). Subsequent inclusion of r, x and
vm in the regression model at step 5 (not shown in Table
2) only accounted for an additional 7% (r2 = 0.767) of
the variation. When 1992 data were excluded, only pb
(P = 0.001) explained a significant portion of the varia-
tion in (Ne/N)S estimates, while vf (P = 0.209), r (P =
0.657), x (P = 0.287) and vm (P = 0.180) did not. This

Table 1. Annual estimates (sample size), arithmetic mean (x), standard error (SEM) and harmonic mean (x̄ h) of five demographic parameters
and annual effective population size/census population size ((Ne/N)S) for adult San Joaquin kit foxes on the Naval Petroleum Reserves in
California, 1981–1995

Demographic parameter estimate

Year r vf vm x pb (Ne/N)S

1981 0.43 (23) 0.67 (21) 0.69 (22) 4.40 (5) 0.36 (11) 0.52
1982 0.58 (36) 0.52 (35) 0.49 (22) 4.78 (11) 0.73 (15) 0.77
1983 0.63 (8) 0.40 (34) 0.42 (23) 3.71 (14) 0.75 (16) 0.78
1984 0.33 (21) 0.22 (47) 0.21 (31) 4.45 (13) 0.91 (11) 0.93
1985 0.35 (17) 0.35 (28) 0.30 (22) 3.17 (13) 0.75 (12) 0.81
1986 0.64 (39) 0.14 (13) 0.45 (11) 3.50 (6) 0.50 (2) 0.65
1987 0.44 (25) 0.35 (11) 0.27 (20) 4.00 (8) 1.00 (5) 1.15
1988 0.37 (30) 0.56 (11) 0.26 (17) 3.63 (9) 0.43 (7) 0.58
1989 0.53 (30) 0.50 (7) 0.25 (6) 3.75 (5) 0.50 (4) 0.72
1990 0.37 (27) 0.39 (29) 0.40 (7) 3.50 (3) 0.33 (6) 0.47
1991 0.67 (6) 0.62 (19) 0.55 (6) 3.00 (1) 0.20 (15) 0.31
1992 0.78 (9) 1.00 (8) 0.44 (3) 3.83 (6) 1.00 (6) 0.22
1993 — — — — — —
1994 0.55 (67) 0.53 (18) 0.48 (13) 3.00 (8) 0.78 (9) 0.83
1995 0.51 (53) 0.20 (20) 0.37 (15) 2.00 (1) 0.29 (7) 0.61
x̄ 0.51 0.46 0.40 3.62 0.61 0.67
SEM 0.04 0.06 0.04 0.19 0.07 0.07
x̄ h — — — — — 0.55

Demographic variables are: male proportion of the sex ratio (r), female survival rate (vf), male survival rate (vm), mean litter size (x) and female reproductive success rate
(pb). n is given in parentheses for each variable. Insufficient data were collected in 1993 to estimate any of the demographic or population parameters.
182
Table 2. Forward stepwise multiple regression analysis to determine the demographic parameters significantly influencing variation in annual
estimates of effective population size/census population size ((Ne/N)s) for adult San Joaquin kit foxes on the Naval Petroleum Reserves in
California, 1981–1995
Step 0
Parameter Coefficient P Coefficient
vf –0.597a 0.024
pb 0.516a 0.059
r –0.466a 0.093
x 0.226a 0.436
vm –0.499a 0.069
r2
a
Partial correlation coefficient.
Demographic variables are: male proportion of the sex ratio (r), female survival rate (vf), male survival rate (vm), mean litter size (x) and female reproductive success rate
(pb). Values presented are regression coefficients, t test P values (P) and proportion of variability accounted for (r2).
pattern suggests that, while pb appears to strongly influ-
ence variation in (Ne/N)S, the influence of other para-
meters (particularly vf) may partly depend on the amount
of variation in the parameter itself. Annual female repro-
ductive success accounted for approximately 90% (r2 =
0.896) of the observed variation in (Ne/N)S when 1992
data were excluded.
