FACTORS AFFECTING CANNIBALISM AMONG NEWLY HATCHED WOLF SPIDERS (LYCOSIDAE, PARDOSA AMENTATA)

2005·7.8 MB·English

by  Aino Hvam

2005. The Journal of Arachnology 33:377-383
FACTORS AFFECTING CANNIBALISM AMONG
NEWLY HATCHED WOLF SPIDERS
(LYCOSIDAE, PARDOSA AMENTATA)
Aino Hvam', David Mayntz^2,3 Rikke Kruse Nielsen*: “Department of Ecology
and Genetics, University of Aarhus, Bldg. 540, DK-8000 Arhus C, Denmark;
^Department of Zoology, University of Oxford, South Parks Road, Oxford
0X1 3PS, UK
ABSTRACT. Cannibalism is a common phenomenon among young wolf spiders (Lycosidae). The pur-
pose of this study was to investigate how various factors influence cannibalistic tendencies in hatchlings
ofPardosa amentata (Clerk 1757). The basic experimental approach was toplace pairs ofunfedhatchlings
ofsimilar body mass in small containers without prey and to measure ifand when cannibalism happened.
From the data, we identified three different cannibalistic strategies. One large group of hatchlings never
cannibalized and thus died from starvation. Another group cannibalized shortly before the time at which
they were predicted to die from starvation. In these spiders, there was a strong positive relationship
between average body mass ofthe contestants and their latency to cannibalize. A third group cannibalized
quickly and the latency to cannibalize in these spiders was independent of body mass. We also tested if
cannibalistic tendencies were higher among unrelated pairs than among pairs of siblings, but we did not
find any support for this hypothesis. In another experiment we tested if maternal effects influenced can-
nibalism, i.e. if siblings from certain mothers were more cannibalistic than siblings from others. We did
not find any evidence that maternal effects influenced whether or not cannibalism occurred. However,
when cannibalism did occur, the latency to cannibalize varied significantly among siblings from different
mothers beyond what would have been predicted solely from hatchling body mass.
Keywords: Lycosidae, intraspecific predation, spiderlings, kin recognition, maternal effects
Cannibalism among wolf spiders is often nibals are facing prey with similar predatory
observed in the field. One example is Pardosa abilities, an obvious cost ofcannibalism is the
lugubris (Walckenaer 1802) which seems to risk of retaliation. Another intriguing cost of
be the most important predator ofits own spe- cannibalism is the potential loss of inclusive
cies (Edgar 1969), and it was estimated that fitness when a cannibal kills a genetically re-
&
juveniles of this species included conspecifics lated individual (Elgar Crespi 1992; Pfen-
as 29% of their total diet (Hallander 1970). nig & Sherman 1995). Ifthis cost is large, we
Other examples are Schizocosa ocreata would predict spiders to be able to distinguish
(Hentz 1844) and Pardosa milvina (Hentz between kin and non-kin and to treat kin and
1844) in which cannibalism is assumed to be non-kin differently (Pfennig & Sherman
an important regulating factor on population 1995). Kin recognition has been shown to oc-
density (Wagner & Wise 1996; Buddie et al. cur in many cannibalistic animals (see refer-
2003). ences in Pfennig 1997) and also some spiders
A variety of factors has been suggested as seem to be able to identify and subsequently
potential selective forces, promoting or inhib- avoid eating a close relative (Evans 1999; Bil-
iting cannibalistic behavior. The most obvious de & Lubin 2001; Anthony 2003; Roberts et
advantage connected with cannibalism is that al. 2003). Wolf spider females carry their
the cannibal gains a meal in addition to the young on the abdomen for about a week.
normal diet and cannibals often show higher Thus, hatchlings have a good opportunity to
growth and survival rates than their non-can- learn chemical or visual cues, which could lat-
nibalistic conspecifics (Polis 1981). As can- er be used to recognize siblings from non-sib-
lings.
