Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Ecology, 83(1), 2002, pp.

164–175
q 2002 by the Ecological Society of America

THE EFFECT OF HOST DENSITY ON ECTOPARASITE DISTRIBUTION:


AN EXAMPLE OF A RODENT PARASITIZED BY FLEAS
BORIS KRASNOV,1,3 IRINA KHOKHLOVA,2 AND GEORGY SHENBROT1
1Ramon Science Center, Jacob Blaustein Institute for Desert Research, Ben-Gurion University of the Negev,
P.O. Box 194, Mizpe Ramon, Israel
2Wyler Department of Desert Agriculture, Jacob Blaustein Institute for Desert Research,

Ben-Gurion University of the Negev, Beer Sheva, Israel

Abstract. The pattern of parasitism of the flea species Xenopsylla dipodilli and Nos-
opsyllus iranus theodori on the desert rodent species Gerbillus dasyurus was studied to
test the hypothesis that the relationships between flea abundance and host density conform
to pre-existing models of R. M. Anderson and R. M. May, with the correction that the
density of those host individuals that possess permanent burrows (residents) is substituted
for the overall host density. It was predicted that: (1) the intensity of flea infestation would
increase in curvilinear fashion with increase of host density to a plateau that would be
attained at a lower level of host density than would be expected from the basic model, and
(2) the prevalence of flea infestation plotted against host density would be hump-shaped.
The results indicated that intensity of flea infestation increased in either curvilinear fashion
to an asymptote (for X. dipodilli) or linearly (for N. i. theodori) with increase of host
density. As host density increased, the prevalence of infestation changed either unimodally
(X. dipodilli) or logarithmically (N. i. theodori). In addition, there was a positive relationship
between the mean number of fleas per host and the percentage of hosts infested. Both basic
and corrected models describing the relationships between flea burden and host density fit
the observational data well. However, simulations of the fraction of resident hosts dem-
onstrated that this parameter influences the relationship between host density and flea burden
only when residents comprise #50% of all host individuals.
Key words: ectoparasite; flea burden; Gerbillus dasyurus; host density; Nosopsyllus iranus theo-
dori; parasitism; rodent; Xenopsylla dipodilli.

INTRODUCTION the mean number of parasites per infested host (parasite


Unlike most free-living species, the geographic burden or intensity of infestation) would increase in a
range of parasites consists of a set of more or less curvilinear fashion to a plateau with increasing host
uniform inhabited ‘‘islands’’ or patches, represented by density (Dobson 1990). This is due to the increased
their host organisms, and the environment between probability that a parasite transmission stage would
these patches is absolutely unfavorable. The distribu- meet a host. However, only a few empirical studies of
tion of a parasite population across a host population this prediction have been conducted (Haukisalmi and
is characterized by their aggregation, or overdispersion. Hentonnen 1990, Arneberg et al. 1998). Another com-
Most individuals of the overdispersed parasite occur ponent of parasite abundance, namely, prevalence of
on a few host individuals, while most host individuals infestation (percentage of infested hosts), was not con-
have a only a few parasites or none at all (Anderson sidered in the framework of the model of Anderson and
and May 1978, Poulin 1993). This particular distri- May (1978). Nevertheless, the prevalence of infestation
bution of parasite individuals among hosts has impor- could be expected to increase with increasing host den-
tant consequences for different aspects of the evolu- sity. The argument for this can be the same as that for
tionary ecology of parasites. For example, overdisper- the intensity of infestation, namely, that under high host
sion can explain the density dependence of the intensity density there is an increased probability that the par-
of intraspecific competition (Shostak and Scott 1993) asite transmission stage will be transmitted. Another
and the biased sex ratio (Morand et al. 1993) in par- explanation follows metapopulation theory: parasites
infesting different host individuals are analogous to
asites.
free-living organisms inhabiting discrete patches, and
The basic models of Anderson and May (Anderson
the percentage of occupation of the latter increases with
and May 1978, May and Anderson 1978), implying the
the decrease of patch isolation (Thomas and Hanski
overdispersed distribution of parasites, predicted that
1997). This correlation between host density and prev-
alence of infestation has been supported by empirical
Manuscript received 17 April 2000; revised 20 December
2000; accepted 17 January 2001. studies (Arneberg et al. 1998).
3 Present address: Ramon Science Center, P.O. Box 194, Most studies related to parasite evolutionary ecology
Mizpe Ramon 80600 Israel. E-mail: krasnov@bgumail.bgu.ac.il and host–parasite relationships were made on endo-
164
January 2002 HOST DENSITY AND ECTOPARASITE DISTRIBUTION 165

PLATE 1. Wagner’s gerbil (Gerbillus dasyurus). Photo by G. Shenbrot.

parasites, mainly helminths. A host represents an ul- particularly in solitary species, often results in a surplus
timate habitat for an endoparasite, providing it with a of individuals that have no individual home ranges
place for living, foraging, and mating. Thus, hosts may (Brandt 1992, Gliwicz 1992 and references therein). In
be considered as habitat patches for most endopara- theory, these homeless individuals should not be pu-
sites. Unlike endoparasites, ectoparasites are influ- tative hosts for fleas, because they do not possess bur-
enced not only by characters of the host, but also by rows that are necessary for flea reproduction and de-
characters of their off-host environment, because their velopment of pre-imaginal stages. In turn, the density
contact with the host is usually intermittent. The range of resident hosts is determined by the carrying capacity
of off-host:on-host ratio of different ectoparasite taxa of the given habitat. For example, this carrying capacity
is very broad, varying from almost constant occurrence can be dependent on the number of available burrows
on the host lice (Anoplura), to sporadically attacking or places for burrowing, all else being equal. If the flea
host ticks (Ixodidae) and sand flies (Phlebotomidae). burden increases in a curvilinear fashion to a plateau
Fleas (Siphonaptera) take a median position in this with increasing host density, the breakpoint of this
range, alternating between periods when they occur on curve should be determined by the density of resident
the body of their host and periods when they occur in hosts rather than by the overall density of the hosts.
its burrow or nest. Fleas are permanent satellites of Consequently, the flea burden should decrease with the
higher vertebrates, being most abundant and diverse on proportion of resident individuals, under the same over-
small burrowing mammals. The female flea of some all abundance of the host and the same overall abun-
species oviposits while on the host, but the eggs typ- dance of parasites. The plateau of the flea burden curve
ically drop off into the nest or burrow (e.g., Pectin- should be lower than is expected from the basic model.
octenus pavlovskii, Leptopsylla segnis [Vasiliev 1961, If the resident hosts alone support the flea popula-
cited by Vatschenok 1988]). Other species mate and tion, the prevalence of infestation should increase with
oviposit both on-host and off-host (e.g., Xenopsylla the increase of host density, until the habitat becomes
cheopis). Nevertheless, in nearly all cases, larval and saturated with residents (e.g., all available burrows oc-
pupal development is entirely off-host. The larvae are cupied). Under higher host density, the prevalence of
not parasitic (except for a single species, Uropsylla flea infestation should decrease, since the resident ro-
tasmanica [see Vatschenok 1988]) and feed on debris dents would compose an ever-decreasing fraction of
and materials found in the burrow and/or nest of the the overall host population. Thus, the prevalence of
host. The larvae of the third instar spin a cocoon, cam- flea infestation plotted against host density is expected
ouflaging it by adhering particles of the substrate. to be hump-shaped.
Therefore, the habitat patch of a flea should be con- The small gerbilline rodent, Gerbillus dasyurus
sidered not only as a particular host, but rather as a (Wagner, 1842) (body mass is 20–24 g), occupies a
particular host with a particular burrow or nest in a variety of habitats in the Negev Highlands, Israel
particular habitat. This complex host–habitat–parasite (Shenbrot et al. 1997; see Plate 1). The density of G.
relationship can change the expectations derived from dasyurus varies across habitats, being highest in loess
the basic models. hills and lowest on gravel plains (Krasnov et al. 1996).
Indeed, the increase of density in rodent populations, The fraction of transient (nonresident) individuals
166 BORIS KRASNOV ET AL. Ecology, Vol. 83, No. 1