DISCUSSION
Our results indicate that demographic parameter values
for San Joaquin kit foxes on NPRC varied considerably
on an annual basis. While some variation may be due to
limited sample sizes for some years (particularly with
regard to estimates of x), annual variation in demo-
graphic parameters appears to be typical of most San
Joaquin kit fox populations. Annual variation in female
reproductive success, adult survival, and litter sizes has
been observed during other kit fox studies (White &
Ralls, 1993; Spiegel & Disney, 1996; Spiegel & Tom,
1996; White, Vanderbilt White & Ralls, 1996) and is
likely a response to fluctuating prey abundances (Cypher
et al., in review). We believe that most of the yearly
demographic parameter estimates are reasonable and fall
within the range of values likely to occur in the NPRC
population over time. A few extreme estimates, how-
ever, such as vf and r in 1992, are unusual and are prob-
ably unlikely to occur in the population as a whole.
Extreme values of pb, although unusual, have occurred
in other kit fox populations (White & Ralls, 1993;
Spiegel & Tom, 1996; White et al., 1996), and are also
probably a result of vixens responding to variable lev-
els of prey availability.
The substantial temporal fluctuation observed in
demographic parameters made use of the minimal demo-
graphic model (Nunney & Elam, 1994) somewhat prob-
lematic. The model estimates lifetime demographic
parameters from data collected over few breeding sea-
sons, and contains no mechanism to account for the
effects of parameter fluctuations on Ne. Large fluctua-
tion in some demographic parameters can profoundly
influence Ne (Nunney, 1996). One approach to this prob-
lem is to estimate a single Ne/N ratio from cumulative
mean parameter values. This approach, however, ignores
M. R. M. OTTEN & B. L. CYPHER
Step 1 Step 2
P Coefficient P
–0.660 0.024 –0.736 0.002
0.732a 0.004 0.539 0.004
–0.276a 0.361 –0.434a 0.159
0.471a 0.104 0.264a 0.406
–0.346a 0.247 –0.032a 0.922
0.356 0.701
temporal fluctuations in both demographic parameters
and N and is likely to overestimate Ne (Frankham, 1995;
Vucetich et al., 1997). Alternately, the population can
be partitioned and separate estimates of Ne/N calculated
based on lifetime parameter values. For species with
overlapping generations, cohorts can be considered sep-
arate units (Hill, 1972, 1979), and are probably the most
appropriate grouping. Concerns over small sample sizes
resulting from species-specific and study-related factors
(e.g. variable reproduction, replacement of radiocollars,
variable adult survival and individual trap response),
however, made dividing our data by cohort impractical.
We recognize that estimating Ne/N on an annual basis
probably underestimates Ne/N, and has limited utility for
estimating long-term Ne in the population (L. Nunney,
pers. comm.). However, annual estimates are useful for
gauging the magnitude of variation in (Ne/N)S likely to
be exhibited by a population.
We calculated the harmonic mean of (Ne/N)S estimates
to identify a ‘typical’ annual Ne/N ratio for the San
Joaquin kit fox population on NPRC. Given the relative
isolation of individual populations, this rate is potentially
useful as an indicator of patch susceptibility to short-
and long-term inbreeding effects. Fluctuations in demo-
graphic parameters observed during this study and at
other locations suggests that wide variation in (Ne/N)S
ratios may represent an intrinsic characteristic of all San
Joaquin kit fox populations. As a result, use of the har-
monic mean as an indicator of a ‘typical’ (Ne/N)S ratio
may not be practical. We caution, therefore, against for-
mulating species conservation or recovery strategies
based on just one or few (Ne/N)S estimates (see also
Vucetich & Waite, in press). Indeed any successful strat-
egy for maintaining the genetic viability of San Joaquin
kit fox populations must be robust enough to account
for expected variation in (Ne/N)S. As a conservative
guideline we suggest assuming an effective population
size of no more than ~20–30% of adult census popula-
tion size in any given year. This rate is similar to the
Ne/N assumed by Mace & Lande (1991) in their criteria
for categorizing threat to endangered species.
Our estimates of (Ne/N)S and the associated harmonic
mean should not be viewed as equivalent to an estimate
of long-term Ne/N. Recent research indicates that fluc-
 Effective population size in kit foxes
tuations in population size have a significant negative
affect on long-term Ne/N (Frankham, 1995; Vucetich et
al., 1997). In an analysis of 192 previously published
estimates, Frankham (1995) concluded that, through
time, the effects of fluctuations in population size, fam-
ily size and sex ratio are likely to decrease long-term
Ne/N to ~0.1. Our estimates of (Ne/N)S and the associated
harmonic mean, however, do not account for fluctuations
in population size and are not intended to be estimates
of long-term Ne/N. While estimates of adult-only abun-
dance on NPRC were not available, kit fox population
trend data (Cypher et al., in review) indicates that pop-
ulation size has fluctuated widely over the study period.
