ANIMAL BEHAVIOUR, 2002, 63, 857–870 doi:10.1006/anbe.2002.2003, available online at http://www.idealibrary.com on

Value of male remating and functional sterility in redback

MAYDIANNE C. B. ANDRADE & ERIN M. BANTA Department of Neurobiology and Behavior, Cornell University

(Received 20 October 2001; initial acceptance 20 February 2001; final acceptance 7 November 2001; MS. number: A8911R)

In the Australian redback , Latrodectus hasselti, males typically use their paired copulatory organs (palps) to copulate twice with a single female then sacrifice themselves to their cannibalistic mates in a strategy that increases their paternity in that one mating, but leads to death. This type of terminal investment in one mating is predicted only if the expected value of future matings is low for males relative to the value of repeated mating with the same female. In this laboratory study, we quantified the reproductive value of mating more than once with the same female (repeated mating) and mating with more than one female (multiple mating) for male redback spiders. We tested two natural selection hypotheses for repeated mating, sperm limitation and reproductive insurance, but found no support for either hypothesis. We show that, in the absence of sperm competition or cannibalism, male lifetime reproductive output is the same whether a male copulates once, twice, or several times with a given female. Repeated mating does not increase the probability of successful fertilization, nor does it increase the number of offspring produced in successful matings. Although male repeated mating is not favoured because of increased fertility of mates, other studies suggest it may be important in sperm competition. Here we show that the relative reproductive value of the first two copulations is very high for redback males because they are functionally sterile after each palp has been used once; nonvirgin males are unable to father offspring. Functional sterility and repeated mating by male redbacks may be favoured by the same factors that lead to male sacrifice behaviour: ecological constraints on multiple mating combined with competitive benefits of maximal investment in the first mating.  2002 The Association for the Study of Animal Behaviour. Published by Elsevier Science Ltd. All rights reserved.

Current theory easily explains the observation that males expending the same effort in securing new reproductive of most species allocate considerable effort to multiple opportunities (Dewsbury 1982; Parker 1998). Individuals mating (copulations with different individuals, Petrie must decide how many times to copulate with a given et al. 1992), but little to repeated mating (copulations partner before ending the interaction and searching for with the same individual, Hunter et al. 1993; see other partners. In systems without paternal care, the Andersson 1994 for a review). In comparison, male strat- currency of this decision is the number of offspring egies that maximize success with one or a few females but expected as a result of each possible strategy (Trivers rarely involve multiple mating are less well understood. 1972). Thus, ecological or social factors that affect mating The two types of remating are linked because the time success or the reproductive payoff for mating (Emlen & and resources available to males are finite; thus, increased Oring 1977) can determine the degree to which effort at mating repeatedly with one female will usually males engage in repeated mating rather than seeking result in decreased opportunity for multiple mating and copulations with multiple partners. vice versa (Williams 1966; Dewsbury 1982; Simmons The spiders (Araneidae) are a particularly interesting et al. 1992; Smith 1995; Parker 1998; Alonzo & Warner group for examining how the relative value of repeated 1999). At any given stage in a male’s life history, the mating and multiple mating affect observed mating strat- pattern of allocation of reproductive effort (sensu Trivers egies. Male spiders possess paired copulatory organs 1972) between these two forms of remating will be (palps) and female spiders have two independent sperm affected by the fitness benefit of copulating repeatedly storage organs, each with its own insemination opening with the current partner compared to the benefit of (Foelix 1996). However, because sperm from the two spermathecae are released to a common area along with Correspondence and present address: M. C. B. Andrade, Division of Life Sciences, University of Toronto at Scarborough, 1265 Military eggs at fertilization (Foelix 1996), a male that makes only Trail, Scarborough, ON M1C 1A4, Canada (email: mandrade@ one insertion could theoretically fertilize all of his mate’s scar.utoronto.ca). E. M. Banta is now at 147 San Carlos Avenue, eggs. Each individual palp insertion is thus equivalent to El Cerrito, CA 94530, U.S.A. one copulation. While the evolution of independent 857 0003–3472/02/$35.00/0  2002 The Association for the Study of Animal Behaviour. Published by Elsevier Science Ltd. All rights reserved. 858 ANIMAL BEHAVIOUR, 63, 5