varies both among habitats and between years. This cirque is an area of about 200 km2 and is situated at
fraction reaches as high as 50% in high-density years the southern boundary of the Negev Highlands (308359
in loess hills (Gromov et al. 2000) and is even higher N, 348459 E). The diverse habitats represented in the
on gravel plains, where the majority of individuals are Ramon cirque, range from sandy dunes in the north-
transient (Khokhlova et al. 1994). Young individuals eastern parts of the cirque to sandstone rocks on the
of G. dasyurus have been observed mainly from April rims. The climate is characterized by hot, dry summers
to September, whereas the last pregnant females oc- (mean daily air temperature of July is 348C) and rel-
curred in October (Shenbrot et al. 1997). This indicates atively cold winters (mean daily temperature of January
that the spring and summer parts of breeding period is 12.58C). There is a sharp decrease in annual rainfall
were successful, while the pups of late autumn repro- from 100 mm on the north rim to 56 mm in the bottom
duction tended not to survive. In addition, the dispersal of the cirque.
of young gerbils occurred mainly in late spring and
summer (I. S. Khokhlova and B. R. Krasnov, unpub- Rodent trapping and flea collection
lished data). Two types of burrows of G. dasyurus have Rodents were trapped on seven 1-ha plots from sum-
been described (Shenbrot et al. 1997): (1) simple bur- mer 1992 until summer 1993 and on 16 1-ha plots from
rows of small depth, without nests, and (2) complex, summer 1994 until summer 1995. The first seven plots
deeper, burrows with a spherical nest made from dried were sampled every 2–3 mo. Other plots were sampled
plant material. It appears that resident individuals use every 6 mo. Plots were selected to represent the main
complex burrows, whereas transients have to be sat- substrate and vegetation gradients. Each plot was sam-
isfied with simple burrows, if any. We recorded fleas pled during 3–5 consecutive days, using 50 Sherman
in complex burrows of G. dasyurus, whereas no fleas live traps (Sherman Traps, Tallahasse, Florida, USA)
were found in simple burrows (B. R. Krasnov and G. placed in a grid at 5 3 5 stations, with two traps per
I. Shenbrot, unpublished data). Nevertheless, each bur- station and an interval of 20 m between stations. The
row of either type is occupied by one adult individual overall density of G. dasyurus was evaluated as the
(Shenbrot et al. 1997). minimal number of individuals known to be alive. An
G. dasyurus is parasitized mainly by two flea species, individual was considered resident if it was captured
Xenopsylla dipodilli Smit, 1960, and Nosopsyllus ir- either: (1) at least once in each of three consecutive
anus theodori Smit, 1960 (Krasnov et al. 1998, 1999). trapping periods, each 2–3 mo apart (in 1992–1993),
X. dipodilli occurs on G. dasyurus throughout the year, or (2) at least once in each of two consecutive trapping
although the abundance of this flea in winter is ex- periods, each 6 mo apart (in 1994–1995). This deter-
tremely low (Krasnov et al. 1997). In contrast with X. mination of resident vs. transient status can produce
dipodilli, N. i. theodori is a winter flea. No imago has some bias toward finding transients. In other words,
been found on G. dasyurus between April and Decem- we could erroneously attribute the transient status to
ber (Krasnov et al. 1997). Thus, these two fleas alter- resident gerbil (e.g., if it died by some unknown rea-
nate on G. dasyurus. These flea species occur also on sons during the study period) but could not erroneously
other rodents in the area (Krasnov et al. 1997, 1999).
attribute the resident status to transient gerbil. How-
However, extremely low abundances and sporadicity
ever, the individual composition of G. daysurus pop-
of these fleas on rodents other than G. dasyurus suggest
ulations has been shown to be very stable (Shenbrot et
that this has no any significant effect on the host–par-
al. 1997), suggesting that this potential bias could be
asite population dynamics.
considered negligible.
We hypothesized that the relationships between flea
We collected fleas from each individual rodent only
abundance and host density conform to the basic mod-
when it was captured the first time. The animal’s fur
els of Anderson and May (1978), with the following
was combed thoroughly, using a toothbrush over a
correction: flea abundance is influenced by the abun-
white plastic bath, and fleas were carefully collected.
dance of resident hosts, rather than overall host density.
We predicted that the intensity of flea infestation would Each rodent was sexed, weighed, marked by toe clip-
increase with an increase in host density in a curvilinear ping, and released. In total, 657 gerbils were captured,
fashion to a plateau that would be attained at a lower and 945 X. dipodilli and 320 N. i. theodori were col-
level of host density than expected from the basic mod- lected.
el. In addition, we predicted that the prevalence of flea The models
infestation plotted against host density would be hump-
shaped. To test these predictions, we studied the pattern We used the models presented by Anderson and May
of parasitism of the fleas X. dipodilli and N. i. theodori (1978), May and Anderson (1978), Grenfell (1992),
on G. dasyurus in the Negev desert. and Arneberg et al. (1998). These models describe the
dynamics of populations of host (H ) and parasite (P).
METHODS The main assumptions of the basic models are that: (1)
Study area host population growth is density dependent, given that
The study was conducted in the Ramon erosion there is no parasite-induced reduction of host repro-
cirque, Negev Highlands, Israel. The Ramon erosion duction, (2) hosts are long-lived, relative to their par-
January 2002 HOST DENSITY AND ECTOPARASITE DISTRIBUTION 167

TABLE 1. Description of the parameters of the basic model (adapted from Anderson and May
1978, Dobson 1990, Arneberg et al. 1998).

