Behavioral Ecology Vol. 12 No. 5: 569-576
© 2001 International Society for Behavioral Ecology
Superfluous killing in spiders: a consequence of adaptation to food-limited environments?
Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996-1610, USA
Address correspondence to S.E. Riechert, Department of Ecology and Evolutionary Biology, 569 Dabney Hall, Knoxville, TN 37996-1610, USA. E-mail: sriecher{at}utk.edu . J.L. Maupin is currently at the Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USA.
Received 13 March 2000; revised 13 November 2000; accepted 13 November 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The hypothesis that superfluous killing, partial consumption, and abandonment of prey is a consequence of adaptation to food-limited environments was tested in two feeding trials on a desert spider, Agelenopsis aperta. First, we made comparisons among populations inhabiting sites of high prey (HP) or low prey (LP) availability that differed in their degree of genetic isolation. Typically, A. aperta entirely consumed one or two of the prey items it captured in a feeding bout. Additional prey were partially consumed or abandoned without eating. Spiders from the genetically isolated HP population, however, captured fewer prey and showed a higher incidence of full feeding on prey than did individuals from the other populations. Only one spider from this population captured a prey item that it failed to feed on, whereas spiders from LP populations failed to feed on high numbers of captured prey. The greatest variability in feeding behavior was exhibited in the HP population that experienced gene flow. The second test was based on the finding that aggressiveness is largely a sex-linked trait in A. aperta: the aggressiveness of the female parent only is inherited by male offspring, whereas both parents contribute to this trait in female offspring. All female F1 hybrids between LP and HP parental types exhibited high levels of superfluous killing, as did male F1 hybrids derived from LP females. F1 hybrid males derived from HP females exhibited extremely low levels of superfluous killing. Superfluous killing thus has its basis in the genetic control of levels of aggression.
Key words: Agelenopsis aperta, behavioral variation, genetic determination, predation, predator populations, spiders, superfluous killing.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Geographically isolated populations often have evolved into distinct ecotypes that differ in morphological, life history, or behavioral traits (Krebs and Davies, 1993
;
Mayr, 1970;
Riechert, 1999). Such
variation may reflect experiential conditions, but it may also have a
substantial genetic basis. Genetic differences in behavioral traits between
populations have been documented in many organisms (see, e.g., reviews in
Foster and Endler, 1999).
We investigated the adaptive significance of superfluous killing in
spiders. This term refers to the fact that animals may only partially consume
or abandon intact prey they have captured
(Conover, 1966). Superfluous
killing has been reported for a diverse group of animals, including
zooplankton (Conover, 1966),
stoats and weasels (Erhlinge et al., 1974a,b;
Oksanen and Oksanen, 1981),
damselfly naiads (Johnson et al.,
1975), wolves (Bjarvall and
Nilsson, 1976; Miller et al.,
1985), predaceous mites (Metz
et al., 1988), and spiders
(Riechert and Maupin, 1998;
Samu and
Bíró,
1993; Smith and Wellington,
1986).
In delineating the functional response curve of the orb-web spider
Araneus diadematus L. (Araneidae), Smith and Wellington
(1986) observed abandonment of
intact prey in some trials. This and other reports of potential surplus
killing by spiders led Samu and
Bíró
(1993) to investigate the
behavior in a wandering spider, Pardosa hortensis Thorell
(Lycosidae). This wolf spider exhibited significant levels of both partial
feeding and prey abandonment at high rates of encounter with prey. Riechert
and Maupin (1998) also
observed both components of superfluous killing in five web-building species
with different web structures: Argiope trifasciata (Forskal;
Araneidae), which has a chemically sticky orb web; Dictyna volucripes
Keyserling (Dictynidae), which has a mechanically sticky hackled-band web;
Achaearanea tepidariorum (C. L. Koch; Theridiidae), which has a
scaffold web; Florinda coccinea (Linyphiidae), which has a sheet-line
web; and Agelenopsis aperta (Gertsch; Agelenidae), which has a funnel
web.
Our study of the desert spider Agelenopsis aperta tested the possibility that superfluous killing is a consequence of selection for aggressiveness toward prey in food-limited environments. Two approaches were taken: (1) comparisons of the trait in populations occupying different selective environments and experiencing different levels of genetic isolation and (2) a genetic breeding experiment involving population crosses. We completed feeding trials on individuals from four populations of A. aperta, comprising a set of different selection and genetic mixing conditions. Two of the populations were derived from food-limited environments and two from food-rich environments. Further, one population of each feeding regime was genetically isolated, while the other population experienced at least some gene flow from a population having a different feeding environment.
We also completed a breeding experiment that explicitly tested the extent
to which superfluous killing is genetically determined. This test was based on
our growing understanding of the genetic bases of fitness-linked behavioral
traits in the A. aperta system (reviewed in
Riechert, 1999). Briefly,
individual aggressiveness on a continuum from low to high aggression varies
among populations occupying different feeding and competitive environments.