The precise effect of these fluctuations is not known, but
long-term Ne/N for kit foxes on NPRC and at other loca-
tions are likely to be considerably less than the harmonic
mean value we calculated. Recent work by Vucetich &
Waite (in press) may prove useful for incorporating fluc-
tuations in population size into demographic models for
assessing long-term Ne/N in small populations.
A suite of demographic characteristics including mat-
ing system, sex ratio, individual reproductive success,
longevity, fecundity and age at reproductive maturity can
potentially influence estimates of Ne/N (Franklin, 1980;
Lande & Barrowclough, 1987; Waite & Parker, 1996).
In order to quantify or assess these influences, previous
authors have focused primarily on how changes in
demography affect variance in progeny production (e.g.
Wright, 1938; Chepko-Sade & Shields, 1987; Nunney,
1996). Increased variance in progeny production results
in the unequal representation of genotypes in successive
generations, which deceases average heterozygosity in
the population (Nunney, 1993). Our results were con-
sistent with previous research findings and theoretical
expectations for monogamous mammalian species with
overlapping generations. We found that pb exerted a
strong, consistent, positive influence on variation in
(Ne/N)S estimates. Increased reproductive success rates
result in a decline in the mean relatedness of produced
offspring (Chepko-Sade & Shields, 1987), particularly if
litter sizes exhibit little variability. In monogamous
species, increased annual adult survival (vf and vm)
decreases Ne/N through the lengthening of generation
time, when the age at which individuals reach maturity
remains constant (Nunney, 1993; Waite & Parker, 1996).
While these effects are expected to be small (Nunney,
1993), our results suggest that vf, and possibly vm, may
explain significant portions of the variation in (Ne/N)S
estimates if annual survival is highly variable. Similarly,
r probably has little influence on Ne/N, unless it is atyp-
ical or heavily biased (Chepko-Sade & Shields, 1987;
Nunney, 1993).
Demographic parameters significantly affecting vari-
ation in (Ne/N)S estimates were identified as a potential
means of developing strategies to increase Ne/N ratios in
San Joaquin kit fox population patches. Realistically,
however, few of these parameters could be manipulated
through active management to produce the desired
effect. Some parameters (Mf, Mm and x) are wholly or
partly controlled by physiological processes and cannot
183
be substantially altered. In addition, manipulation of
some parameters to increase Ne/N, such as decreasing vf
and vm, would likely result in undesirable declines in
overall abundance. Of the demographic parameters we
measured, only pb could potentially be manipulated to
increase Ne/N without decreasing abundance. While they
may be warranted under certain circumstances, tech-
niques for increasing pb among kit foxes (e.g. habitat
management for prey species and predator control) may
be costly or impractical on the scale necessary to sig-
nificantly increase Ne/N.
One of the salient considerations for the conservation
of small or fragmented sexual populations is the mini-
mum Ne required to ensure genetic viability over time.
Franklin (1980) recognized two consequences of loss in
heterozygosity from random genetic drift: short-term
loss of fitness through inbreeding depression, and long-
term loss of genetic adaptive potential through the accu-
mulation of deleterious mutations. The minimum Ne
required for a population to retain sufficient short and
long-term genetic diversity, however, is not clear.
Preliminary work by Franklin (1980) and Soulé (1980)
suggested Ne ≥ 50 to ensure short-term viability. Recent
research by Latter & Mulley (1995), however, suggests
that an Ne of 50 is probably not adequate to prevent loss
of fitness due to inbreeding depression. Across longer
time scales, Lynch, Conery & Bürger (1995) and Lande
(1995) suggested an Ne of >1000 and ~5000, respec-
tively, to minimize the probability of extinction from the
accrual of mildly deleterious alleles. Experimental work
by Gilligan et al. (1997), however, suggests that, over
45–50 generations, populations with Ne values as low as
25 may be more susceptible to extinction from stochas-
tic factors than from mutation accumulation. Regardless
of the minimum Ne required, population conservation
goals are similar. Ultimately, both short and long-term
population persistence and viability will be enhanced by
maximizing population size and minimizing fragmenta-
tion and isolation of patches (Lynch et al., 1995).