Parameter Description
a Instantaneous birth rate of host (per host per unit time)
b Instantaneous death rate of host due to causes other than parasite (per
host per unit time)
a Instantaneous host death rate due to parasite influence (per host per unit
time)
b Instantaneous rate of host reproduction reductions due to parasite influ-
ence (per host per unit time)
à Severity of density dependence in host population growth
l Instantaneous birth rate of parasite transmission stages (per parasite per
unit time) (the transmission stage for fleas is the adult imago)
m Instantaneous death rate of parasites due to either natural or host induced
(e.g., autogrooming) causes
H0 Transmission efficiency constant, varying inversely with the proportion of
parasite transmission stages that infest individuals of the host popula-
tion (see Anderson and May [1978] for further explanations)
k Parameter of the negative binomial distribution, varying inversely with
the degree of overdispersion; it measures the degree of aggregation of
the parasites within host population

asites, and (3) the frequency of parasites within hosts and transient host individuals. This transmission can
follows the negative binomial distribution. occur either when individuals visit alien burrows or
The basic model utilizes the following equations: through direct contact between host individuals (Kras-
nov and Knyazeva 1983). Consequently,
dH
5 (a 2 b 2 ÃH )H 2 (a 1 b)P (1)
dt dH
5 (a 2 ÃS)S 2 bH (3)
dt
dP lPH dP lPS (k 1 1)P 2
5 2 (b 1 ÃH 1 a 1 m)P 5 2 (b 1 ÃS 1 m)P 2 m . (4)
dt H0 1 H dt H0 1 H kH
The parasite burden (M ) is the mean number of par-
(k 1 1)P 2 asites per infested host, so the rate of changes in par-
2 (a 1 m) . (2)
kH asite burden is
The parameters of the model are presented in Table 1. dM dP /H
5
The model assumes that a parasite induces mortality dt dt
of a host (Anderson and May 1978, May and Anderson
lMS m(k 1 1) 2
1978). However, Arneberg et al. (1998) argued that the 5 2 (b 1 ÃS 1 m)M 2 M
positive relationship between host density and parasite H0 1 H k
abundance would also be expected when parasites have S S2
no pathogenic effects. Although the negative influence 2a M 1 Ã M 1 bM. (5)
H H
of flea parasitism on the fitness of a rodent host has
been reported for Synosternus cleopatrae on Gerbillus At equilibrium (dH/dt 5 dp/dt 5 dM/dt 5 0),
andersoni allenbyi (Lehmann 1992), no effect of flea
infestation on body conditions or on the survival of G.
dasyurus was found (Krasnov et al. 1997).
Thus, given that fleas neither induce host mortality
M5
[ lS
H0 1 H
S
2m2a 2 12
H
S
H
ÃS 1
k
m(k 1 1)
.2 ][ ]
(6)
nor reduce host reproduction, a 5 b 5 0 is assumed.
Further, we assumed that the dynamics of fleas is in- Eq. 6 was applied for X. dipodilli in spring and summer.
fluenced by the density of those host individuals that N. i. theodori occurs on its host during the period when
are resident and have permanent burrows (S ). This den- host does not reproduce (a 5 0). Consequently, this
sity is limited by the carrying capacity of the habitat, equation for N. i. theodori was transformed as
for example, by the number of available burrows. In
addition, only resident host individuals reproduce. Fur-
thermore, we assume that all host individuals are equal-
M5
[ lS
H0 1 H
2m 121
S
H
ÃS 2
k
m(k 1 1)][
.
] (7)

ly susceptible to parasitism by fleas, and thus, the trans- Hereafter, Eqs. 6 and 7 are referred to as ‘‘corrected’’
mission of fleas from individual to individual, or from models. They are collapsed into the equations of the
burrow to burrow, can be performed by both resident basic model, if S 5 H.
168 BORIS KRASNOV ET AL. Ecology, Vol. 83, No. 1

TABLE 2. Parameters of frequency distribution of X. dipo- to-mean ratio s2/X̄, and (3) standardized Morisita index
dilli and N. i. theodori on G. dasyurus
Mo (Krebs 1989) using the NEGBINOM program.
Negative We applied least-squares estimation procedures via
binomial the quasi-Newton algorithm for fitting the models of
Flea species N 2
s /X̄ Mo k x2 P the relationships between host density and flea burden.
X. dipodilli 945 7.25 0.53 0.39 29.05 0.27 In addition, we used regression analysis for the de-
N. i. theodori 320 3.79 0.51 0.51 18.9 0.45 scription of the relationships between both flea burden
Note: See Statistical processing for explanation of param- and prevalence of flea infestation, and both overall host
eters. density and density of resident hosts. Samples with G.
dasyurus density ,2 were excluded from the analyses
of prevalence of infestation. Model simulations for X.
We assigned values of the birth rate of G. dasyurus dipodilli were analyzed using a nonlinear estimation of
(a 5 2) based on our data both from field observations the breakpoint of piecewise linear regression. Those
(on average, two litters per breeding female per repro- for N. i. theodori were analyzed using linear regression.
duction period, 1.4:1 male:female ratio, 60% repro- All these calculations were performed with the statis-
ductively active adult females; Shenbrot et al. 1997) tical package STATISTICA (StatSoft, 1995).
and from laboratory breeding (on average, four pups/
litter; I. S. Khokhlova and B. R. Krasnov, unpublished RESULTS
data). No data are available on birth rate of transmis-
Flea distribution among host individuals
sion stages either of X. dipodilli or N. i. theodori.
Therefore, we used data on congeneric species, namely, Distributions of both flea species on G. dasyurus
Xenopsylla cheopis (Samarina et al. 1968), Xenopsylla were aggregated. The aggregation is indicated by the
conformis (Vatschenok 1988; B. R. Krasnov and I. S. values of the standardized Morisita index of dispersion,
Khokhlova, unpublished data), and Nosopsyllus con- the variance-to-mean ratio and k for negative binomial
similis (Alekseev 1961, Vatschenok 1988). We calcu- (Table 2). The negative binomial model successfully
lated the birth rate of flea transmission stages based on fit the observed frequency distribution of both X. di-
the rate of oviposition per female per breeding season, podilli and N. theodori on G. dasyurus in spring–sum-
given that survival of pre-imaginal stages is 80%. Thus, mer and winter, respectively (Table 2).
l was considered to be equal to 100 for X. dipodilli,
and 140 for N. i. theodori. Parameters H0, Ã, and m The effect of host density on flea burden
were estimated from the model-fitting procedures (see The burden of X. dipodilli (Mobs) on G. dasyurus
Statistical processing). increased in a curvilinear fashion to a plateau with
Analysis of the model concentrated on the variation increase in host density. This relationship could be de-
in flea burden and prevalence of infestation with chang- scribed significantly by logarithmic fit Mobs 5 0.4 1
es in the fraction of resident host individuals under the 0.73logH (F1,34 5 80.45, r2 5 0.69, P , 0.001). Change
same overall density. We simulated changes in this in host density was accompanied by a rapid change in
fraction from 5% to 100% with a 5% step. flea burden until the former attained ;20 individuals/
ha. Further increase in host density did not cause
Statistical processing change in flea burden. Both corrected and basic models
Whether the negative binomial can be used as a mod- described this relationship well (Table 3, Fig. 1), yield-
el for the distribution of X. dipodilli and N. i. theodori ing similar values for the parameters H0 and m.
on G. dasyurus was tested using the NEGBINOM com- Manipulation of the fraction of host residents while
puter software (Krebs 1989). The acceptance of the null holding all other parameters constant, produced a num-
hypothesis that the negative binomial distribution fits ber of curves (Fig. 2). It may be envisaged that the
the data was performed using a chi-square test. In ad- basic and corrected models differed from one another
dition, three measures of aggregation were calculated: when the fraction of residents was $50%. The break-
(1) k of the negative binomial distribution, (2) variance- point of the piecewise linear regressions of flea burden

TABLE 3. Parameters of the corrected and basic model when fitted to the observed burden of
X. dipodilli or N. i. theodori in relation to density of G. dasyurus.