^Corresponding author. Adult wolf spiders can survive starvation
377
378 THE JOURNAL OF ARACHNOLOGY
for several months (Anderson 1974). Newly
hatched spiderlings on the otherhand can only
survive a few days or weeks before their nu-
trient reserves are depleted (Wagner & Wise
1996; Toft & Wise 1999). This means that the
first meal is of utmost importance for spider-
lings and cannibalism can therefore be impor-
tant for juvenile survival.
In the present experiments we investigated
cannibalistic tendencies among equally sized
pairs of unfed hatchlings and provided them
no choice other than to cannibalize or to die
from starvation. Using this approach we eval-
uated different hypotheses about what influ-
ences cannibalistic tendencies in the hatch-
lings. In the first experiment, we paired sibling
and non-sibling hatchlings in order to test if
cannibalism was dependent on kinship. From
these results we also describe three apparently
different strategies among hatchlings. Poten- Average bodymass (mg)
tially, a mother of a brood can affect the con-
dition of her spiderlings and thus also their
2
cannibalistic propensities, for example
through her nutritional status before reproduc-
tion. The rates of cannibalism may also vary
between closely related species or among and 6.5
even within populations, due to genetic dif- O
ferences (Thibault 1974; Stevens & Mertz o>
1985; Tarpley et al. 1993). In a second exper- 6.0
iment we therefore tested if there was varia- U)
tion in cannibalistic tendencies among hatch- (/)
Q 5.5
lings descending from different mothers.
Starvation Cannibalism
METHODS
Cause ofdeath
— —
The wolf spider. The wolf spider Par- Figures 1-2. 1. Effects ofbody mass and cause
dosa amentata is abundant in Europe in many of death on the survival time of wolf spider hatch- '
open, humid habitats, especially grasslands lings (Experiment 1). Survival time was measured
and agricultural fields with a well-developed from the time spiderlings were paired. Each point
litter layer (Alderweireldt & Maelfait 1988). represents the time passed until one ofthe two spi-
Reproduction takes place in May-July. Fe- ders in a pair died from starvation (black triangles)
males carry the eggsac for 2-3 weeks and or from cannibalism (open circles). Body mass is
the average mass of the two hatchlings in a pair.
hatchlings spend about one week on their Regression lines are based on spiders thatdied after
mother’s abdomen before they disperse (Rob- day 3, i.e > 7days old (above horizontal dotted
erts 1995). — line); death from starvation = broken regression
Experiment 1. The purpose of this ex- line, cannibalism = solidregressionline. 2. Number
periment was to test if kinship affected can- of days survived adjusted for average body mass ^
f
nibalism and to describe the cannibalistic ten- (least squares means, calculated on spiders dying
dency ofhatchlings in general. Subadult male after day 3 in the experiment, i.e. > 7 days old).
and female P. amentata spiders were collected
in spring in a meadow at Stjaer, Denmark
(56°07'N, 9°9UE), and brought to the labo- frequently to maintain a permanent high hu- “
ratory. They were housed individually in plas- midity in the container. The spiders were fed
tic containers (diameter 35 mm, height 80 wild type Drosophila melanogaster (Meigen)
mm) with a plaster bottom, which was wetted in excess until maturity. Fruit flies were raised
HVAM ET AL.—CANNIBALISM AMONG WOLF SPIDER HATCHLINGS 379
weight difference = 14pg). Spideiiing age
(days since hatching) at the start of the ex-
periment varied up to three days within a pair.
Experimental conditions were set at 25 ±0.1
°C; 16L:8D. The spiderlings never received
any food but had constant access to water
from the plaster. Spiders were checked for
deaths twice daily. Cannibalism left clear
marks of partly or fully digested body parts
and a stereomicroscope was used in case of
doubt. An outcome of the experiment was re-
corded when one of the two hatchlings was
dead, due to starv—ation or cannibalism.