Given the high degree of variation observed in demo-
graphic parameters and in estimates of (Ne/N)S, it is
uncertain whether individual San Joaquin kit fox popu-
lation patches can be maintained at an Ne necessary to
retain adequate short- and long-term heterozygosity. As
a result, some kit fox populations may be at considerable
risk to loss of fitness through inbreeding depression, and
to local extirpation through demographic/environmental
stochasticity or random catastrophic events. In the short
term, small, isolated populations in danger of rapid
inbreeding could require immediate attention, and may
benefit from the translocation of as few as one individ-
ual per generation (Mills & Allendorf, 1996). Long-term
conservation and recovery of the species, however, will
probably rely primarily on maintaining and expanding (if
possible) the size and connectivity of existing population
patches. Increasing population size, dispersal and inter-
population gene flow will typically result in an attendant
increase in Ne (Lande & Barrowclough, 1987). As a
result, conservation efforts and management techniques
such as increasing carrying capacity on existing habitats,
184 M. R. M. OTTEN & B. L. CYPHER
establishing and protecting permanent dispersal corri-
dors, and rehabilitating degraded or marginal habitats
should be given serious consideration. Large temporal
variation in estimates of (Ne/N)S, however, suggests that
basing recovery strategies on one or few estimates
should be avoided.
Acknowledgements
We thank K. Ralls, R. Frankham, L. Nunney and an
anonymous reviewer for valuable comments on drafts of
this manuscript. We also thank T. Kato, J. Vucetich and
G. Warrick for timely discourse and correspondence.
Demographic data were collected by EG&G Energy
Measurements, Inc. personnel between 1981 and 1995.
Data collection and manuscript preparation were spon-
sored by the United States Department of Energy, Naval
Petroleum Reserves in California and Chevron USA
Production Company.
REFERENCES
Caballero, A. (1994). Developments in the prediction of effective
population size. Heredity 73: 657–679.
Chepko-Sade, B. D. & Shields, W. M. (1987). The effects of dis-
persal and social structure on effective population size. In
Mammalian dispersal patterns: the effects of social structure on
population genetics: 287–321. Chepko-Sade, B. D. & Halpin,
Z. T. (Eds). Chicago: The University of Chicago Press.
Cypher, B. L., Warrick, G. D., Otten, M. R. M., O’Farrell, T. P.,
Berry, W. H., Harris, C. E., Kato, T. T., McCue, P. M., Scrivner,
J. H. & Zoellick, B. W. (In review). Population dynamics of San
Joaquin kit foxes at the Naval Petroleum Reserves in California.
Wildl. Monogr.
Egoscue, H. J. (1975). Dynamics of the kit fox in western Utah.
Bull. South. Calif. Acad. Sci. 74: 122–127.
Frankham, R. (1995). Effective population size/adult population
size ratios in wildlife: a review. Genet. Res. 66: 95–107.
Franklin, I. R. (1980). Evolutionary change in small populations.
In Conservation biology: an evolutionary–ecological perspec-
tive: 135–150. Soulé, M. E. & Wilcox, B. A. (Eds). Sunderland:
Sinauer Press.
Gilligan, D. M., Woodworth, L. M., Montgomery, M. E., Briscoe,
D. A. & Frankham, R. (1997). Is mutation accumulation a threat
to the survival of endangered species? Conserv. Biol. 11:
1235–1241.
Hill, W. G. (1972). Effective size of populations with overlap-
ping generations. Theoret. Pop. Biol. 3: 278–289.
Hill, W. G. (1979). A note on effective population size with over-
lapping generations. Genetics 92: 317–322.
Kimura, M. & Crow, J. F. (1963). The measurement of effective
population number. Evolution 17: 279–288.
Lande, R. (1995). Mutation and conservation. Conserv. Biol. 9:
782–791.
Lande, R. & Barrowclough, G. F. (1987). Effective population
size, genetic variation, and their use in population monitoring.
In Viable populations for conservation: 87–123. Soulé, M. E.
(Ed.). Cambridge: Cambridge University Press.
Latter, B. D. H. & Mulley, J. C. (1995). Genetic adaptation to
captivity and inbreeding depression in small laboratory popula-
tions of Drosophila melanogaster. Genetics 139: 255–266.