Level of aggressiveness in this spider underlies the territory size demanded
by individual A. aperta, as well as their aggressiveness toward prey,
boldness toward predators, and agonistic or fighting behavior. These traits
are phenotypically correlated and thus appear to be pleiotropic effects of the
same genes (Riechert and Hedrick,
1993). Maynard Smith and Riechert
(1984) developed a genetic
model that explained level of aggressiveness in terms of the classic
ethological construct of conflicting tendencies: two scalars (i.e., tendency
to flee and tendency to attack) operate antagonistically in determining the
level of aggression an individual exhibits in particular contexts. In crossing
individuals from two populations occupying different environments (high
competition—low predation risk vs. low competition—high predation
risk) with corresponding behavioral differences, it was possible to unravel
the genetic system underlying the fitness-linked traits listed above. The
tendency-to-attack component makes a greater contribution to an individual's
aggressiveness than the tendency-to-flee component. Further, the
tendency-to-attack component is sex linked (inherited on the sex chromosomes),
whereas the tendency-to-flee component is a quantitative autosomal trait.
Because male A. aperta have only one copy of each of two X sex
chromosomes (X1- and X2-), while females have two copies
of each sex chromosome (X1X1, X2X2), males inherit the tendency-to-attack
level of the female parent. Female A. aperta, in contrast, inherit
tendency to attack from both parents. This relationship formed the basis of
our breeding experiment. If superfluous killing has an underlying genetic
basis, males should exhibit the behavior appropriate to the population to
which the female parent belongs in reciprocal, between-population crosses. The
female offspring serve as the control in this case, as they are not expected
to show a directional effect in their superfluous killing scores.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study animals
Agelenopsis aperta is an arid-lands spider that occupies a variety of habitats in the southwestern United States. Its web is composed of a nonsticky sheet, on which prey capture and agonistic interactions occur, and a funnel retreat that leads into a protected area, such as a crevice, or under a rock or stump. A foraging spider sits just inside the funnel retreat facing the web surface to facilitate the location and capture of prey. As prey encounter with the web is detected, the spider runs out of the funnel and actively captures the prey by biting it and wrapping it in silk (Riechert and Maupin, 1998
).
When not foraging, A. aperta retreats into the web funnel to seek
protection from predators, heat, wind, and so on. Extensive work by Riechert
and colleagues over the past 20 years has led to a wealth of information
regarding the ecology and behavior of this species as well as the role of
population ecology in behavioral differentiation (see reviews in Riechert,
1993a,
1999).
Population comparison
The four populations used in this study were:
- a riparian population, high prey (HP), genetically isolated (GI), Blanco
County, south-central Texas;
- a riparian population, HP, with significant levels of gene flow (GF) from
more arid habitats, Cochise County, southeastern Arizona;
- a desert grassland population, low prey (LP), genetically isolated (GI),
Lincoln County, south-central New Mexico;
- an evergreen dry-woodland population, LP, limited GF from a riparian
habitat, Cochise County, southeastern Arizona.
The two riparian populations occupy mesic habitats where prey is abundant
and favorable thermal conditions allow for extended foraging periods
(Riechert, 1993c). Populations
3 and 4, the desert grassland and dry woodland populations, occupy more arid
habitats where encounter with prey is limited because prey numbers are low and
environmental conditions restrict time spent foraging
(Riechert, 1993b). Gene flow
occurs between the arid woodland and more mesic riparian populations in
southeastern Arizona (Riechert,
1993b), but is not a factor in the arid grassland (New Mexico
site) and mesic riparian (Texas site) populations. This combination of
environments and genetic influences allows us to examine the possible effects
of local adaptation and gene flow on the superfluous killing behavior
evidenced by A. aperta (Riechert
and Maupin, 1998).
Study areas
The species range of A. aperta extends from central Texas to
California and from northern Wyoming south to southern Mexico. The HP/GI
population inhabits a riparian area along the Pedernales River at the eastern
edge of the species range in south central Texas. Because the riparian habitat
in south-central Texas is large and is surrounded by similar habitats, it is
not subject to gene flow from more arid areas. The site is characterized by an
abundance of prey (mean dry weight = 115.7 mg per day) associated with its
proximity to water, relatively high levels of precipitation (80-100 cm a
year), and a complex vegetation structure
(Riechert, 1993c). The forest
floor is covered with leaves and grass, and the canopy consists of cottonwood
(Populis deltoides), willow (Salix nigra), and pecan
(Carya illinoiensis). Oak (Quercus fusiformis) and juniper
(Juniperus ashei) dominate areas that are not immediately adjacent to
the river. An abundance of canopy cover provides favorable habitat for birds,
which often feed upon A. aperta in this area
(Riechert, 1993c). HP/GI
spiders demonstrate predatory, antipredatory, and territorial behaviors
appropriate to a consistent and abundant supply of food and to high predation
pressure by birds (Riechert,
1993c).