spermathecae and copulatory organs is itself a puzzle, it is back males should gain by investing all their resources in also puzzling that species that possess these paired struc- their first mating, and thus mate repeatedly, possibly at tures do not inevitably engage in repeated copulation the cost of lowered fertility in (rare) future matings with (Foelix 1996). There is considerable variation within and other females. This hypothesis of terminal investment in across species in the number of copulations that occur one mating predicts that the reproductive output of during mating (e.g. Cohn 1990; Watson 1990; Prenter redback males will be high in their first mating (consist- et al. 1994; Jackson & Pollard 1997; Knoflach & van ing of two copulations), but decrease in subsequent Harten 2000; Johnson 2001). Causes of this variation matings (if any occur). We tested this hypothesis by have not been identified, perhaps because few data are comparing the reproductive output of males that mate as available on how repeated copulations affect male or virgins to the output of males that mate as nonvirgins. female reproductive success, and what fitness trade-offs Even if this male mating strategy represents terminal might be involved with respect to opportunities or pay- investment in one mating, this does not explain why that offs for multiple mating. one mating involves two copulations. One clear benefit of We studied the reproductive consequences of repeated copulating twice for the male is the insemination of both mating and multiple mating for males of the Australian spermathecae, which would potentially increase total , Latrodectus hasselti. Almost all redback paternity under sperm competition (i.e. repeated mating males attempt two palp insertions during mating, sug- may be sexually selected). However, it is also possible that gesting that repeated mating is beneficial for males repeated copulation is favoured by natural selection. To (Andrade 1998; Forster 1992, 1995). Although most mat- complement our investigation of the value of multiple ings involve two copulations, some males achieve only mating, we examined two natural selection hypotheses one (16.7–30.4%, Andrade 1996, 1998). Quantifying the for repeated copulation: sperm limitation (Arnqvist & relative value of multiple mating is important for under- Henrikkson 1997) and reproductive insurance (Dewsbury standing redback males’ unusual mating strategy, which 1982). usually involves self-sacrifice by the male during sperm The sperm limitation hypothesis proposes that males transfer. Male redbacks always twist 180 during inser- do not transfer enough sperm in one copulation to tion, and place their abdomens directly above the fangs fertilize all the eggs produced by a typical female of their mates (=the copulatory somersault: Cariaso 1967; (Dewsbury 1982; for an overview of possible mechanisms Forster 1992). Many females begin to consume males see MacDiarmid & Butler 1999). Repeated mating could during the first insertion, while males are in this vulner- thus increase male reproductive output by ensuring life- able posture, and while sperm transfer proceeds. Most time fertility of females to which that male has mated. males (83%) survive the first insertion and pull free of The hypothesis predicts that an increase in the number of their mate (Andrade 1998), although females could easily within-pair copulations will increase the lifetime repro- restrain males in this extremely size-dimorphic species ductive output of a male’s mate. This may occur if females (Forster 1992, 1995; Andrade 1998). After a period of that receive many insertions from one male produce additional courtship, males return to the female for a more offspring per reproductive bout, or have more second insertion, during which they again somersault lifetime reproductive bouts than females that receive one onto their mate’s fangs. Thus a ‘normal mating’ for insertion. This hypothesis is supported by studies show- redbacks involves two copulations, and because 65% of ing that females of some spider species require more than males are killed during and after the second insertion, one palp insertion to fertilize all their eggs (Suter & male mating behaviour usually eliminates the oppor- Parkhill 1990; Arnqvist & Henriksson 1997). The sperm tunity for copulations with other females (Cariaso 1967; limitation hypothesis may be particularly relevant to Forster 1992, 1995; Andrade 1996, 1998). spiders because females store sperm and may continue Male redbacks must compete for fertilizations; several producing eggs for long periods after mating (e.g. 1 year males are often present on the web of one female at the or more, Kaston 1970). If some proportion of sperm die in same time, and some females copulate with more than the spermatheca during storage (e.g. in crabs, Paul 1984), one male in their lifetime (Andrade 1996, 1998). Self- a large volume of sperm might be required to fertilize all sacrifice may be adaptive for redback males because can- the eggs produced by a female throughout her life. How- nibalized males can more than double their paternity ever, there are theoretical challenges to this hypothesis relative to males that are not consumed during copula- (e.g. Parker 1998), and tests for sperm limitation have tion (through an increase in copulation duration during produced contradictory results even within one genus cannibalism), and females are less likely to be receptive (e.g. Dolmedes: Arnqvist & Henriksson 1997; Johnson after a cannibalistic mating compared with a noncanni- 2001). Evaluating whether sperm limitation occurs is balistic mating (Andrade 1996). However, these paternity important not only for understanding the factors favour- advantages in a single mating could only lead to adaptive ing repeated mating (for females as well as males), but male complicity in cannibalism if the value of increased also because a correlation between number of palp inser- paternity exceeds the residual reproductive value of the tions and fertility is a key assumption of one of the main male following that mating (Polis 1981; Buskirk et al. models for the evolution of sexual cannibalism (Newman 1984; Elgar 1992; Schneider & Lubin 1998). Male oppor- & Elgar 1991). tunities for multiple mating are apparently limited by The reproductive insurance hypothesis proposes high mortality during mate search (over 80% of males die that individual copulations do not always result in without finding a mate; Andrade 2000), therefore, red- fertilization (e.g. Dewsbury 1978, 1979; Bauer 1992). ANDRADE & BANTA: FUNCTIONAL STERILITY IN SPIDERS 859