Lynch, M., Conery, J. & Bürger, R. (1995). Mutation accumula-
tion and the extinction of small populations. Am. Nat. 146:
489–518.
Mace, G. M. & Lande, R. (1991). Assessing extinction threats:
toward a reevaluation of IUCN threatened species categories.
Conserv. Biol. 5: 148–157.
Major, J. (1977). California climates in relation to vegetation. In
Terrestrial vegetation of California: 11–74. Barbour, M. G. &
Major, J. (Eds). New York: John Wiley and Sons.
McGrew, J. C. (1979). Vulpes macrotis. Mamm. Species 123: 1–6.
Mills, L. S. & Allendorf, F. W. (1996). The one-migrant-per-gen-
eration rule in conservation and management. Conserv. Biol. 10:
1509–1518.
Morrell, S. (1972). Life history of the San Joaquin kit fox. Calif.
Fish and Game 58: 162–174.
National Oceanic and Atmospheric Administration (1996). Local
climatological data, annual summary with comparative data,
Bakersfield, California. Report No. ISSN 0198–0696. Asheville:
US Department of Commerce, National Climatic Data Center.
Nunney, L. (1993). The influence of mating system and overlap-
ping generations on effective population size. Evolution 47:
1329–1341.
Nunney, L. (1996). The influence of variation in female fecun-
dity on effective population size. Biol. J. Linn. Soc. 59: 411–425.
Nunney, L. & Elam, D. R. (1994). Estimating the effective pop-
ulation size of conserved populations. Conserv. Biol. 8:
175–184.
Ralls, K. & White, P. J. (1995). Predation on San Joaquin kit
foxes by larger canids. J. Mammal. 76: 723–729.
Sawyer, J. O. & Keeler–Wolf, T. (1995). A manual of California
vegetation. Sacramento: California Native Plant Society.
Soulé, M. E. (1980). Thresholds for survival: maintaining fitness
and evolutionary potential. In Conservation biology: an evolu-
tionary–ecological perspective: 135–150. Soulé, M. E. &
Wilcox, B. A. (Eds). Sunderland: Sinauer Press.
Spiegel, L. K. & Disney, M. (1996). Mortality sources and sur-
vival rates of San Joaquin kit fox in oil-developed and unde-
veloped lands of southwestern Kern County, California. In
Studies of the San Joaquin kit fox in undeveloped and oil-devel-
oped areas: 71–91. Spiegel, L. K. (Ed.). Sacramento: California
Energy Commission.
Spiegel, L. K. & Tom, J. (1996). Reproduction of San Joaquin
kit fox in undeveloped and oil-developed habitats of Kern
County, California. In Studies of the San Joaquin kit fox in unde-
veloped and oil-developed areas: 53–69. Spiegel, L. K. (Ed.).
Sacramento: California Energy Commission.
US Department of the Interior (1983). San Joaquin kit fox recov-
ery plan. Portland: US. Fish and Wildlife Service.
Vucetich, J. A. & Waite, T. A. (In press). Number of censuses
required for demographic estimation of effective population size.
Conserv. Biol.
Vucetich, J. A., Waite, T. A. & Nunney, L. (1997). Fluctuating
population size and the ratio of effective to census population
size. Evolution 51: 2015–2019.
Waite, T. A. & Parker, P. G. (1996). Dimensionless life histories
and effective population size. Conserv. Biol. 10: 1456–1462.
White, P. J. & Ralls, K. (1993). Reproduction and spacing pat-
terns of kit foxes relative to changing prey availability. J. Wildl.
Mgmt. 57:861–867.
White, P. J., Vanderbilt White, C. A. & Ralls, K. (1996).
Functional and numerical responses of kit foxes to a short-term
decline in mammalian prey. J. Mammal. 77: 370–376.
Williams, D. F., Cypher, E. A., Kelly, P. A., Norvell, N., Johnson,
C. D., Colliver, G. C. & Miller, K. J. (1997). Recovery plan for
upland species of the San Joaquin Valley, California. Portland:
US Fish and Wildlife Service.
Wright, S. (1931). Evolution in Mendelian populations. Genetics
16: 97–159.
Wright, S. (1938). Size of population and breeding structure in
relation to evolution. Science 87: 430–431.
Zar, J. H. (1984). Biostatistical analysis. 2nd edn. Englewood
Cliffs: Prentice-Hall, Inc.