The HP/GF population occupies a strip of riparian habitat bordering either
side of a stream on the eastern slope of the Chiricahua Mountains of Cochise
County, Arizona. The permanent spring-fed stream supports a tree canopy of box
elder (Acer negundo), juniper (Juniperus osteosperma), and
sycamore (Platanus wrightii). Environmental conditions experienced by
spiders in this habitat closely resemble those of the riparian population in
our Texas study area. These spiders, too, have an unlimited food supply and
are protected by a tree canopy from harsh abiotic factors such as the
temperature extremes and strong winds characteristic of arid areas (Riechert,
1979,
1993b,c).
They also are subjected to predation by avian predators. However, the riparian
spiders from this site in Arizona are bolder toward predators, more aggressive
on average in prey capture and territory defense, and maintain larger
territories than evolutionarily stable strategy and optimal foraging models
would predict (Hammerstein and Riechert,
1988; Riechert,
1991). Behavioral variation within this population is also greater
than would be expected for a population at adaptive equilibrium, suggesting
that some factor is facilitating the maintenance of aberrant behavior.
Riechert (1993b) identified
these deviations from expectation as consequences of gene flow from adjacent
arid habitats.
The LP/GF population is situated in an arid woodland habitat just uphill of
the riparian zone occupied by the HP/GF population. The short distance (100-m
transition) between the HP/GF and LP/GF populations at this Arizona study site
favors the differential movement of LP/GF spiders (primarily adult males in
search of mates) into the more favorable riparian zone
(Riechert, 1993b). The
evergreen woodland habitat is dominated by widely dispersed live oaks
(Quercus emoryi is the most common species) and the pine, Pinus
leiophylla var chihuahuana, is also present. The habitat is
classified as arid because the live oaks and pines provide little shade and
there is no litter accumulation. The substrate consists of sparsely
distributed grasses on bare ground and gravel. The LP/GF spiders are food
limited, do not experience significant levels of predation by birds, and must
compete for suitable web sites in an environment that provides little
protection from the sun and wind
(Riechert, 1993b).
The LP/GI population is located in desert grassland habitat bordering the
Carrizozo Lava Beds in south central New Mexico. The habitat is large, and the
A. aperta population, therefore, does not experience measurable gene
flow from riparian areas. A dense sod of the drop-seed grass, Sporobolus
flexuosus, characterizes this area. Prey levels are low and closely
approximate those recorded for the dry woodland site in Arizona
(Riechert, 1993b;
Riechert and Tracy, 1975).
Riechert and Hedrick (1990)
did not detect measurable levels of bird predation on A. aperta in
this habitat. Behavioral tests completed on LP/GI spiders indicate that their
aggressive reactions to prey, predators, and conspecifics are consistent with
adaptive equilibrium to a food-limited and highly competitive environment with
negligible levels of predation by birds (Riechert,
1979,
1981,
1991;
Riechert and Hedrick,
1990).
Superfluous killing measures
The superfluous killing trials were completed on at least 25 spiders from
each of the four populations (n, HP/GI = 35, HP/GF = 25, LP/GI = 39,
LP/GF = 25). We tested penultimate instars of both sexes and adult females,
but not adult males because they cease foraging upon reaching sexual maturity.
Riechert and Maupin (1998)
found no age- and sexclass differences in trial outcomes in their preliminary
study using the HP/GF population.
At capture, each spider was weighed and placed in its own plastic container, measuring 16 cm x 30 cm x 9 cm. We maintained the test spiders on European crickets (Gryllus domesticus, average mass = 37.9 mg) offered ad libitum at 3-day intervals. Individual spiders were tested when the following conditions were met: they had been in the laboratory for a minimum of 1 week; they had established a functional sheet-web; and 2 days had elapsed since their last feeding.
We offered each test spider sitting in a foraging mode at its funnel entrance a minimum of five weighed and individually paint-marked crickets. These were offered one at a time at an interval of 3 min. The 3-min time lapse between introductions was extended if a spider required additional time to capture and subdue a given prey item. Preliminary trials completed on LP/GF spiders identified no significant differences between an introduction interval of 3 min versus one of 20-min duration (t test comparisons: no. of prey captured, t = 1.18, p <.25; no. of partial feeds, t = 0.47, p <.65; no. of intact prey abandoned, t = 1.39, p <.20). We use the 3-min interval in the study reported here as it permitted a greater number of trials to be completed than would have been possible using a longer interval.