Reproductive failures (where no offspring are produced tion occurs prior to the first mating, and then again after after an apparently normal copulation) may occur if each mating. Postcopulatory sperm induction presum- insemination or sperm storage is prevented by chance ably ensures the palps are full of sperm at each ejaculation irregularities in copulatory structures or female manipu- (e.g. Watson 1986, 1990; Suter 1990). lation of sperm (e.g. cryptic female choice, Thornhill In Latrodectus spiders, a small, sclerotized structure at 1983). This hypothesis predicts that the probability of the tip of the male’s palp (the palpal sclerite) breaks off reproductive failure will decrease as the number of and remains inside the female’s genitalia after copulation within-pair copulations increases. This may occur for two (Bhatnagar & Rempel 1962; Andrade 1998; Berendonck & reasons. First, because the genitalia of spiders are often Greven, in press). The function of the sclerite is unknown complex (Eberhard 1985; Foelix 1996), slight morpho- in redbacks, but its loss does not prevent redback males logical irregularities or damage in one copulatory organ from copulating again, nor does it appear to function as a or insemination opening might make insemination on sperm plug in the female (Andrade 1996). that side difficult (e.g. Cohn 1990). Copulating a second time with a second set of genitalia may increase the probability of successful insemination. Second, there is MATERIALS AND METHODS evidence of cryptic female choice based on male copu- Experimental spiders were from a laboratory population latory behaviour in spiders: female spiny orb weavers , of Latrodectus hasselti started and maintained with adults Micrathena gracilis, store more sperm when the male’s collected in and around Perth, Western in second (but not first) copulation is prolonged (Bukowski February 1996, 1997 and 1998. Spiders were shipped to & Christenson 1997). Thus in this species, repeated Cornell University and kept at 279 C, 7010% RH, copulation could increase insemination success. under a 12:12 h light:dark cycle. Experimental spiders We quantified the reproductive value of repeated mat- were reared from egg sacs produced by field-collected ing and multiple mating for redback males in the labora- individuals. Offspring from each egg sac were reared tory to test critical predictions of the sperm limitation, communally with full siblings and fed Drosophila sp. reproductive insurance and terminal investment hypoth- twice per week until the fourth instar, when each spider eses. We complemented the laboratory tests of these was placed in a separate cage to ensure they were virgins hypotheses with (1) an examination of the postcopu- at the time of experimental trials. Once separated, males latory behaviour of surviving males to determine whether were fed Drosophila and the much larger females were fed they engage in preparatory behaviours that would ensure house crickets (Acheta domesticus) twice per week until future fertility (see below) and (2) assessment of the maturity. Spiders were used in mating trials within 2 reproductive output of field-collected females with differ- weeks (males) or 1 month (females) of their final moult. ent mating histories for comparison to laboratory-derived Our laboratory population included 15–20 outbred family data. lines, and we ensured that only unrelated individuals In this study, we concentrated on quantifying the were paired in mating trials. effects of repeated mating on male reproductive output. All experimental spiders were preserved in 95% ethanol Since this necessarily involves assessing the reproductive after death. Because female size or condition might affect output of females that have mated with a single male, our reproductive output (e.g. Andrade 1995; Spence et al. results also provide information on possible benefits of 1996; Uhl 1998), we weighed females prior to trials (on a repeated mating for females (see Petrie et al. 1992; Hunter Mettler balance accurate to 0.1 mg) and measured the et al. 1993). maximum cephalothorax width of preserved females (using an ocular micrometer) for comparison across NATURAL HISTORY treatments. Size was estimated as cephalothorax width2 and condition was calculated as the residuals from a More detailed information about redback natural history regression of weight on size (Jakob et al. 1996). is available elsewhere (Forster 1992, 1995). Redback males mature earlier than females (males: 5 instars in 45 days; females: 7–8 instars in 75 days) and adult males have a Mating Trials shorter life span in the laboratory (a maximum of 8 weeks compared with 2 years for adult females; Forster 1984; We paired males with virgin females in the laboratory Andrade 1996). Mature females generally inhabit one web and manipulated male mating history (virgin or non- for most of their adult lives (M. C. B. Andrade, unpub- virgin) and the number of palp insertions allowed in each lished data), whereas mature males abandon the web on mating (one, two or ad libitum copulation). We quanti- which they develop as juveniles to search for a potential fied male reproductive success after each mating by mate. Prior to leaving their juvenile web, redback males counting the number of offspring produced in the life- fill their copulatory organs (palps) with sperm in a process time of his mate. This allowed us to determine how male called sperm induction. Male spiders have no direct reproductive success was affected by repeated mating in connection between their gonads and their palps so the absence of sperm competition, quantify the expected sperm induction involves building a special sperm web, reproductive output of a nonvirgin male, and test the ejaculating (from the abdominal gonopore) onto the web predictions outlined in Table 1. and drawing the sperm into the palps (Foelix 1996). In Experimental females were chosen randomly from many spiders where males are polygynous, sperm induc- among virgin females, fed three times per week in the 860 ANIMAL BEHAVIOUR, 63, 5

Table 1. Outline of experimental treatments where males of different mating history were allowed a variable number of palp insertions with virgin females to test predictions of three hypotheses about remating Reproductive insurance Sperm limitation hypothesis prediction: Terminal investment Number Male hypothesis prediction: proportion of hypothesis prediction: Treatment N* of inserts status† reproductive output reproductive failures reproductive output

One-palp First 15 1 V Lowest Highest High Second 15 1 V Two-palps 10 2 V High Low High Two-palps+‡ 11 2+ V High Lowest High One palp+‡ 11 1+ NV — — Lowest

*Total number of mating trials included in data analysis. Trials were excluded if females acted aggressively towards males during courtship (Andrade 1996) or if males were damaged or killed by females during trials (see Methods). †At the start of the trial, V=virgin (at least one unused palp); NV=nonvirgin (each palp already used once). ‡In ‘+’ treatments, males were left on the female web for 1–4 days after the first copulation.

month prior to the trials, then offered one cricket each 3. Two+ palp treatment day in the 2 days preceding the trial. These additional Virgin males (N=11) were observed while they achieved feedings reduced the likelihood of cannibalism by two palp insertions (one with each palp), then the pair improving female condition (Andrade 1998). Matings was left together for 4 days or until the death of the male were staged in arenas (353015 cm) containing (most males died within 4 days of their first mating). V-shaped wooden frames for female webs (Kavale 1986). Spiders in this treatment would be likely to have a higher Females were placed in arenas 2–4 days prior to a trial and number of insertions than pairs in the two-palp treat- always produced webs in this time. For trials, males were ment (during periodic checks after the first two placed on female webs, observed while they achieved the copulations, males were often observed courting and intended number of insertions, and then removed from sometimes mating with females). This ad libitum treat- the web to terminate the trial. Males were virgins at the ment also allowed ample time for males to go through start of treatments 1–3, which varied in the number of postcopulatory sperm induction prior to additional palp insertions permitted to each male. In treatment 4, repeated matings. We included ad libitum treatments in matings were between virgin females and males that had our analysis only if the male survived for least 24 h after already copulated twice. the observed mating.