Experiment 2. The purpose of this ex-
periment was to test if the tendency to can-
nibalize varied among spiderlings from differ-
ent eggsacs. Females of P. amentata with an
eggsac were collected from the same location
as in experiment 1, thus, we were only able
to test for maternal effects on cannibalism and
Average bodymass (mg)
not for paternal effects. The spiders were tak-
—
Figure 3. Relationship between average body en to the laboratory and kept as in experiment
mass of hatchling pairs and their latency to canni- 1. At 4 days of age, hatchlings from 19 egg-
balize (Experiment 2). The two spiderlings always sacs were weighed. At day 6, hatchlings of
originated from the same eggsac. Each point rep- approximately the same body mass were
resents the time passed until cannibalism occurred. paired (mean body mass ± SE = 519 ± 3 pg;
The regression line is based on spiders cannibaliz- mean weight difference ± SE = 2.7 ± 0.2pg;
i>ng7afdtaeyrsdaolyd.1A(baobuotve6h4o%rizoofnttahledsoptitdeedrslindei)d, ni.oet. max. weight difference = 13pg). In all pairs,
cannibalize at all and are not shown in the figure. the two hatchlings descended from the same
eggsac {n = 205 pairs, 4-17 pairs from each
eggsac). Experimental conditions and proce-
on instant Drosophila medium (formula 4-24 dures were the sa—me as in experiment 1.
Plain, Carolina Biological Supply, Burlington Data analysis. Differences in the cause of
NC), mixed with crushed dogfood (Techni-Cal death between kin and non-kin hatchling pairs
ADULT®, Martin Pet Foods, Canada). Fully were analyzed using the Pearson statistic. The
matured females were mated with a single latency to cannibalize between kin and non-
male and different males were used for each kin were analyzed using Student’s t-test, after
female in order to avoid offspring from dif- testing for equal variances (Bartlett’s Test, a
ferent females being half-siblings. The young > 0.05). Linearregression was used to test for
descending from eight eggsacs were chosen correlation between mean body mass of pairs
for the experiment. These eggsacs hatched and the time spent before one of the two
within a period of four days and when hatch- hatchlings died from either cannibalism or
lings were 4 ± 1 days old, they were weighed starvation. Regression lines were analyzed us-
to the nearest pug. Pairs ofhatchlings descend- ing Analysis of Covariance (ANCOVA), with
ing from the same eggsac {n = 71) or from body mass as the covariate. First, we tested
different eggsacs {n = 71) were then placed for equal slopes and if they were not signifi-
in the same plastic tube (diameter 20 mm, cantly different, we tested if intercepts were
height 60 mm). The tubes contained a plaster equal and calculated means adjusted for the
bottom, which was wetted frequently to main- covariate (least squares means). We used lo-
tain a permanent high humidity in the con- gistic regression to test ifhatchlings from dif-
tainer. Body mass asymmetry was avoided by ferent eggsacs differed in their probability to
pairing spiders of almost equal body mass cannibalize or to die from starvation. All sta-
(mean body mass ± SE = 421 ±4 pg; mean tistical analyses were performed with IMP 5.0
weight difference ± SE = 2,6 ± 0,2pg; max. for Windows (SAS Institute).
380 THE JOURNAL OF ARACHNOLOGY
RESULTS occurred (linear regression, t = 0.93, w = 14,
resEuxltpienrgiimnenatca1n.n—ibaTlhiestipcroepvoernttiownasofnotpaiarfs- P E=xp0.e3r7i,mRe2nt=2.0—.07W).e found the same three
cannibalistic patterns in this experiment as de-
fected by kinship, i.e. whether or not the two
spiders in a pair originated from the same scribed from experiment 1 (Fig. 3). Either spi-
eggsac (Pearson = 0.26, P = 0.61; siblings ders did not cannibalize at all (64.4%); they
4m1ore%,,tnhoen-tsiimbelipnagsssi4n5g u%n,tilwa=can1n4i2b)a.liFsutritchearc-t Pc(a2n5<n.i40b%.a,0l0il0zi1en,deaRri2nrea=grb0eo.s2ds9iy)o;nm,oarsnsth=edye5p2ce,anndtnei=nbtal4iw.z5ae7y,d
occurred did not differ between sibling and
lnsiionbngl-sisnigb=sli=8n.g989p.a8i±2rs±0(.t4-0t9.es4td0,aydDsaFys±=±S5E9)S,.E,PThn>uosn-,0s.i8wb0e;- Pfw(oi1rt0=eh.i2b0n%e.,i8tn6hlg,ein7fRei2radsrat-yrdsea0goy.rl0ed0os)2fs)irt.oehnge,arendxlpe=esrsi2om1fe,nbttod(=iy.e0m..ab1se8s-,
found no evidence that relatedness affected
Mother identity did not affect whether or
the cannibalistic tendency in hatchlings. In our
not cannibalism occurred within a pair of
description of general patterns of cannibalism hatchlings (logistic regression, Wald ~
below, we therefore pool data from siblings 23.05, DF = 18, R = 0.19). However, when
and non-siblings.