The fate of each prey item was recorded as either a capture if the spider attacked and subsequently subdued it or a miss if the cricket escaped attack or was ignored by the spider. If the test spider failed to attack all of the first five prey items offered, we terminated the trial. If all prey in the series were attacked, however, we continued to offer additional crickets at 3-min intervals until the spider failed to attack a prey within 5 min of introduction. Note that because the web of this spider is not sticky, prey that are not actively captured by the spider escape. We removed all escaped prey from the box housing the test subject's web at the end of the trial.
No spider was observed feeding on prey 24 h after capture in the feeding
trials completed by Riechert and Maupin
(1998). This was expected, as
in nature A. aperta does not cache extra prey but casts all items out
of its web after completion of a feeding bout. This may be due to the fact
that prey remains attract foraging ant columns. The ants consume the prey and
also cause A. aperta to abandon their webs (Riechert, personal
observations). Another reason that prey items are not cached is that they
rapidly desiccate in the low humidity environment of the desert southwest and
thus are not available for return feeding-bout visits. Based on this
information and the results of previous feeding trials, we checked the status
of captured prey items after 24 h had elapsed after the completion of each
feeding trial.
Through visual inspection, we assigned each prey item captured as abandoned
without feeding (prey intact), partially consumed, or fully consumed. We then
weighed all individual prey remains and added the weight of indistinguishable
parts of prey to these to obtain estimates of total mass left uneaten. Using
the individual weight determinations and a regression relationship determined
for prey desiccation in the absence of feeding, we tested the validity of our
visual assignment of particular prey. The weight loss in prey attributed to
desiccation over a 24-h period following mortality is:
 |
).
Prey remains of intermediate mass were assigned the status of partially
consumed. There was a close correspondence between our individual assignment
of crickets to the categories listed above and their final assignment based on
weighing and correction for desiccation. Only a couple of prey items were
reassigned on the basis of the weight determinations.
Breeding experiment
The genetic classes of individuals used in this study were derived from
reciprocal crosses of riparian and arid-land spiders. We collected 25 egg
masses from desert grassland habitat in New Mexico and another 25 egg masses
from riparian habitat in Arizona. In the controlled environment of a walk-in
chamber at the University of Tennessee, we reared 20 offspring from each egg
case in isolation at 29°C on a 12-h day clock and at 20°C on a 12-h
night clock. Each individual was provided unlimited food at 3-day intervals
through maturation (alternating flies, moths, and crickets with tenebrionid
beetle larvae and termites that provide a needed source of juvenile hormone to
A. aperta). Individuals initially were reared in plastic trays
housing 10 individual jelly containers measuring 2.5 cm x 4.6 cm x
3 cm. At this age, A. aperta is pinhead size and weighs < 1 mg. We
subsequently transferred the individuals to mailing-tube containers and then
to individual boxes measuring 16 cm x 30 cm x 9 cm as they grew in
size from 50 mg to mature individuals in excess of 150 mg. Maturation in the
laboratory environment takes approximately 6 months.
At maturity, we selected individuals at random from the two population
lines, arid and riparian. We staged matings of these in a reciprocal breeding
design, which produced the following two classes of individuals: (1) female
spiders with equal representation of arid and riparian tendencies to both
attack and to flee, and (2) male spiders that either had riparian or arid
tendencies to attack (sex chromosome contribution from mother), but an equal
contribution of riparian and arid tendencies to flee (autosomal contribution).
This breeding experiment is based on the predictions of the two-tendency
genetic model of Riechert and Maynard Smith
(1989) and empirical support
of this model (summarized in Riechert,
1999). From this prior work, we know that the males have different
sex chromosomal contributions on the same autosomal background, depending on
the source of the female parent (arid sex chromosome when female is arid and
riparian when female parent is of riparian origin). On the other hand, females
receive the same sex chromosomal and autosomal contributions regardless of the
direction of the cross. Eleven riparian (female parent) x arid (male
parent) sib groups and seven arid (female parent) x riparian (male
parent) sib groups were produced and reared as described above.
All males and females were scored in the superfluous killing trials at the penultimate stage of development, using the protocol already described. No more than 10 males and 10 females were scored from each sib group. Due to the large sample sizes involved in the breeding experiment and the results of our analyses of the population comparison, we limited our data collection to the conservative test of superfluous killing (i.e., frequency of abandonment of captured prey without feeding).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Population comparison
Spiders collected from the isolated LP environment captured significantly more prey (mean = 6.6 ± 0.44; t = 6.0, p <.0001) than did spiders collected from the isolated HP environment (mean = 3.3 ± 0.33). Prey mass captured showed the same significant result (LP/GI = 381.8 ± 25.3 mg, HP/GI = 153.2 ± 21.0 mg; t = 7.0, p <.0001), though consumption of prey did not vary between the two classes of spiders (LP/GI = 56.7 ± 8.5 mg, HP/GI = 76.6 ± 12.9 mg; t = 0.78, p <.08). These results suggest that the potential for superfluous killing exists in populations occupying prey-limited environments.