Variance
explained
Flea Model (%) R H0 Ã m
X. dipodilli Basic 71.5 0.84 3.6 6 1.0 9.2 6 0.7
Corrected 69.7 0.81 3.4 6 1.8 24.9 6 2.6 8.7 6 1.2
N. i. theodori Basic 81.5 0.90 17.6 6 1.9 3.7 6 0.8
Corrected 83.0 0.91 19.1 6 1.2 29.6 6 1.8 3.1 6 0.7
Note: The à term is not present in the basic model (because S 5 H).
January 2002 HOST DENSITY AND ECTOPARASITE DISTRIBUTION 169

FIG. 1. Dependence of X. dipodilli burden on density of


G. dasyurus (observed, triangles and solid line; expected from FIG. 3. Values of the breakpoint of piecewise linear re-
corrected model, circles and dashed line; expected from basic gression (triangles) and the threshold of establishment (cir-
model, diamonds and bold dashed line). cles), calculated for simulation curves of X. dipodilli burden–
G. dasyurus density in dependence on the simulated per-
centage of resident host individuals.

against host density changed logarithmically with a dif-


ferent fraction of residents attaining a plateau when rected models differed from one another when the frac-
this fraction achieved 50–60% (F1,18 5 2247.5, r2 5 tion of residents exceeded 50%. The higher the resident
0.99, P , 0.001; Fig. 3). In addition, the described fraction, the faster was the rate of change in flea burden,
simulation demonstrated that the threshold of estab- indicated by the value of the slope of linear regression,
lishment (the minimal size of host population necessary until the fraction of residents attained 50–60% (F1,18
to sustain a parasite population) was higher when there 5 73.4, r2 5 0.80, P , 0.001; Fig. 6). As was the case
was a lower (5–30%) fraction of residents in a host for X. dipodilli, the threshold of N. i. theodori estab-
population (F1,18 5 38.2, r2 5 0.69, P , 0.001; Fig. lishment was greater with a lower fraction of residents
3). Furthermore, X. dipodilli burden decreased weakly, in a host population (F1,18 5 16.8, r2 5 0.77, P , 0.001;
albeit significantly, with an increase in the observed Fig. 6). The observed flea burden was correlated neg-
percentage of resident G. dasyurus (F1,34 5 19.8, r2 5 atively with the observed percentage of resident gerbils
0.35, P , 0.001). In contrast with X. dipodilli, the (F1,20 5 19.7, r2 5 0.49, P , 0.001).
burden of N. i. theodori increased linearly with an in-
crease in G. dasyurus density (F1,20 5 85.8, r2 5 0.8, The effect of host density on flea prevalence
P , 0.001). As was the case with X. dipodilli, both Prevalence of infestation by X. dipodilli changed un-
corrected and basic models fit this pattern well (Table imodally with an increase in overall density of G. das-
3, Fig. 4). yurus. The fraction of infested gerbils peaked at a den-
The results of the manipulation of the fraction of sity of 20 individuals/ha, whereas it was lower under
resident hosts while holding all other parameters con- both lower and higher gerbil density (Fig. 7). This re-
stant are presented in Fig. 5. Again, the basic and cor- lationship can be described by the equation PI 5 50.1

FIG. 2. Relationships between X. dipodilli burden and G. FIG. 4. Dependence of N. i. theodori burden on density
dasyurus density produced by manipulation of the fraction of of G. dasyurus (observed, triangles and solid line; expected
host residents from 5% to 100%, while holding all other pa- from corrected model, circles and dashed line; expected from
rameters constant. basic model, diamonds and bold dashed line).
170 BORIS KRASNOV ET AL. Ecology, Vol. 83, No. 1

FIG. 5. Relationships between G. dasyurus density and


N. i. theodori burden produced by manipulation of the fraction
of host residents from 5% to 100% while holding all other FIG. 7. Relationship between the density of G. dasyurus
parameters constant. and prevalence of infestation by X. dipodilli, calculated for
resident gerbils only (triangles and dashed line) and for both
resident and nonresident gerbils (circles and solid line).
1 3.15H 2 0.1H2, where PI is prevalence of infestation
and H is host density (F2,30 5 26.8, r2 5 0.62, P ,
25.41, r2 5 0.56, P , 0.001; Fig. 8). Although these
0.001). Prevalence of X. dipodilli infestation plotted
two regressions did not differ significantly ( t 5 0.45
against density of resident hosts was also hump-shaped,
for intercepts and t 5 0.29 for slopes), the density of
with a peak of prevalence at 12–13 individuals/ha (Fig.
resident hosts seemed to be a better predictor of prev-
7). The curve of prevalence against resident density
alence of infestation than overall host density (note
was steeper than the curve of prevalence against overall
higher r2 values). In contrast with X. dipodilli, no cor-
density, and fitted the equation PI 5 36.0 1 8.6S 2
relation was found between prevalence of infestation
0.36S2, where S is density of resident individuals (F2,30
by N. i. theodori and the fraction of resident individuals
5 20.9, r2 5 0.58, P , 0.001). Coefficients of these
in the host population (F1,20 5 0.76, P . 0.3).
two regressions differed significantly ( t 5 3.64 and t
5 3.95, P , 0.01). The prevalence of infestation re- Relationships between flea burden and
sponded weakly, albeit significantly, to the fraction of prevalence of infestation
resident individuals in the host population being higher The burden of X. dipodilli was not correlated with
with a greater percentage of residents (F1,31 5 9.6, r2 the prevalence of infestation when the former was cal-
5 0.23, P , 0.001). culated as the mean number of fleas per infested gerbil
Prevalence of infestation by N. i. theodori increased (F1,30 5 0.1, P . 0.9). However, when this parameter
logarithmically with an increase in host density (F1,20 was substituted with mean number of fleas across all
5 16.8, r2 5 0.46, P , 0.001; Fig. 8). The same was hosts, the relationship between flea abundance per host
true for the relationship between the prevalence of in-
festation and the density of resident hosts (F1,20 5