1. One-palp treatment 4. Nonvirgin male treatment Virgin males (N=30) were allowed a single palp inser- Nonvirgin males had previously been allowed to use tion. Because males have two palps, and the first and each palp once in two separate copulations, each with a second insertion have been shown to have different different virgin female. Then, for the fourth treatment, reproductive consequences in some spiders (Suter 1990; each nonvirgin male (N=11) was placed on the web of a Bukowski & Christenson 1997), we split this treatment third virgin female and observed until at least one palp into two subgroups to compare the effect of an insertion insertion occurred. The pair was then left together for 1–4 from the first palp to the effect of an insertion from the days to provide ad libitum mating access, as in the two+ second palp used by a male. palp treatment. We terminated half of the trials (N=15) after the male We excluded trials from our analysis based on the achieved his first palp insertion, and assessed male repro- following criteria: criterion 1: mating did not occur ductive success from the reproductive output of this first within 5 h of the introduction of a courting male (Forster mate (first-palp matings). The remaining 15 males were 1995 reports an average courtship duration of  transferred to the web of a second virgin female immedi- 5.03 0.84 h); criterion 2: females showed repeated rejec- ately after their first copulation. These males were permit- tion behaviours during courtship (such as repeatedly ted a single palp insertion with these new females striking at the male, Andrade 1996); or criterion 3: a male (second-palp matings, N=15). Second-palp matings were was consumed or injured (i.e. integument punctured) always completed with the previously unused (virgin) prior to or during the assigned number of palp insertions palp. We assessed mating success of second-palp males (assessed by visual examination). Although well-fed based only on the reproductive output of their second females are less likely to kill males during copulation mate. (Andrade 1998), many matings still involve some female- caused injury to the male, which disqualified trials from our consideration. 2. Two-palp treatment Our criteria ensured that data were included only (a) Virgin males (N=10) were allowed to achieve two palp when there was no evidence of female precopulatory insertions with one female (always one with each palp). discrimination against males (criterion 1, criterion 2), and ANDRADE & BANTA: FUNCTIONAL STERILITY IN SPIDERS 861

(b) when treatment males were undamaged by the female 1995, and January and February 1996. Whenever a web when they copulated (criterion 3). In addition, criterion 3 was found intact with the remains of the dead female, its ensured that any treatment effect we might see was not refuge was cut free and collected. We also collected due to effects of cannibalism. Our strict criteria for data refuges from marked webs from which the female had inclusion reduced sample sizes by 40–60% in each treat- disappeared (and was assumed to have been depredated). ment, so the data reported here were accumulated over 3 Refuges were dissected in the laboratory and the number years (1996, 1997, 1998). of egg sacs in each was counted. Because we did not know the mating history of the females whose webs were marked, we include only data for those webs that con- Reproductive Output in the Laboratory tained at least one egg sac (N=27). Although this ensures that females in this sample were mated prior to death, it We quantified male lifetime reproductive output as the may overestimate the expected number of egg sacs pro- output of his mate in her postmating lifetime (number of duced per mated female since it ignores cases where egg sacs and spiderlings produced). Following trials, mated females died prior to oviposition. mated females were placed in separate cages and fed two adult crickets (Acheta domesticus) each week. Egg sacs were removed as they were produced and placed in individual Reproductive Failure in the Laboratory cages until hatching, when observers who were blind with respect to treatment counted the spiderlings that In addition to quantifying total reproductive output, emerged. Direct counts were obtained by freezing groups we noted the number of experimental matings that ended of spiderlings shortly after they hatched and counting in complete reproductive failure, where no spiderlings them under a dissecting microscope. To avoid euthaniz- were produced after an apparently normal copulation (i.e. ing large numbers of spiders, we also counted some an insertion with a full copulatory somersault, Forster spiderlings by briefly chilling them, videotaping the 1992). groups through the clear cage walls, then counting spid- erlings in a paused video image on a monitor with a grid overlay. All videotaped spiderlings were counted twice (at Reproductive Failure in the Field least 1 week apart, using the same video images), and the We estimated the frequency of reproductive failures as rounded average of the two counts was used in data a function of the number of virgin copulations for field- analysis. Both counting methods were used throughout mated females. Sexually mature females were collected in the experiment (a coin toss determined which was and around Perth, Western Australia in February 1997, assigned to each egg sac), and they yielded comparable and in February–March 1998, then shipped to Cornell results. Counts of 25 randomly chosen egg sacs using University, and kept as described above. Egg sacs were each method were highly correlated (r2=0.979, P<0.001, removed as they were produced, and we noted whether Y intercept= 1.098, slope=1.010) and the 99% confi- spiderlings emerged from those sacs. Although we did not dence interval of a regression line relating the two counts know female mating histories when they were collected, for each sac overlapped a line with slope of 1 and a Y we confirmed that females had mated in the field by intercept of 0. dissecting their spermathecae after death and looking for palpal sclerites. We considered the number of sclerites Reproductive Output in the Field found in a female’s genitalia to be equal to the number of times that female had received an insertion from a virgin, Any changes in reproductive output we saw over time or previously unused, palp (a reasonable assumption, see in the laboratory would only be relevant to selection if Andrade 1996, 1998). However, we were not able to females would normally live long enough to experience determine the number of different males with which each the change in nature. For comparison to laboratory female had mated. For example, a female with two scler- results, we estimated the number of egg sacs produced by ites may have mated with one or two males, a female with females in the field by counting sacs in the webs of adult three sclerites may have mated with two or three males. females after their death. Females mainly produce their We include these data because the reproductive insurance egg sacs in a protected refuge portion of their webs (Main hypothesis predicts that, for females, the frequency of 1980). The refuge is likely to contain all the sacs produced reproductive failure should decrease with increases in the by a female in her lifetime because the outer covering number of insertions received, regardless of how many of an egg sac is a tightly woven shell (Kaston 1970; different mates are involved. Foelix 1996) that remains intact for years after eggs hatch Field-collected females were considered to have experi- (M. C. B. Andrade, personal observation), and most enced reproductive failure if their genitalia contained at female redbacks remain in the same web throughout their least one sclerite but they did not produce any viable reproductive lives (M. C. B. Andrade, unpublished data). offspring in the laboratory. We considered only females In early January 1995, we marked the webs of 40 adult that survived at least 3 months in captivity (N=46) to females of unknown age at a variety of sites in and around ensure they had ample time to produce an egg sac (mated Perth, Western Australia. The spiders were left undis- females held on a laboratory diet generally produce one turbed, and webs that still contained adult females in egg sac each month, M. C. B. Andrade, unpublished data), early December 1995 were revisited in late December and that they were not senescent (females may cease 862 ANIMAL BEHAVIOUR, 63, 5