Feeding strategy representation
Figure 1 presents a
comparison of the frequency of full consumption of all prey captured versus
the exhibition of superfluous killing as evidenced by partial feeding on
multiple prey items and/or the abandonment of intact prey for all the
populations studied. Note that we included individuals that partially fed on a
single prey item in the full consumption category. This is because a spider
might reach satiation through partial feeding on a single large cricket, or it
may require part of an additional prey item to reach satiation after full
consumption of the first prey. The majority of the individuals from the LP
populations and HP population experiencing gene flow from an LP population
exhibited superfluous killing. The reverse was true for the HP population that
was genetically isolated. Chi-square tests completed on the frequency
distribution identified significant among-population differences in feeding
strategy (
2 = 41.3, df = 3, p <.0001). The fact
that the majority of individuals belonging to the HP/GI population fully
consumed all captured prey contributed high cell values to this significant
test result (HP/GI full consumption, 14.8; superfluous killing, 9.6).
|
Partial feeding on multiple prey and abandonment of intact prey were
exhibited by approximately the same proportion of individuals among the
populations showing a high incidence of superfluous killing
(Figure 2). The
among-population test for differences in the frequency representation of
partial feeding on multiple prey versus intact prey abandonment was
nevertheless significant (2 = 9.1, df = 3, p
<.03). The higher incidence of partial feeding compared to the abandonment
of intact prey in the case of the HP/GI population contributed prominently to
this significant test result (HP/GI cell 2 values: partial
feed = 2.7, abandon intact prey = 3.2). Only 20% of the individuals
representing this population exhibited superfluous killing. Although most
spiders from the HP/GF, LP/GF, and LP/GI populations captured prey that they
abandoned without consumption, only one spider from the HP/GI population left
a prey item completely uneaten. The little superfluous killing that was
exhibited in this population was in the form of partial feeding
(Figure 2).
|
The representation of full feeding, partial feeding, and prey abandonment within individual feeding bouts involving mixed-strategy representation is presented in Figure 3. We completed this analysis on those trials involving the capture of multiple prey: 75% of HP/GF individuals in this population's feeding trials, 20% of HP/GI, 100% of LP/GF, and 74% of LP/GI. Spiders from the LP environments fully consumed a smaller proportion of the prey items captured than did individuals representing the HP/GI population. The HP/GF population was intermediate in this respect and also in the proportion of prey captured that were abandoned without feeding (Figure 3). Although all of the individuals that attacked multiple prey partially fed on approximately the same proportion of the prey they had captured, a far greater proportion of intact prey were abandoned without feeding in trials involving individuals from LP environments (Figure 3). We completed an ANOVA to test for the influence of the number of prey items captured on the rate of prey abandonment. Significant prey capture number (F ratio = 28.2, p <.00001) and population by prey capture number interaction (F ratio = 6.3, p <.0006) effects contributed to a highly significant overall ANOVA test result (F ratio = 32.2, p <.00001). There was no significant population effect (F ratio = 0.96, p >.5).
|
10% but < 90% body mass remains; and uneaten, 
of
0.05). HP, high prey; LP, low prey; GI, genetically isolated; GF, gene flow.
Mean number of prey captured, HP/GI = 3.9 ± 0.3; HP/GF = 7.8 ±
0.7; LP/GI = 7.8 ± 0.6, and LP/GF = 7.5 ± 0.8.
Potential gene flow effects
We applied Shannon's equitability estimate (EH;
Magurran, 1988) to the bouts
recorded for each of the four populations:
 |
Breeding experiment
We used ANOVA to test for potential familial effects in the numbers of prey
abandoned in the feeding trials. Because no significant familial effects were
detected (F ratio = 1.4, p <.15), we pooled individuals
from different sib groups within the respective genetic classes in the
analyses reported here. A chi-square test performed on the frequency of trials
in which individual spiders abandoned at least one captured intact prey
without feeding detected a highly significant genotype difference
(2 = 186.3, df = 3, p <.0001). Males with the
riparian sex chromosome contribution exhibited a much lower frequency of
trials involving prey abandonment than did the male class with arid sex
chromosomes or the two female classes with both arid and riparian sex
chromosome contributions (Figure
4).