cannibalism did occur, the latency to do so
Fig. 1 shows the relationship between av- varied significantly among hatchlings from
erage body weight of hatchling pairs and the
different eggsacs, after correcting for the ef-
time until one of the two hatchlings died. The fect of body mass (ANCOVA on the latency
data indicate a presence ofthree different can- to cannibalize with mean body mass as cov-
nibalistic strategies. One group of hatchlings ariate, DF = 13, SS = 186.5, F = 5.20, P <
never cannibalized (57%) and consequently
0.0001, Fig. 4); five eggsacs in which fewer
died from other reasons than cannibalism. In than three pairs cannibalized were omitted
this non-cannibalistic group, there was a pos- from this analysis, thus, 14 eggsacs were in-
itive correlation between mean body mass and cluded with a total of 65 pairs.
survival time of the first dying hatchling (lin-
ear regression, t = 8.45, n = 79, P < 0.0001; DISCUSSION
R2 = 0.48). A similar type of positive corre- The results of this study indicate that three
lation was found in pairs where cannibalism different cannibalistic strategies exist in the
happened after 3 days ofthe experiment (33% wolf spider hatchlings. Either we observed no
of all pairs; linear regression, t = 6.95, n — cannibalism, late and size dependent canni-
47, P < 0.0001; R2 = 0.52). The regression balism, or early and size-independent canni-
line of spiders cannibalizing after day 3 ofthe balism. This pattern appeared in two separate
experiment (i.e. > 7 days old) and the regres- experiments, which suggests that it is a gen-
sion line of spiders dying from other reasons eral pattern of this wolf spider species.
than cannibalism did not have significantly More than half of the spiderlings belonged
different slopes (ANCOVA, SS = 0.04, F = to the group that never cannibalized and con-
0.03, P = 0.85, Fig. 1), but the intercepts of sequently died from other causes than canni-
the two regression lines differed significantly balism. As all spiderlings were deprived of
(ANCOVA, SS = 4.53, F = 4.20, P = 0.04). food we expect that the main part of these
The least squares means of survival days ad- non-cannibalizing spiders died from starva-
justed for body mass showed that spiders can- tion. The body mass of an animal probably
nibalizing after day 3 (>7 days old), did so correlates positively with the amount of nu-
on average 0.4 days (i.e. less than 10 h) before trient reserves that are stored in the body. Fur-
equal sized spiders would die from other rea- thermore, light animals are often found to
sons than cannibalism (Fig. 2). Besides the have proportionally higher specific metabolic
two strategies where spiders either died or rate than heavier animals (Edwards 1946;
cannibalized in a size dependent way, 10% of Phillipson 1963). Together, these two factors
the pairs cannibalized early, within the first 3 may explain the observed pattern of lighter
days of the experiment. Among these pairs, spiders dying from starvation sooner than
there was no correlation between mean body heavier spiders.
mass and the time passing until cannibalism The spiders that did cannibalize could be
HVAM ET AL.—CANNIBALISM AMONG WOLF SPIDER HATCHLINGS 381
81
Eggsacs
—
Figure 4. Effect of eggsac origin on the latency to cannibalize in pairs of equal sized siblings (Ex-
periment 2). Points show the time passing until cannibalism occurred adjusted for the effect ofbody mass
{n = 3--9 pairs of siblings per eggsac, five eggsacs were not included because less than 3 pairs of spi-
derlings from these eggsacs cannibalized).