FIG. 6. Values of the slope of linear regression (triangles)


and threshold of establishment (circles) calculated for sim- FIG. 8. Relationship between the density of G. dasyurus
ulation curves of X. dipodilli burden–G. dasyurus density in and prevalence of infestation by N. i. theodori, calculated for
dependence on the simulated percentage of resident host in- resident gerbils only (triangles and dashed line) and for both
dividuals. resident and nonresident gerbils (circles and solid line).
January 2002 HOST DENSITY AND ECTOPARASITE DISTRIBUTION 171

and prevalence of infestation was significantly positive Flea burden and host density
(F1,30 5 21.11, r2 5 0.41, P , 0.001). Furthermore, the
Our field data did not allow us to successfully dis-
prevalence of infestation of resident gerbils increased
tinguish between the corrected and basic models of
significantly with the burden of X. dipodilli calculated
host–parasite dynamics. However, the manipulation of
for resident individuals only (F1,30 5 37.27, r2 5 0.55,
the fraction of host residents provided a set of curves
P , 0.001). In order to test if these relationships were
that permitted us to discriminate between these models.
sampling artifacts, the sampling model of Hanski et al.
A positive relationship between host density and par-
(1993) was applied to our observational data. This
asite burden was expected from the increase of parasite
model explained only 32% of the variance in the frac-
transmission rate under higher host densities (Anderson
tion of uninfested gerbils across sampling sites. This
and May 1978, May and Anderson 1978, Arneberg et
suggests that the increase of the prevalence of infes-
al. 1998). The greater the host density, the greater the
tation with the increase of flea burden cannot be con-
sidered as sampling artifact. probability that each parasite individual or respective
A positive relationship has been found between the transmission state (egg, larva, or imago) will contact
burden of N. i. theodori and the prevalence of infes- a host. Indeed, a positive correlation between parasite
tation when the burden was calculated both for infested burden and host density has been reported for both
gerbils and for all gerbils (F1,20 5 18.53, r2 5 0.48, P endoparasites (Haukisalmi and Hentonnen 1990) and
, 0.001, and F1,20 5 41.6, r2 5 0.68, P , 0.001 re- ectoparasites (Zhonglai and Yaoxing 1997, 1998), al-
spectively). The intensity of infestation of resident G. though other studies did not support this prediction
dasyurus was an even better predictor of the prevalence (Sorci et al. 1997).
of infestation (F1,20 5 52.79, r2 5 0.73, P , 0.001). The plateau attained by the curve of X. dipodilli bur-
The sampling model explained only 44% of variance den with an increase in host density can be explained
of the fraction of uninfested gerbils across sampling most easily by the limited carrying capacity of host
sites. individuals that can harbor a limited number of para-
sites only. However, this suggests a negative impact of
DISCUSSION parasites on hosts (Brown and Brown 1986, Lehmann
The present study of patterns of flea parasitism on 1993). No indication of an effect of fleas on gerbil body
G. dasyurus produced the following answers to ques- conditions and survival was found (Krasnov et al.
tions we set out to address. The intensity of flea in- 1997), and no effect was assumed in the present mod-
festation increases either in a curvilinear fashion to a els. Thus, the only possible explanation implicating
plateau (for X. dipodilli) or linearly (for N. i. theodori) host carrying capacity can be that this parasite thresh-
with an increase in host density. The difference be- old is very narrow, and that gerbil response to flea
tween overall host density and the density of resident number is stepwise rather than gradual. Flea abundance
hosts is important for the determination of parasite bur- below some threshold does not influence the health of
den only if residents comprise ,50% of all hosts. The the gerbil, whereas flea abundance above the threshold
percentage of infested hosts changes either unimodally is lethal. Consequently, the observed number of fleas
(X. dipodilli) or logarithmically (N. i. theodori) as host on every captured gerbil was below the threshold. Fur-
density increases. There was a positive relationship be- thermore, the relationship between N. i. theodori and
tween the mean number of fleas per host and the per- gerbil density was linearly positive, and there was no
centage of hosts infested. The answers were obtained plateau under high gerbil density. In addition, the mean
from the field data and from the analysis of a flea– number of fleas per infested rodent was higher for N.
rodent relationship model, incorporating among-host i. theodori than for X. dipodilli (mean 6 1 SE; 4.8 6
functional differences that are important for flea re- 0.3 fleas/rodent vs. 2.7 6 0.1 fleas/rodent). This result
production. suggests that the explanation of a plateau caused by
We recognize that in the real world, flea population host mortality when flea intensity exceeds a specific
dynamics will be affected, among other factors, by abi- threshold is unsatisfactory.
otic parameters (ambient temperature, relative humid- An alternative explanation implies the heterogeneity
ity, etc.) that may directly influence flea reproduction of hosts in relation to burrows as habitats for successful
and survival (especially during pre-imaginal develop- flea breeding, as well as the carrying capacity of host
ment; B. R. Krasnov and I. S. Khokhlova, unpublished habitat. Flea populations can be established and main-
data). Nevertheless, our omission of the abiotic con- tained on those host individuals that own burrows (res-
straints on the flea population dynamics is deliberate, idents), whereas homeless hosts can take part in flea
since our main aim is to provide quantitative insights transmission but are unable to sustain fleas. As overall
into the mechanisms by which hosts regulate parasite host density increases, the number of residents attains
populations. Therefore, we excluded the concept of en- a particular level determined by the carrying capacity
vironmental mediation of rodent–flea relationships to of host habitat. Further increases in host density result
simplify the manipulations. Such simplification eluci- in an increase in the number of nonresidents, while the
dates predictions of biological concern. number of residents remains stable. Because nonresi-
172 BORIS KRASNOV ET AL. Ecology, Vol. 83, No. 1