|
A similar relationship was observed when considering the numbers of intact prey that were abandoned after capture by individual spiders (Figure 5). An ANOVA performed on the numbers of prey captured but subsequently abandoned without feeding identified significant sex effects (F ratio = 64.4, p <.00001), genetic cross effects (F ratio = 18.2, p <.00001), and interaction effects (F ratio = 25.7, p <.00001). The significant sex and interaction effects reflect the genetic model prediction that F1 hybrids with both arid and riparian components for the tendency-to-attack trait will be more aggressive than either parental line, pure riparian or pure arid. Remember that this test is made possible by the fact that males possess only the sex chromosomes of their female parent, and tendency to attack (aggressiveness) is inherited on the sex chromosomes. The significant genetic cross effect reflects the extremely low frequency of abandonment of prey by hybrid males with riparian female parents (i.e., low tendency-to-attack genetic contributions).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
As observed in tests with other spider species (e.g., an ambush spider: Haynes and Sisojevic, 1966
; a
wandering spider: Samu and
Bíró,
1993; and a variety of web-building spiders:
Riechert and Maupin, 1998),
the desert spider, A. aperta, frequently captures many more prey than
it consumes. However, our laboratory trials indicate that this behavior is
less prevalent in populations that occupy high prey-level environments. Only
20% of the spiders from a riparian area in Texas that offered both high prey
levels and genetic isolation from low prey environments captured prey that
they failed to fully consume. In contrast, approximately 75% of the
individuals in the two low prey populations sampled in this study exhibited
superfluous killing in our laboratory trials. Partial consumption of prey items and abandonment of intact captured prey contribute equally to the superfluous killing exhibited by A. aperta except in the case of the HP/GI population from riparian habitat in Texas. With the exception of one individual in this population, partial feeding was equated with superfluous killing.
Hypotheses addressing adaptive function of superfluous killing
It is difficult to envision an adaptive function for abandoning prey that a
spider has expended effort to capture. Prey abandonment without immediate
feeding is correlated with food caching in small mammalian and avian predators
(Oksanen et al., 1985). This
possibility is not valid in the case of A. aperta because it engages
in web-cleaning behavior, which includes discarding the remains of all
captured prey after cessation of feeding. In the feeding trials completed in
this study, many A. aperta discarded intact prey along with the
remains of prey that had been partially fed on from the web. Similar web
cleaning that included removal of intact prey was also observed in the other
four web-building species studied by Riechert and Maupin
(1998).
Partial feeding, however, may be adaptive if selective feeding occurs on
the most easily digested or most nutritionally valuable parts of individual
prey. Pollard (1989) suggested
that an asymptotic feeding curve (decelerating feeding rate) is exhibited in
arthropods with external digestion because the viscosity of a prey's contents
increases as feeding progresses. An asymptotic feeding curve has been observed
in a number of arthropods with external digestion, including spiders
(Bailey, 1986;
Giller, 1980;
Griffiths, 1982;
Lucas and Grafen, 1985;
Pollard, 1989;
Riechert and Maupin, 1998;
Samu, 1993). However, the
interrupted feeding experiments completed by Riechert and Maupin
(1998) failed to observe this
pattern in A. aperta. Only one of the five species tested in the
study exhibited greater feeding efficiency earlier in a feeding bout than
later. Feeding efficiency does not appear to underlie the partial consumption
of multiple prey by A. aperta.
Another adaptive explanation for partial feeding has to do with the
nutritional inequality of the materials in a prey item and the fact that
enzymes attack the more digestible materials first
(Cook and Cockrell, 1978;
Formanowitz, 1984). The
chitinous exoskeleton of a cricket makes up between 5 and 6% of a cricket's
mass (Riechert, 1991).
Agelenopsis would be expected to feed on prey items in their entirety
in a food-limited environment as long as the nutritional gain is a positive
one. However, when faced with abundant prey, it might use only those portions
of prey that offer the highest nutritional rewards. The high rates of partial
feeding by A. aperta from prey-limited environments and full feeding
in spiders from the high prey environment is the opposite pattern from what
would be expected to occur under the nutritional value hypothesis.
Delineated feeding pattern
In our study, the majority of individuals exhibited the following feeding
pattern: full consumption of the first crickets fed on, with partial feeding
and abandonment of subsequent prey. Partial consumption and abandonment of
some prey, then, may merely be a consequence of the fact that spiders are
nearly sated after fully feeding on the first one or two prey. Samu
(1993) analyzed the fate of
individual prey fed to the wolf spider, Pardosa hortensis, which does
not build a web and, thus, completes consumption of a given prey item before
engaging in capture of the next. He allowed the spiders to feed to conclusion
on each prey item before offering the next item. He observed initial full
consumption of prey followed by partial consumption of later prey as well as
some abandonment of intact prey. From these results he suggested that full
consumption of initial prey is a consequence of adaptation to food scarcity
with partial consumption and abandonment of later prey, reflecting the effect
of gut satiation. If an individual is already sated, however, why attack
additional prey at all?