divided in two groups: a group where the on- and spadefoot toad tadpoles (Pfennig et al.
set of cannibalism was dependent on body 1993) individuals can be divided into canni-
mass, and a group, which cannibalized early. balistic and non-canriibalistic forms and can-
In the body mass dependent cannibalism ligh- nibalistic individuals are often characterized
ter pairs cannibalized earlier than heavier by actual morphological and physiological
pairs, which suggests that the latency to can- differences that enhance this feeding strategy.
nibalize depended on their level of nutrient Field studies have shown that conspecifics
reserves. In fact, this group ofspiderlings gen- comprise a large part of the diet in juvenile
erally waited to cannibalize almost until the and adult wolfspiders (Edgar 1969; Hallander
time when they were predicted to die from, 1970). However, in this experiment spider-
starvation, which suggests that they chose to lings were ratherreluctantto cannibalize, even
cannibalize as a very last option. In a rela- though they were kept in the same container
tively small proportion of spider pairs (ca. with no escape possibilities. Why did the ma-
10%) cannibalism appeared early in the ex- jority ofhatchlings refuse to cannibalize when
periment regardless of their body mass. This the consequence of such a decision is death
group of spiders did not seem to be under se- from starvation? Our experimental setup does
vere food stress when the cannibalism oc- not provide a clear answer to that question.
curred, suggesting that these spiders had a One likely explanation is that they fear the
higher keenness to cannibalize. Different can- cost of retaliation.
nibalistic strategies among individuals within The risk associated with attacking decreases
a species have also been observed in other as the asymmetry in body mass/size increases.
animals. In salamanders (Laenoo et al. 1989) Samu et al. (1999) found that the body mass
382 THE JOURNAL OF ARACHNOLOGY
ratio between two juvenile spiders was the sibling from a non-sibling, or they do not care
most important factor influencing cannibal- and cannibalize nevertheless. These results are
ism, and cannibalism was not observed within also in contrast to data on social (Diaea er-
24 hours if the body mass ratio was less than gandros Evans 1995) and sub-social (Stego-
2:1 (predatoriprey). Here we paired spider- dyphus lineatus Latreille 1817) spiders (Evans
lings of equal body mass, which in principle 1999; Bilde & Lubin 2001), in which the stud-
have similar predatory abilities and therefore ies showed kin recognition and kin discrimi-
provide roughly 50/50 chance ofdying, unless nating cannibalistic behavior. Compared to
there are different risks associated with being solitary spiders, social spiders and sub-social
an attacker or a defender. The fact that a large spiderlings spend long periods of time close
proportion of the spiders postponed cannibal- to relatives and it is possible that such fre-
ism almost until they died from starvation in- quent encounters with relatives are a require-
dicates that risk of retaliation or other factors ment for the evolution ofkin recognition (Bil-
&
inhibit cannibalism. It is possible that canni- de Eubin 2001). Spiderlings of a clutch do
balism occurred when the risk of dying from not leave their mother’s abdomen at the same
starvation had outweighed these risks. We time but dispersal is distributed over several
cannot exclude the possibility that some ofthe days and over a relatively large area (D.
cannibalistic events happened after one of the Mayntz, pers. obs.). Thus, the only time wolf
spiders was dead or almost dead from star- spiders have a high chance ofmeeting siblings
vation. If so, then cannibalistic acts should is when the spiderlings are gathered on their
only confer little or no risk of retaliation. An- mother’s abdomen. Avoiding cannibalism of
other potential cost of cannibalism is the risk kin may possibly be accomplished during oth-
of receiving pathogens from conspecific prey er routes than actual kin recognition. For ex-
(Pfennig et al. 1998). If this is a real cost in ample, intra-brood cannibalism in Pardosa
the field, it would explain the general reluc- pseudocmnidata (Bosenberg & Strand, 1906)
tance to cannibalize in the majority ofthe spi- rarely occurred due to the small size differ-
ders. However, we are not aware of any path- ence within the brood (lida 2003). Moreover,
ogens that might cause such a risk in wolf P. pseiidoannulata did not seem to cannibalize
spiders, especially not among young hatch- siblings less frequently than non-siblings (i.e.
lings. no evidence for kin recognition).