dent hosts do not participate in the siphonapteran life when residents comprise #50% of all host individuals.
cycle, the overall increase in host density after satu- This was true for both flea species. The only possible
ration of a habitat by residents has no effect on the flea role of nonresident hosts in flea dynamics is their effect
burden. This can also explain the absence of a plateau on flea transmission, given that they do not support flea
in the burden of N. i. theodori with a high density of reproduction. It appears that when the percentage of
gerbils. In contrast to X. dipodilli, N. i. theodori is a nonresidents is relatively low, they do not contribute
winter flea, reproducing in periods when neither re- heavily toward flea transmission, so their influence on
production nor dispersal of G. dasyurus occurs, so most flea dynamics and distribution is negligible (Khokhlova
host individuals in a given habitat are residents and Knyazeva 1983). Thus, the introduction of a pa-
(Khokhlova et al. 1994, Shenbrot et al. 1997). If so, rameter for nonresident density into the model does not
each habitat in each given period is saturated by res- produce any significant shift in comparison with the
idents, whereas transient homeless individuals are ab- basic model. However, when the fraction in the host
sent. Thus, the flea burden increases linearly with host population is relatively high, their contribution to flea
density. Indeed, the mean percentage of individuals transmission becomes significant. In such a case, a
considered to be nonresidents was 17.4 6 3.1% non- model that takes this nonresident component into ac-
resident individuals in spring–summer (when G. das- count, should describe the observational data better
yurus was parasitized by X. dipodilli) and 3.1 6 1.1% than that which considers only the overall host density.
nonresident individuals in winter (when G. dasyurus The role of resident hosts in flea dynamics is also sup-
was parasitized by N. i. theodori). ported by the fact that flea burden was correlated neg-
Finally, it is possible that an increase in flea number atively with the percentage of resident gerbils, showing
above some threshold either arouses an irritation of a that all flea individuals were distributed almost exclu-
gerbil’s skin and, consequently, excessive grooming sively among resident hosts and did not occur on non-
activity of a gerbil, or induces an immune response of residents. In addition, the manipulation of the fraction
a gerbil to flea salivary gland-derived molecules (Wikel of resident hosts demonstrated that the breakpoint of
1996). Grooming activity of a host has been shown to the flea burden vs. host-density curve is affected by
be one of the most significant host-induced mortality this fraction, thus conforming to our prediction, at least
factors for ectoparasites (Marshall 1981), and is pos- in relation to X. dipodilli. The plateaux of the simulated
itively correlated with the severity of infestation (Hin- curves of flea burden were lower with a lower fraction
kle et al. 1998). This is why the number of fleas on an of resident hosts under the same overall host density.
individual host does not exceed some threshold. Fur- This also fits well with the hypothesis about the role
thermore, in contrast to N. i. theodori, X. dipodilli has of resident hosts in flea distribution.
no ctenidia that permit the flea to anchor itself within It is noteworthy that the threshold of establishment
host fur, enabling it to resist the host’s grooming effort (the minimal size of a host population necessary to
(Humphries 1967, Traub 1972). This can partly explain sustain a parasite population) was affected by the frac-
why the mean number of fleas per infested rodent was tion of residents in the host population. The lower the
higher for N. i. theodori than for X. dipodilli. percentage of resident gerbils, the greater the number
Host immune responses can be also important in of them necessary to maintain flea reproduction. Again,
modulating ectoparasite numbers. It has been reported this is consistent with the assumption that only resi-
that different taxa of blood-feeding arthropods, includ- dents support flea populations. For example, if it is
ing fleas, stimulate a variety of host innate and specific surmised that the density of G. dasyurus is 2 individ-
acquired immune responses (Jones 1996, Wikel 1996). uals/ha, but both of them are residents, then this pro-
Consequently, host–ectoparasite associations are usu- duces two burrows for the reproduction of X. dipodilli
ally characterized by development of some level of with a threshold of establishment equaling two. If this
acquired resistance, which could potentially limit the is modified to assume that the density of G. dasyurus
level of infestation. However, it is unknown whether is 20 individuals/ha, but that only two are residents,
G. dasyurus is able to develop immune responses to this again produces two burrows for the reproduction
of X. dipodilli, but the threshold of establishment now
flea infestation. Also, a higher level of N. i. theodori
equals 20. The same relationship between the fraction
infestation in comparison with X. dipodilli infestation
of resident hosts and threshold of establishment was
seems to contradict the explanation implying host im-
produced by manipulations of the percentage of resi-
mune response, given the absence of the plateau in the
dents for N. i. theodori. However, this pattern can be
former case and the presence of the plateau in the latter
considered as a theoretical exercise only, because non-
case.
resident individuals comprise a very small proportion
As mentioned previously, both basic and corrected
of overall gerbil population in the period when N. i.
models describing the relationships between flea bur-
theodori occurs.
den and host density fitted the observation data well.
However, the simulations of the fraction of residents Flea prevalence and host density
demonstrated that this parameter influences the rela- Theoretically, the relationship between host density
tionship between host density and flea burden only and parasite prevalence is expected to be positive (at-
January 2002 HOST DENSITY AND ECTOPARASITE DISTRIBUTION 173

taining a plateau under high host density), given that above-mentioned discussion, namely that there is an
there are no parasite extinction events across host in- increased probability that the transmission stage is
dividuals. The explanation is that, once host density transmitted under high host density. The more distantly
becomes high enough for all parasites to find a host, a located the patches are (i.e., density is low), the less
further increase in host density would be inconsequen- frequently they are occupied, due to greater difficulty
tial. However, field data on the relationship between of colonization (Hanski 1994). In the case of the flea
flea prevalence and host density are contradictory. For N. i. theodori, the number of habitat patches (hosts) at
example, the increase and stabilization of flea preva- each given location does not increase, so the absence
lence around 100% with an increase in host density of an unimodal curve of prevalence plotted against host
has been reported for Spermophilus richardsoni para- density is to be expected.
sitized by Opisocrostis bruneri and Oropsylla rupestris
(Lindsay and Galloway 1997), whereas prevalence of Flea burden and flea prevalence
Xenopsylla bantorum increased with the decrease in The relationship between flea burden and flea prev-
density of its host, Arvicanthis niloticus (Schwan alence is analogous to the intraspecific abundance–oc-
1986). Conversely, prevalence cannot be lower than a cupancy relationship of free-living animals. The ex-
specific eradication threshold (sensu Nee et al. 1997), istence of positive correlations between local density
because a minimum number of hosts is required for the and regional or geographic occupation have been re-
parasite population to persist, analogous to a minimum ported for a variety of taxa (Gaston et al. 1997, Gaston
number of patches for free-living animals (Nee 1994, 1999 and references therein). Contrary to studies of
Kareiva and Wennergren 1995, Hanski et al. 1996). interspecific abundance–occupancy relationships, abun-
The prediction about the unimodal response of flea dance in the studies of intraspecific abundance–occu-
prevalence to an increase in host density appeared to pancy relationships is usually averaged across all (oc-
be true for X. dipodilli, but not for N. i. theodori. If cupied and unoccupied) sites (e.g., Kuno 1991 and ref-
only resident host individuals support the flea popu- erences therein). These relationships have been found
lation, the prevalence of infestation increases with an to be positive, although contrary cases also have been
increase in host density until all resident individuals reported (Boecken and Shachak 1998). In general, an
are infested and the habitat is saturated with residents empirical relationship between the fraction of sites (or
(all available burrows are occupied or individual home samples) where a given species occurs and its mean
ranges collapsed to minimal possible size). Under high- density can be modeled as a logistic curve (Gaston
er host density, the prevalence of flea infestation de- 1999).
creases, because the resident and infested gerbils com- Intraspecific studies of parasite burden (equivalent
pose an ever-decreasing fraction of the overall popu- of abundance) vs. parasite prevalence (equivalent of
lation. Conversely, for the above-mentioned reasons, percentage occupancy or Diamond’s [1975] incidence
flea prevalence plotted against resident host density is of a species) of a parasite species across individual
expected to increase, reaching a plateau. This was not hosts are related to relatively small scale (see Gaston
the case; the curve of prevalence against resident den- 1999). Thus, the model can be collapsed to its linear
sity was also hump-shaped, although steeper than the version. Positive relationships have been repeatedly re-
curve of prevalence against overall density. An expla- ported for helminth burden and prevalence (e.g., Bush
nation for this result might be that flea reproduction and Holmes 1986, Haukisalmi and Hentonnen 1994).
and transmission have lower rates in comparison with Analogous data for fleas have not been found.
reproduction and dispersal of gerbils. In other words, The present results demonstrated that the burden of
the rate of establishment of new patches (previously both fleas correlated positively with the prevalence of
vacant or new burrows that are being occupied either infestation, when it was calculated as the mean number
by dispersing young gerbils or by transient adults) is of fleas across all host individuals. These positive re-
faster than the rate of their infestation. Consequently, lationships could be sampling artifacts; the number of
it is probable that, under high host density, a fraction patches at which a species is recorded (prevalence)
of resident hosts remains underused by the fleas, merely could be a positive function of the mean abundance of
because they cannot keep pace with gerbil reproduction the species (burden) simply because locally rare species
and dispersal. Indeed, the active breeding period of flea are more difficult to detect than are locally abundant
Xenopsylla hirtipes has been reported to fall behind species. However, this was not the case. The sampling
active breeding and dispersal of its main host, Rhom- model of Hanski et al. (1993) explained a relatively
bomys opimus, for 2–3 wk (Kiriakova et al. 1970). In low proportion of variance of the fraction of uninfested
addition, species of Xenopsylla have only 1–3 gener- gerbils across sampling sites, in spite of the small scale
ations/yr (Soldatkin et al. 1967 for X. gerbilli caspica, of present considerations. The explanation for the pos-
and Zolotova et al. 1978 for X. skrjabini), so the above itive relationship between flea burden and prevalence
scenario is feasible. may be straightforward. The probability that a flea in-
The positive relationship between prevalence and fests a new host individual (patch) is higher when the
host density for N. i. theodori can be explained by the flea burden is higher. Flea colonization of new host
174 BORIS KRASNOV ET AL. Ecology, Vol. 83, No. 1