Full feeding reflects phylogenetic inertia in high prey-level
environments
From the results of our work with A. aperta, we suggest that
phylogenetic inertia explains feeding to satiation on the first prey or two by
spiders occupying high prey environments. Spiders show many adaptations to
food-limited environments, including low metabolic rates, extra-oral
digestion, a highly expandable abdomen, and a sit-and-wait foraging strategy
(see review in Riechert,
1992). Full consumption of the first few items as long as the
energetic reward is positive is another strategy that is consistent with
adaptation to food-limited environments. Sticky trap records summarized in
Riechert
(1993b,c)
indicate that encounters with multiple prey at rates similar to those we
offered spiders in our trials are experienced in only 4-6% of the trap days in
the three arid-land habitats sampled. The risk of low prey encounter feeding
bouts, then, is high, and full feeding on the first items captured is
expected. On most days, however, riparian A. aperta experience high
rates of encounter with prey. Here full feeding is not expected if partial
consumption of several items provides a higher return. The spiders from the
high prey populations, nevertheless, exhibited full feeding on the first
items.
Superfluous killing as a consequence of adaptation to food-limited
environments
We believe that the capture of excessive numbers and biomass of prey is a
consequence of the level of aggressiveness toward prey that is selected for in
different prey availability environments. The results of the feeding trials
for both the comparative population and breeding experiment studies are
consistent with the idea that populations of A. aperta exhibit levels
of aggressiveness toward prey that are adaptive to local prey availability
levels. Where encounter with prey is typically low, the spiders need to
exhibit a rapid and aggressive response to prey encountering their nonsticky
web-sheets. Where prey availability is high, spiders may exhibit a more
cautious response to potential prey items encountering their webs
(Riechert, 1991). Often
potential prey items may turn into potential predators. There is another cost
to attacking prey that takes a long time to capture for the energy reward
received: time spent on the web involved in prey capture exposes spiders to
predation and environmental extremes (e.g., high temperatures and winds).
The performance of different genetic classes clearly shows the influence of genotype on excess capture of prey. Both the proportion of individuals that abandoned at least one prey without feeding on it and the numbers of intact prey abandoned showed close correspondence to the presence of the dominant high tendency-to-attack genetic component of aggressiveness that is inherited on the sex chromosomes. Hybrid females regardless of the direction of the cross and males possessing sex chromosomes of the arid, high tendency-to-attack type showed high levels of excess killing (> 90% of the individuals with an average number of 4 prey items abandoned/trial). Males with the recessive low tendency-to-attack component of aggressiveness, in contrast, exhibited almost no intact prey abandonment (< 15% of the individuals with an average number of 0.19 prey items/trial).
Conclusions
Adaptation to food-limited environments appears to lead to the killing of
multiple prey when high-density patches of prey are encountered spatially or
temporally. In these instances, prey are attacked and subdued in quantities
that are well beyond the consumptive capabilities of individual spiders. This
superfluous killing probably is not itself adaptive, but rather is a
consequence of selection for aggressiveness toward prey. Studies of the
impacts of predators on associated prey populations are often based on
estimates of the numbers of predators and prey. Superfluous killing effects
will produce significant errors in model predictions if they are not
incorporated in the analyses.
|
ACKNOWLEDGEMENTS |
|---|
We thank the University of Tennessee, Knoxville, spider group for helpful comments on earlier drafts. J.M.'s work on this project was supported by a Howard Hughes Medical Institute undergraduate curriculum development grant to the University of Tennessee. Additional support was provided through Population Biology and Animal Behavior Program grants from the National Science Foundation to S.R.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bailey PCE, 1986. The feeding behaviour of a sit-and-wait predator, Nanatra dispar (Heteroptera: Nepidae): optimal foraging and feeding dynamics. Oecologia 68: 291-297.
Bjarvall A, Nilsson E, 1976. Surplus-killing of reindeer by wolves. Solna, Sweden: National Environment Protection Board.
Carbyn LN, 1983. Wolf predation on elk in Riding Mountain National Park, Manitoba. J Wildl Manag 47: 963-976.
Conover RJ, 1966. Factors affecting the assimilation of organic matter by zooplankton and the question of superfluous feeding. Limnol Oceanogr 11: 346-354.
Cook RM, Cockrell BJ, 1978. Predator ingestion rate and its bearing on feeding time and the theory of optimal diets. J Anim Ecol 47: 529-547.
Ehrlinge S, Bergsten B, Kristiansson B, 1974a. The stoat and its prey: hunting behavior and escape reactions. Fauna Flora 69: 203-211.
Ehrlinge S, Jonsson B, Willstedt H, 1974b. Hunting behavior and prey choice of the weasel. Fauna Flora 69: 95-101.
Formanowitz DR Jr, 1984. Foraging tactics of an aquatic insect: partial consumption of prey. Anim Behav 32: 774-781.
Foster SA, Endler JA (eds), 1999. Geographic variation in behavior. Oxford: Oxford University Press.
Giller PS, 1980. The control of handling time and its effects on the foraging strategy of a heteropteran predator, Notonecta. J Anim Ecol 49: 699-712.
Griffiths D, 1982. Test of alternative models of prey consumption by predators, using antlion larvae. J Anim Ecol 51: 363-373.
Hammerstein P, Riechert SE, 1988. Payoffs and strategies in spider territorial contests: ESS-analyses of two ecotypes. Evol Ecol 2: 115-138.