A general inhibition of cannibalism can be When we tested for variation in cannibal-
an indirect method to avoid eating relatives. istic tendencies among hatchlings from differ-
Where such an inhibition has been demon- ent eggsacs, we did not find any evidence that
strated, it is often expressed in certain life maternal effects influenced whether or not
stages. Filial cannibalism, for example, is in- cannibalism happened. However, when can-
hibited in reproductively active females ofthe nibalism did occur, hatchlings from different
wolf spider Scizocosa ocreata (Wagner 1995). eggsacs showed variable latencies to do so
Moreover, cannibalism was less frequent in (Fig. 4). We collected the eggsacs in the field.
the 2nd instar ofthe wolfspiderHogna helluo This made it impossible for us to assess the
(Walckenaer 1837), compared to 3rd instar genetic influence from the fathers, and pre-
spiderlings (Roberts et al. 2003). Avoidance vented us from separating genetic effects from
of related prey can also be direct through kin other maternal effects that might have affected
recognition where relatives are recognized and the hatchlings’ tendency to cannibalize. Be-
disregarded as prey (Pfennig 1997). There is yond pure genetic factors, possible maternal
one study that supports kin discrimination factors affecting cannibalistic tendency may
among young spiderlings in a wolf spider include the nutritional history of the mother,
(Roberts et al. 2003). In this species a higher the age of mother, or size of the brood. Her-
frequency of cannibalism was observed in itability of cannibalistic behavior has been
pairs ofnon-siblings compared to pairs ofsib- shown in fish, flour beetles, corn borers and
lings. In the present experiment, we did not ladybird beetles (Thibault 1974; Stevens &
find any evidence supporting the hypothesis Mertz 1985; Tarpley et al. 1993; Wagner et al.
that siblings cannibalized each other less fre- 1999) but so far not in spiders. Half-sib ex-
quently than non-siblings. Thus, either Par- periments or actual selection experiments are
dosa amentata hatchlings cannot recognize a needed before we can clarify how much ge-
hvam et al=—cannibalism among wolf spider hatchlings 383
netic effects contribute to the observed vari= Pfennig, D.W. 1997. Kinship and cannibalism. Bio-
ation in the latency to cannibalize. Pfesncniiegn,ceD.4W7.:,667S-.G6.75.Ho & E.A. Hoffman. 1998.
ACKNOWLEDGMENTS
Pathogen transmission as a selective force
We are indebted to S0ren Toft, Gitte Skov- against cannibalism. Animal Behaviour 55:
ciuonmdmeanntdsJeornoeetheVamnadneuescBroirprte afnodr vtaoluEalbslee Pfe1n2ni5g5,-1D2.6W1., H.K. Reeve & RW. Sherman. 1993.
Bomholt Rasmussen for assistance in the lab- Kin recognition and cannibalism in spadefoot
toad tadpoles. Animal Behaviour 46:87-94.
oratory. Pfennig, D.W. & RW. Sherman. 1995. Kin recog-
LITERATURE CITED nition. Scientific American 272:68-73.
Alderweireldt, M. & J.-R Maelfait. 1988. Life cy- Phillipson, J. 1963. The use of respiratory data in
cle, habitat choice and distribution of Pardosa estimating annual respiratory metabolism, with
amentata (Clerck, 1757) in Belgium (Araneae, particular reference to Leiobunum rotundum
Lycosidae). Bulletin de laSociete Scientifiquede (Latr.) (Phalangiida). Oikos 14:212-223.
Bretagne 59:7--15. Poiis, G.A. 1981. The evolution and dynamics of
Anderson, J.E 1974. Responses to starvation in the ietraspecific predation. Annual Review of Ecol-
spiders Lycosa lenta Hentz and Filistata hiber- ogy and Systematics 12:225-251.
nalis (Hentz). Ecology 55:576-585. Roberts, M.J, 1995. Spiders ofBritain andNorthern
Europe. Harper Collins Publishers, London.