individuals is passive rather than active, and occurs Gliwicz, J. 1992. Patterns of dispersal in non-cyclic popu-
when host individuals come into direct contact with lations of small rodents. Pages 147–159 in N. C. Stenseth
and W. Z. Lidicker, editors. Animal dispersal. Small mam-
one another or visit each other’s burrows. It is clear mals as a model. Chapman and Hall, London, UK.
that the higher the flea abundance on a flea donor, the Grenfell, B. T. 1992. Parasitism and the dynamics of ungulate
higher the probability of infestation in a flea recipient. grazing systems. American Naturalist 139:907–929.
Moreover, fleas can infest nonresident hosts, but these Gromov, V. S., B. R. Krasnov, and G. I. Shenbrot. 2000.
Space use in Wagner’s gerbil Gerbillus dasyurus (Wagner,
hosts are not able to support flea populations. This
1842) in the Negev Highlands, Israel. Acta Theriologica
explains why the intensity of infestation of resident 45:175–182.
gerbils was found to be the best predictor of the prev- Hanski, I. 1994. Patch-occupancy dynamics in fragmented
alence of infestation. These results once more accen- landscapes. Trends in Ecology and Evolution 9:131–135.
tuate the effect of the density of resident host individ- Hanski, I., J. Kouki, and A. Halkka. 1993. Three explanations
of the positive relationship between distribution and abun-
uals on flea distribution. dance of species. Pages 108–116 in R. E. Ricklefs and D.
ACKNOWLEDGMENTS Schluter, editors. Species diversity in ecological commu-
nities. Historical and geographical perspectives. The Uni-
We thank M. Hastriter (Brigham Young University, Utah, versity of Chicago Press, Chicago, Illinois, USA.
USA), S. Medvedev and V. Vatschenok (both from Zoological Hanski, I., A. Moilanen, and M. Gyllenberg. 1996. Minimum
Institute RAS, St. Petersburg, Russia) for their help in flea viable metapopulation size. American Naturalist 147:527–
identification. David Ward (Ben-Gurion University of the Ne- 541.
gev, Israel) and Neil Springate (Museum of Natural History, Haukisalmi, V., and H. Hentonnen. 1990. The impact of cli-
Great Britain) read an earlier version of the manuscript and matic factors and host density on the long-term population
made helpful comments. We also thank J. S. Brown (Uni- dynamics of vole helminths. Oecologia 83:309–315.
versity of Illinois at Chicago, USA) and two anonymous ref- Haukisalmi, V., and H. Hentonnen. 1994. Distribution pat-
erees for their comments. Financial support during this study terns and microhabitat segregation in gastrointestinal hel-
was provided by Israel Ministry of Science, Israel Ministry minths of Sorex shrews. Oecologia 97:236–242.
for New Immigrant Absorption and Local Council of Mizpe Hinkle, N. C., P. G. Koehler, and R. S. Patterson. 1998. Host
Ramon. This is publication no. 102 of the Ramon Science grooming efficiency for regulation of cat flea (Siphonap-
Center. tera: Pulicidae) populations. Journal of Medical Entomol-
LITERATURE CITED ogy 35:266–269.
Humphries, D. A. 1967. Function of combs in ectoparasites.
Alekseev, A. N. 1961. Biology of the flea Ceratophyllus
Nature 215:319.
(Nosopsyllus) consimilis Wagn., 1898 (Ceratophyllidae,
Jones, C. J. 1996. Immune responses to fleas, bugs and suck-
Aphaniptera). [In Russian.] Zoologicheskii Zhurnal 40:
ing lice. Pages 150–174 in S. K. Wikel, editor. The im-
1840–1847.
munology of host-ectoparasitic arthropod relationships.
Anderson, R. M., and R. M. May. 1978. Regulation and
CAB International, Wallingford, UK.
stability of host-parasite population interactions. I. Regu-
Kareiva, P., and U. Wennergren. 1995. Connecting landscape
latory processes. Journal of Animal Ecology 47:219–247.
patterns to ecosystem and population processes. Nature
Arneberg, P., A. Skorping, B. Grenfell, and A. F. Read. 1998.
373:299–302.
Host densities as determinants of abundance in parasite
communities. Proceedings of the Royal Society of London Khokhlova, I. S., and T. V. Knyazeva. 1983. The effect of
B 265:1283–1289. spatial and social structure of house mice populations on
Boecken, B., and M. Shachak. 1998. The dynamics of abun- the flea assemblages. Pages 165–167 in I. F. Taran, editor.
dance and incidence of annual plant species richness during Prophylactics of diseases in the Natural Nidi. [In Russian.]
colonization in a desert. Ecography 21:63–73. Stavropol Anti-Plague Institute, Stavropol, Russia.
Brandt, C. A. 1992. Social factors in immigration and emi- Khokhlova, I. S., B. R. Krasnov, G. I. Shenbrot, and A. A.
gration. Pages 96–141 in N. C. Stenseth and W. Z. Lidicker, Degen. 1994. Seasonal body mass changes and habitat dis-
editors. Animal dispersal. Small mammals as a model. tribution in several rodent species from Ramon erosion
Chapman and Hall, London, UK. cirque, Negev Highlands, Israel. [In Russian.] Zoologi-
Brown, C. R., and M. B. Brown. 1986. Ectoparasitism as a cheskii Zhurnal 73:115–121.
cost of coloniality in cliff swallows (Hirundo rustica). Kiriakova, A. N., L. A. Koptzev, and Z. G. Koptzeva. 1970.
Ecology 67:1206–1218. Number of generations in fleas of the genus Xenopsylla
Bush, A. O., and J. C. Holmes. 1986. Intestinal helminths of within a year in Norther Kyzylkum. [In Russian.] Parazi-
lesser scaup ducks: patterns of association. Canadian Jour- tologiya 4:528–536.
nal of Zoology 64:132–141. Krasnov, B. R., M. Hastriter, S. G. Medvedev, G. I. Shenbrot,
Diamond, J. M. 1975. Assembly of species communities. I. S. Khokhlova, and V. S. Vaschenok. 1999. Additional
Pages 342–444 in M. L. Cody and J. M. Diamond, editors. records of fleas (Siphonaptera) on wild rodents in the south-
Ecology and evolution of communities. Harvard University ern part of Israel. Israel Journal of Zoology 45:333–340.
Press, Cambridge, Massachusetts, USA. Krasnov, B. R., and T. V. Knyaseva. 1983. Ectoparasite ex-
Dobson, A. P. 1990. Models for multi-species parasite–host change between the midday jird, Meriones meridianus, and
communities. Pages 261–288 in G. Esch, A. O. Bush, and the house mouse, Mus musculus, under experiment. Pages
J. M. Aho, editors. Parasite communities: patterns and pro- 243–244 in I. F. Taran, editor. Prophylactics of diseases in
cesses. Chapman and Hall, London, UK. the Natural Nidi. [In Russian.] Stavropol Anti-Plague In-
Gaston, K. J. 1999. Implications of interspecific and intra- stitute Publications, Stavropol, Russia.
specific abundance–occupancy relationships. Oikos 86: Krasnov, B. R., G. I. Shenbrot, I. S. Khokhlova, and E. Yu.
195–207. Ivanitskaya. 1996. Spatial structure of rodent community
Gaston, K. J., T. M. Blackburn, and J. H. Lawton. 1997. in the Ramon erosion cirque, Negev Highlands (Israel).
Interspecific abundance–range size relationships: an ap- Journal of Arid Environments 32:319–327.
praisal of mechanisms. Journal of Animal Ecology 66:579– Krasnov, B. R., G. I. Shenbrot, S. G. Medvedev, I. S. Khokh-
601. lova, and V. S. Vatschenok. 1998. Habitat-dependence of
January 2002 HOST DENSITY AND ECTOPARASITE DISTRIBUTION 175