Table 2. Frequency of complete reproductive failure as a function of number of insertions received from virgin males for laboratory, and field-mated females Laboratory (5.9%)* Field (15.2%)* Number of inserts Number of sclerites†

1 2 2+ Group (T1) (T2) (T3) 1 2 >2

N 30 10 11 9 29 8 Number of failures 2 1 0 1 5 1 Percentage of failures 6.7 10 0 11.1 17.2 12.5

T: Treatment. *Percentage of failures is given in parentheses. †Number of sclerites is equivalent to the number of insertions received from virgin palps (but not necessarily the number of different males that mated with that female).

viable egg production as they senesce, see Results). We RESULTS did not compare the total reproductive output of these females as a function of number of insertions because First versus Second Insertion they may have produced egg sacs in the field prior to Male courtship and mating behaviour appeared similar collection. to that previously observed in the field and laboratory (Forster 1992, 1995; Andrade 1996, 1998). Within the Male Postcopulatory Behaviour one-palp treatment (treatment 1) the lifetime reproduc- tive output of females that received an insertion from a We made systematic efforts to observe sperm induction male’s first palp (number of egg sacs =11.471.9; behaviour in nonvirgin males. First, we observed the number of spiderlings=2053.3373.8, N=15), did not behaviour of nonvirgin males in treatment 4 after trans- differ from that of females that received an insertion from ferring them to webs of experimental females. Second, we a male’s second palp (number of egg sacs=11.531.7:  looked for postcopulatory sperm induction in a separate Student’s t test: t28= 0.026, P=0.980; number of spider-  sample of laboratory-reared males (N=21) that were per- lings=1818.0 267.1; t28=0.512, P=0.613) and copu- mitted two complete copulations with well-fed, virgin lation duration was similar for the first and second palp females. These males were returned to their cages and (1) (first insertion: 14.151.7 min., N=13; second insertion:   videotaped during the 6 h immediately following copu- 14.64 1.5 min; t28= 0.221, P=0.827). Since there was lation (N=7), or (2) videotaped during the 7–13 h after no evidence of different reproductive consequences from copulation (N=7), or (3) left in their cages for 15–20 h first compared to second insertions, we pooled the data after copulation, then placed on a web with a virgin from these two subgroups of the one-palp treatment for female, where their activity was videotaped for 6 h (N=7). the remainder of our analyses. All videotapes were reviewed for sperm induction behav- iour, and webs were examined for evidence of a sperm web. Repeated Mating and Reproductive Failure To test the reproductive insurance hypothesis, we com- Analyses pared the frequency of reproductive failures as a function of the number of insertions received from virgin palps. We completed all statistical tests using SYSTAT 8.0 Three of the 51 pairs in treatments 1–3 (virgin males) did (Wilkinson et al. 1997) with consultation from Zar not produce viable offspring despite apparently normal (1984). Data are reported as meansstandard error. Para- copulations that included a full copulatory somersault metric tests (t tests, analyses of variance, ANOVA, and (5.6% failure). However, there was no support for the covariance, ANCOVA) were used if sample sizes were reproductive insurance hypothesis because the pro- greater than 10 and if the Lilliefors test indicated stand- portion of failures showed no predictable association ardized data did not deviate significantly from normality with the number of palp insertions received (Table 2). (P>0.05); otherwise, nonparametric tests (Mann–Whitney Because of small sample sizes, we pooled data to compare U, Kruskal–Wallis) were used. Regression analysis was the frequency of failures in females that received two or used to determine whether female size or condition was more insertions (treatments 2 and 3) to females that correlated with the number of egg sacs or spiderlings received only one insertion (treatment 1) using Fisher’s produced. Where significant correlations were found, exact tests. Field-mated females experienced a higher female traits were used as covariates in ANCOVA to frequency of reproductive failure than laboratory-mated determine the effects of male palp treatments. If ANOVA females (15.2% did not produce viable offspring) but this or ANCOVA showed significant differences among treat- difference was not significant (Table 2; Fisher’s exact test: ments, Bonferroni post hoc tests were used for pairwise P=0.184). As for laboratory-mated females, there was no comparisons. detectable relationship between the number of insertions ANDRADE & BANTA: FUNCTIONAL STERILITY IN SPIDERS 863

field-mated females received from virgin palps (i.e. the 30 number of palpal sclerites found) and the frequency of (a) reproductive failure (Table 2, data for two or more scler- a ites pooled; Fisher’s exact test: P=1.00). Overall, there was a no evidence that females that copulated once were less a,b likely to produce young than females that copulated 20 more than once, even when the field and laboratory data were considered together (Fisher’s exact test: P=0.737).