Anthony, C.D. 2003. Kinship influences cannibal-
iofsmInisnecttheBewhoalvfiosprid1er6,:2P3a-r3d6.osa milvina. Journal RobBeartths,,EJn.Ag.l,anRdW.. Taylor & G.W. Uetz. 2003. Kin-
Bilde, T. & Y. Lubin. 2001. Kin recognition and ship and food availability influence cannibalism
cannibalism in a subsocial spider. Journal of tendency in early-instar wolf spiders (Araneae:
Evolutionary Biology 14:959-966. Lycosidae). Behavioral Ecology and Sociobiol-
& ogy 54:416-422.
Buddie, C.M., S.E. Walker, A.L. Rypstra. 2003.
Cannibalism and density-dependent mortality in Samu, E, S. Toft & B. Kiss, 1999. Factors influ-
the wolf spider Pardosa milvina (Araneae: Ly- encing cannibalism in the wolf spider Pardosa
cosidae). Canadian Journal ofZoology 81:1293- agrestis (Araneae, Lycosidae). Behavioral Ecol-
ogy and Sociobiology 45:349-354,
1297.
&
Edgar, W.D. 1969. Prey and predators of the wolf Stevens, L. D.B. Mertz. 1985. Genetic stability
spider Lycosa lugubris. Journal ofZoology 159: ofcannibalism in Tribolium confusum. Behavior
405-411. Genetics 15:549-559,
Edwards, G.A. 1946. The influence of temperature Tarpley, M.D., F. Breden & G.M. Chippendale.
upon the oxygen consumption of several arthro- 1993. Genetic control ofgeographic variationfor
pods. Journal ofCellular and Comparative Phys- cannibalism in the southwestern corn borer, Dia-
iology 27:53-64. traea grandiosella. Entomologia Experimentalis
Elgar, M.A, & B.J. Crespi. 19—92. Ecology and evo- et Applicata 66:145-152.
lution ofcannibalism. Pp 1 12. InCannibalism: Thibault, R.E. 1974. Genetics of cannibalism in a
Ecology and Evolution Among Diverse Taxa. viviparous fish and its relationship to population
v(eMr.sAi.tyEPlrgeassr,&NeBw.J.YoCrrke.spi, eds.). Oxford Uni- Tofdt,ensS.ity&. DN.aHt.urWeis2e5.1:1193989-.14G0r.owth, development,
Evans,T.A. 1999. Kinrecognitionin asocialspider. and survival of a generalist predator fed single-
Proceedings ofthe Royal Society ofLondon, Se- and mixed-species diets ofdifferentquality. Oec-
ries B"Biological Science 266:287-292. ologia i ]9:191-197.
Hallander, H. 1970. Prey, cannibalism and micro- Wagner, J.D. 1995. Egg sac inhibits filial cannibal-
habitatselectioninthewolfspidersPardosache- ism in the wolf spider, Schizocosa ocreata. An-
lata O. E Miiller and P. pullata Clerck. Oikos imal Behaviour 50:555-557.
21:337-340. Wagner J.D. & D.H. Wise. 1996. Cannibalism reg-
lida, H. 2003. Small within-clutch variance in spi- ulates densities of young wolf spiders: Evidence
derling body size as a mechanism for avoiding from field and laboratory experiments. Ecology
77:639-652.
pssibeluidnogacnannunliabtaali(sAmrainneateh:e wLoylcfosispdiadee)r. PPaorpduolsaa- Wagner, J.D., M.D. Glover, J.B. Moseley & A.J.
tion Ecology 45:1-6. Moore. 1999. Heritability and fitness conse-
Laneoo, M.J., L. Lowcock&J.P. Bogart, 1989. Sib- quences of cannibalism in Harmonia axyridis.
ling cannibalism in eoncaenibal morph Ambys- Evolutionary Ecology Research 1:375-388.
toma tigrinum larvae and its correlation with
high growth rates and early metamorphosis. Ca- Manuscriptreceived22November2004, revised17
nadian Journal of Zoology 67:1911-1914. June 2005.