a parasite–host relationship: flea assemblages in two gerbil aptera) on grassland rodents in Lake Nakuru National Park,
species of the Negev desert. Journal of Medical Entomol- Kenya, and potential for plague transmission. Bulletin of
ogy 35:303–313. Entomological Research 76:633–648.
Krasnov, B. R., G. I. Shenbrot, S. G. Medvedev, V. S. Vat- Shenbrot, G. I., B. R. Krasnov, and I. S. Khokhlova. 1997.
schenok, and I. S. Khokhlova. 1997. Host-habitat relation On the biology of Wagner’s gerbil Gerbillus dasyurus
as an important determinant of spatial distribution of flea (Wagner, 1842) (Rodentia: Gerbillidae) in the Negev High-
assemblages (Siphonaptera) on rodents in the Negev desert. lands, Israel. Mammalia 61:467–486.
Parasitology 114:159–173. Shostak, A. W., and M. E. Scott. 1993. Detection of density-
Krebs, C. J. 1989. Ecological metodology. Harper Collins, dependent growth and fecundity of helminths in natural
New York, New York, USA. infections. Parasitology 76:790–795.
Kuno, E. 1991. Sampling and analysis of insect populations. Soldatkin, I. S., Yu. V. Rudenchik, E. A. Severova, Z. I.
Annual Review of Entomology 36:285–304. Klimova, and N. A. Mokrievich. 1967. Peculiarities of the
Lehmann, T. 1992. Ectoparasite impacts on Gerbillus an- ecology of the summer population of Xenopsylla gerbilli
dersoni allenbyi under natural conditions. Parasitology 104: caspica L. [In Russian.] Zoologicheskii Zhurnal 46:909–
479–488. 914.
Lehmann, T. 1993. Ectoparasites: direct impact on host fit- Sorci, G., M. de Fraipoint, and J. Clobert. 1997. Host density
ness. Parasitology Today 9:8–13. and ectoparasite avoidance in the common lizard (Lacerta
Lindsay, L. R., and T. D. Galloway. 1997. Seasonal activity vivipara). Oecologia 111:183–188.
and temporal separation of four species of fleas (Insecta: StatSoft. 1995. STATISTICA for Windows. Version 5.1.
Siphonaptera) infesting Richardson’s ground squirrels, StatSoft, Tulsa, Oklahoma, USA.
Spermophilus richardsoni (Rodentia: Sciuridae), in Mani- Thomas, C. D., and I. A. Hanski. 1997. Butterfly metapopula-
tions. Pages 359–386 in I. A. Hanski and M. E. Gilpin,
toba, Canada. Canadian Journal of Zoology 75:1310–1322.
editors. Metapopulation biology. Ecology, genetics, and
Marshall, A. G. 1981. The ecology of ectoparasite insects.
evolution. Academic Press, San Diego, California, USA.
Academic Press, London, UK.
Traub, R. 1972. The relationship between the spines, combs
May, R. M., and R. M. Anderson. 1978. Regulation and
and other skeletal features of fleas (Siphonaptera) and the
stability of host–parasite population interactions. II. De- vestiture, affinities and habits of their hosts. Journal of
stabilizing processes. Journal of Animal Ecology 47:455– Medical Entomology 9:601.
461. Vatschenok, V. S. 1988. Fleas: vectors of pathogens causing
Morand, S., J.-P. Pointier, G. Borel, and A. Theron. 1993. diseases in humans and animals. [In Russian.] Nauka, Len-
Pairing probability of schistosomes related to their distri- ingrad, USSR.
bution among the host population. Ecology 74:2444–2449. Wikel, S. K. 1996. Host immunity to ticks. Annual Review
Nee, S. 1994. How populations persist. Nature 367:123–124. of Entomology 41:1–22.
Nee, S., M. P. Hassel, and R. M. May. 1997. Two-species Zhonglai, L., and Z. Yaoxing. 1997. Analysis on the yearly
metapopulation models. Pages 123–147 in I. A. Hanski and dynamics relation between body flea index and population
M. E. Gilpin, editors. Metapopulation biology. Ecology, of Citellus dauricus. [In Chinese.] Acta Entomologica Sin-
genetics, and evolution. Academic Press, San Diego, Cal- ica 40:166–170.
ifornia, USA. Zhonglai, L., and Z. Yaoxing. 1998. The yearly dynamics
Poulin, R. 1993. The disparity between observed and uniform relationship between burrow nest flea index and population
distibutions: a new look at parasite aggregation. Interna- of Citellus dauricus. [In Chinese.] Acta Entomologica Sin-
tional Journal for Parasitology 23:937–944. ica 41:77–81.
Samarina, G. P., A. N. Alekseev, and P. I. Shiranovich. 1968. Zolotova, S. I., Z. P. Maslennikova, and O. V. Afanasieva.
Study of fertility of rat fleas (Xenopsylla cheopis Rothsch. 1978. Number of generations of fleas of the genus Xen-
and Ceratophyllus fasciatus Bosc.) feeding on different host opsylla (Siphonaptera, Pulicidae) in the subzone of north-
species. Zoologicheskii Zhurnal 47:261–268. [In Russian.] ern deserts. [In Russian.] Zoologicheskii Zhurnal 57:1663–
Schwan, T. G. 1986. Seasonal abundance of fleas (Siphon- 1667.

You might also like