Repeated Mating and Lifetime Reproductive 10 b Output Number of egg sacs Contrary to the sperm limitation hypothesis, male lifetime reproductive output did not depend on whether one, two or more than two insertions were achieved with 0 the same female (Fig. 1, comparison of treatments 1, 2 and 3; ANOVA: number of egg sacs: F =1.307, P=0.280; 2,48 1234 number of spiderlings: F =0.939, P=0.398). 2,48 Treatment In spiders, variation in the number of offspring pro- duced is often related to variation in female size or (b) 5000 condition. In this experiment, although female size was a positively correlated with the average number of spider- 2 a lings per sac (treatments 1–3: R =0.267, F1,21=7.654, 4000 a P=0.012), there was no correlation between female size and lifetime reproductive output (treatments 1–3: 2 3000 number of egg sacs: R =0.01, F1,21=0.20, P=0.66; number 2 of spiderlings: R =0.08, F1,21=1.93, P=0.18). However, females in better condition prior to the mating trial 2000 produced more egg sacs in their lifetime (R2=0.241, b F1,18=5.708, P=0.028) and tended to produce more spid- 1000 2 erlings (R =0.206, F1,18=4.684, P=0.044) than females in Number of spiderlings poorer condition. Even when female condition was included as a covariate, there was no detectable effect of 0 number of palp insertions on total reproductive output (treatments 1, 2 and 3: ANCOVA: treatment effect, 1234 number of egg sacs: F2,16=0.642, P=0.539; number of Treatment spiderlings: F2,16=0.064, P=0.938). The sperm limitation hypothesis might also predict Figure 1. Lifetime reproductive output of females in four treatments that a male’s mate will exhaust her sperm stores sooner if in terms of total number of egg sacs produced (a) and total number she receives fewer palp insertions. However the pattern of of viable spiderlings (b). Here and in other figures, box plots show female reproductive output over time was similar across means (filled square), medians (centre horizontal line), 25th and virgin male treatments, regardless of whether one or more 75th percentile (top and bottom of box) and 5th and 95th percen- than one palp insertion was received (Fig. 2a, b). After a tile (tip of each vertical line). In both plots, different letters above boxes indicate significant differences between means, at P<0.05 virgin copulation, males could expect their mates to be (Bonferroni post hoc tests). reproductively active for 80–90% of their postmating lives. We compared the time course of female reproduc- tive output across experimental treatments, while con- stores relative to egg availability; egg sac yields began to trolling for differences in female longevity, by comparing decline below average only after females had produced the relative reproductive life span of each female (number approximately 15 egg sacs, at which point more than 60% of days over which egg sacs containing viable eggs were of experimental females had died (all virgin male treat- produced divided by the total number of days the female ments pooled, Fig. 2c). Females that survive past this survived following copulation). We pooled data from point may be senescing, with an accompanying decrease treatments 2 and 3 to increase power for this test and in reproductive output. Moreover, mated females in our compared treatment 1 (one-palp) to the pooled treatment field sample produced an average of only six egg sacs (see (two-palp, two+ palps). There was no difference in the horizontal line above abscissa, Fig. 2c), so it appears that relative reproductive life span of females that received wild males would be unlikely to experience much of a only one insertion from a male (0.8880.023, N=22) decrease in reproductive output per sac during the compared to those that received two or more insertions lifetime of their mates. (0.8780.029, N=14, Mann–Whitney U test: U=160.00, If sperm limitation did affect reproductive success, and P=0.845). Although reproductive output did decline over longer copulations resulted in the transfer of more sperm time (Fig. 2), this is unlikely to be due to limited sperm (e.g. Andrade 1996), we would have expected a positive 864 ANIMAL BEHAVIOUR, 63, 5

4 (a) 400 b a a,b 3 a,b 300

2 200

1 100

0 0 Female postmating longevity (days)

0 5 10 15 20 25 4 1234 (b) Treatment Figure 3. Female postmating longevity in four mating treatments 3 (different letters above boxes indicate significant differences at P<0.05, Bonferroni post hoc tests).