Haynes DL, Sisojevic P, 1966. Predatory behavior of the Philodromus rufus Walckenaer (Araneae: Thomisidae). Can Entomol 98: 113-133.
Johnson DM, Akre BG, Crowley PH, 1975. Modeling arthropod predation: wasteful killing by damselfly naiads. Ecology 56: 1080-1093.
Krebs JR, Davies NB, 1993. An introduction to behavioral ecology, 3rd ed. Oxford: Blackwell Scientific.
Lucas JR, Grafen A, 1985. Partial prey consumption by antlion larvae. Anim Behav 33: 945-958.
Magurran AE, 1988. Ecological diversity in space and time. Princeton, New Jersey: Princeton University Press.
Maynard Smith J, Riechert SE, 1984. A conflicting tendency model of spider agonistic behaviour: hybrid-pure population line comparisons. Anim Behav 32: 564-578.
Mayr E, 1970. Populations, species, and evolution. An abridgement of Animal species and evolution. Cambridge, Massachusetts: Harvard University Press.
Metz JAJ, Sabelis MW, Kuchlein JH, 1988. Sources of variation in predation rates at high prey densities: an analytic model and a mite example. Exp Appl Acarol 5: 187-205.
Miller KL, Gunn A, Broughton E, 1985. Surplus killing as exemplified by wolf predation on newborn caribou. Can J Zool 63: 295-300.
Oksanen L, Oksanen T, 1981. Lemmings (Lemmus lemmus) and grey-sided voles (Clethrionomys rufocanus) in interaction with their resources and predators on Finnmarksvidda, northern Norway. Rep Kevo Subarct Res Stn 17: 7-31.
Oksanen T, Oksanen L, Fretwell SD, 1985. Surplus killing in the hunting strategy of small predators. Am Nat 126: 328-346.
Pollard SD, 1989. Constraints affecting partial prey consumption by a crab spider, Diea sp. indet (Araneae, Thomisidae). Oecologia 81: 392-396.
Riechert SE, 1979. Games spiders play II. Resource assessment strategies. Behav Ecol Sociobiol 4: 1-8.
Riechert SE, 1981. The consequences of being territorial: spiders, a case study. Am Nat 117: 871-892.
Riechert SE, 1991. Prey abundance versus diet breadth in a spider test system. Evol Ecol 5: 327-338.
Riechert SE, 1992. Spiders as representative sit-and wait predators. In: Biology of natural enemies (Crawley MJ, ed). London: Blackwell; 313-328.
Riechert SE, 1993a. The evolution of behavioral phenotypes: lessons learned from divergent spider populations. Adv Anim Behav 22: 103-134.
Riechert SE, 1993b. Investigation of potential gene flow limitation of behavioral adaptation in an aridlands spider. Behav Ecol Sociobiol 32: 355-363.
Riechert SE, 1993c. A test for phylogenetic constraints on behavioral adaptation in a spider system. Behav Ecol Sociobiol 32: 343-348.
Riechert SE, 1999. The use of behavioral ecotypes in the study of evolutionary processes. In: Geographic variation in behavior (Foster S, Endler J, eds). Oxford: Oxford University Press.
Riechert SE, Hedrick AV, 1990. Levels of predation and genetically-based anti-predatory behavior in the spider, Agelenopsis aperta. Anim Behav 40: 679-687.
Riechert SE, Hedrick AV, 1993. Evidence for phenotypic correlation among a suite of fitness-linked behavioral traits. Anim Behav 46: 669-675.
Riechert SE, Maupin JL, 1998. Spider effects on prey: tests for superfluous killing in five web-builders. In: Proceedings of the 17th European Colloquium of Arachnology, Edinburgh, 1997 (Selden PA, ed). Bucks: British Arachnological Society, Burnham Beeches; 203-210.
Riechert SE, Maynard Smith J, 1989. Genetic analyses of two behavioral traits linked to individual fitness in the desert spider, Agelenopsis aperta. Anim Behav 37: 624-637.
Riechert SE, Tracy CR, 1975. Thermal balance and prey availability: bases for a model relating web-site characteristics to spider reproductive success. Ecology 56: 265-284.
Samu F, 1993. Wolf spider feeding strategies: optimality of prey consumption in Pardosa hortensis. Oecologia 94: 139-145.
Samu F, Bíró Z, 1993. Functional response, multiple feeding, and wasteful killing in a wolf spider (Araneae: Lycosidae). Eur J Entomol 90: 471-476.
Smith RB, Wellington WG, 1986. The functional response of a juvenile orb-weaving spider. In: Proceedings of the Ninth International Congress of Arachnology, Panama, 1983 (Eberhard WG, Lubin YD, Robinson BC, eds). Washington DC: Smithsonian Institution Press; 275-279.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||