2 relationship between copulation duration and the number of spiderlings produced. This was the not the 2 case for matings by virgin males (R =0.011, F1,29=0.309, 1 * P=0.582). * * * * * * * * * * * * * * * Multiple Mating and Male Functional Sterility 0 * * * * As predicted by the terminal investment hypothesis, the mates of non virgin males (i.e. males with each palp 0 5 10 15 20 25 4 previously used once) produced substantially fewer egg (c) sacs and spiderlings than the mates of virgin males (Fig. 1; Number of spiderlings/average number per sac (within females) 1.0 ANCOVA with female condition as covariate, treatment 0.9 effect on: number of egg sacs: F =7.409, P=0.001; 3 0.8 3,24 number of spiderlings: F =8.539, P<0.001). Ten of the 0.7 3,24 11 males in treatment 4 (90.9%) were sterile and fathered 0.6 2 no offspring when they mated as nonvirgins. This is a 0.5 much higher frequency of reproductive failure than 0.4 occurred after laboratory matings with virgin males 1 0.3 (Table 2, treatments 1–3 pooled; Fisher’s exact test: 0.2 P<0.001). The lower siring success of non virgin males 0.1 occurred in spite of significantly greater post mating 0 0.0 longevity of their mates compared to treatment 1 females Proportion of females reproducing * * (Fig. 3; ANOVA: F3,53=3.215, P=0.030; Bonferroni post 0 5 10 15 20 25 hoc test: treatment 1 versus treatment 4: P=0.037). Egg sac number (in chronological order) These results are unlikely to be the result of differences Figure 2. Time course of offspring production for the mates of virgin in the mating behaviour of nonvirgin males (courtship males (treatments 1, 2 and 3, N=42) in terms of relative yield from and copulatory behaviour of nonvirgin males was similar each egg sac in chronological order. Relative egg sac yield equals the to that of virgins). Nonvirgin male courtship duration did number of spiderlings in each sac divided by the average number of not differ from that of virgin males prior to their second spiderlings per sac over the lifetime of each female (horizontal virgin insertion, although courtship duration of virgin dashed line shows relative yield of 1). Change in output was similar males prior to their first insertion was the longest of any for females that received a single insertion (a) or two or more group of males (Fig. 4a; ANOVA: F2,40=39.150, P<0.001; insertions (b). In (c), the time course of reproduction is pooled for all Bonferroni post hoc tests: virgin 1 versus virgin 2: females mated to virgin males (treatments 1–3) and female survivor- P<0.001; virgin 1 versus nonvirgin: P<0.001; virgin 2 ship is the dotted line (right axis). The horizontal line just above the versus nonvirgin: P=0.601). Copulation duration, which abscissa in (c) shows the mean (black circle), 95% confidence interval (whiskers), and two extreme values (asterisks) for the is correlated with paternity under sperm competition number of egg sacs produced by females in the field (N=27). (Andrade 1996), was the same for virgins and nonvirgins (Fig. 4b; ANOVA: F2,42=0.242, P=0.786). ANDRADE & BANTA: FUNCTIONAL STERILITY IN SPIDERS 865

400 4000 (a) Within treatment 4

a 300 3000

b 2000 200

1000 100 Number of spiderlings b Courtship duration (min) 0 0 1 2 3 Insertion number 30 Figure 5. Reproductive output of each of the three females with (b) whom males in treatment 4 mated, in order (dotted line is the single male that fathered offspring in the third insertion).

20 comparison, the mate of every other male in this exper- iment (N=61) contained one sclerite for each virgin insertion they had received. We infer that the first insertion by the anomalous male was not a successful copulation. 10

Male Postmating Behaviour

Copulation duration (min) Although we observed males grooming their palps after 0 copulation, males were not observed to undergo postcopulatory sperm induction (following two virgin insertions) during our experiment (N=11), nor in any of First palp Second palp Nonvirgin our videos of males following unmanipulated copulations virgin virgin (N=17, 4 videos were excluded because the image was not Figure 4. (a) Courtship duration was similar for nonvirgin males and in focus for more than 5 consecutive minutes). We also virgin males prior to their second insertion, but longer prior to the saw no evidence of new sperm webs in male cages or on first virgin insertion (different letters above boxes indicate significant the webs of females. differences at P<0.05; Bonferroni post hoc tests). (b) Copulation duration was similar regardless of male mating experience. DISCUSSION Similarly, male postmating sterility cannot be Sexual Selection for Repeated Mating? explained by a lower overall fertility of males assigned to treatment 4. Figure 5 shows the reproductive output of Repeated mating by male redback spiders does not the three females to whom each nonvirgin male was increase the lifetime reproductive output of males by mated in sequence (where the third mating was the increasing the output of their mates. Our results demon- experimental mating in treatment 4 for which data are strate that redback males transfer sufficient sperm during reported above). Male reproductive output was high in a single insertion to fertilize all the eggs produced in the the first two matings, in each of which one virgin palp lifetime of a female. Although reproductive output per was used, but zero for males on the third, nonvirgin egg sac does decline over time, the relative reproductive mating (repeated measures ANOVA: F2,14=7.895, lifetime of females is unrelated to the number of inser- P=0.005). The single male that did not show this pattern tions received from their mates, and their reproductive of sterility in the third copulation did not produce any decline is more likely to be due to senescence than to offspring as a result of his first copulation (Fig. 5, dashed sperm depletion. It is unlikely that mated females would line). The duration of the first copulation for this anoma- ever experience sperm depletion in nature since they lous male was 49 s, much shorter than average (average produce an average of only six egg sacs in their lifetime copulation duration for males in treatment 1 was and our laboratory results suggest that reproductive out- 14.4 min; 95% confidence limits=12.2–16.7 min, N=27). put remains high (above average) for at least the first 15 Dissections of females after their deaths showed that this egg sacs produced after one copulation (Fig. 2). male’s first mate did not contain a palpal sclerite, whereas Based on theoretical arguments and models, Parker the second and third mates did contain sclerites. In (1998) predicted that female reproduction would rarely 866 ANIMAL BEHAVIOUR, 63, 5

be limited by sperm stores because there would be strong than one male in their lifetime (Table 2 in Andrade 1996), selection on males to ejaculate enough sperm in a single and as we show here, females can store viable sperm for copulation to fertilize all a female’s eggs. Nevertheless, long periods. Moreover, because over 80% of males die there have been empirical demonstrations of sperm limi- during mate search, males that attempt a multiple mating tation after a single copulation in species with direct strategy are unlikely to be successful (Andrade 2000). gamete transfer (e.g. Suter & Parkhill 1990; Arnqvist & Under these circumstances, copulating twice with a single Henriksson 1997; MacDiarmid & Butler 1999). These may female can be the best strategy, especially if (1) repeated represent cases where there are strong energetic con- mating increases paternity under sperm competition (e.g. straints on male sperm production or transfer (Dewsbury Schneider et al. 2000), or (2) females are less likely to 1982; Parker 1998), or where optimal sperm allocation remate after a two-insertion mating compared to a one- decisions result in low sperm transfer in some matings insertion mating. Both of these criteria may hold for (Warner et al. 1995; Gage & Barnard 1996; MacDiarmid & redback males. First, repeated mating effectively doubles Butler 1999). Although few empirical studies have exam- male copulation duration, which could double paternity ined this hypothesis directly, our results support the in each mating relative to a single-insertion strategy conclusion that sperm limitation is unlikely to be a (Andrade 1996; see Schneider et al. 2000). Moreover, as good general explanation for repeated mating (Lake 1975; with all spiders, males that achieve only one insertion Jackson 1980; Birkhead et al. 1989; Birkhead & Møller will leave one entire spermatheca empty, and this may 1992; Hunter et al. 1993). reduce paternity by at least 50% if females mate with We found no evidence that repeated mating provides another male. Second, although the relative risk of female reproductive insurance for males or females. Normal remating has not been measured directly, most field- copulations between virgin males and females did some- caught redback females have had two insertions by the times fail, but the cause of these failures is unclear and end of the mating season, suggesting they prefer two their frequency did not vary with the number of inser- insertions to one (females control mating access by males, tions achieved. This was true for field- and laboratory- Andrade 1996). In addition, there is evidence that female mated females. Although it seems likely that spiders receptivity varies as a result of mating outcomes: 96% of that developed in nature would be more likely to have females remate after brief, noncannibalistic copulations abnormalities in their copulatory organs than would but only 33% remate after longer, cannibalistic copu- laboratory-reared spiders (due to environmental stresses lations (Andrade 1996). Although this study (Fig. 1) or accidents), this did not appear to increase the impor- shows that females do not need two insertions to fertilize tance of repeated mating for successful insemination. all their eggs, there are many other reasons for females to However, it is possible that wild females facultatively mate with more than one male (Petrie et al. 1992; Hunter remated with new males if their first mate did not et al. 1993). Quantification of female remating rates and achieve insemination. This type of female strategy would male paternity as a function of insertion number is have made it difficult for us to detect an effect of number necessary to determine the importance of repeated of insertions from a single male on the risk of reproduc- mating to male paternity. tive failures. This is because counting palpal sclerites allowed us to determine only the maximum possible number of males with which a field-captured female had Multiple Mating and Male Functional Sterility mated. It is possible that subtle effects of the number of Maximizing reproductive output as a result of virgin copulations may play a role in this system, but such insertions is important for redback males because we have effects would only be detectable with larger sample sizes. shown that they are effectively sterile after each palp has Even subtle effects could select for repeated mating been used once (Fig. 1). Because redbacks typically copu- attempts by males because the cost of failure is high. late twice with their first mate, mating with more than Thus, although our data do not suggest support for the one female does not increase male fitness. This extreme reproductive insurance hypothesis, it would be interest- result was predicted by the terminal investment hypoth- ing to see further tests of this idea. This is particularly true esis but would not be predicted by most other hypotheses because there were a number of copulations in the field for male remating strategies (e.g. Andersson 1994). Thus and laboratory that did not result in offspring (Table 2), our study supports the idea that there is selection on and the cause of these failures is unclear. redback males to invest maximally in their first mating, These results also highlight an intriguing question and little selection to maintain fecundity after that first about female mating strategies: why do females permit mating. Such selection pressure could arise from stringent males to achieve two insertions? It appears unlikely that constraints on multiple mating (Andrade 1996). female fertility is affected by repeated mating, but there are a number of other hypotheses for variation in repeated mating that would be interesting to investigate Mechanistic Causes of Functional Sterility (Petrie et al. 1992; Hunter et al. 1993). Redback males may attempt to mate repeatedly to Male functional sterility may arise from the morpho- minimize paternity losses due to sperm competition in an logical damage sustained by the palps during mating environment where searching for another mate is very (Bhatnagar & Rempel 1962; Forster 1992; Andrade 1996; risky. At least 12% of redback females mate with more Berendonck & Greven, in press), from sperm depletion in ANDRADE & BANTA: FUNCTIONAL STERILITY IN SPIDERS 867 the palps (Christenson 1989), or from some combination Male Terminal Investment Strategies of these factors. The loss of the palpal sclerite may be the mechanistic Sterility is often the maladaptive result of a pathology. cost of anchoring the palp inside the female sperm But if all males in a species become functionally sterile storage organ during ejaculation and could result in after mating once, this may reflect a ‘terminal invest- sterility if the damaged organ is unable to properly ment’ strategy that entails skewing allocation of repro- engage the female genitalia (Bhatnagar & Rempel 1962; ductive effort to one or a few copulations. The redback Kaston 1970). Our results provide some support for this male’s strategy involves repeated mating, sterility and hypothesis: the one nonvirgin male that was able to self-sacrifice, all of which may arise because female repro- fertilize a female successfully had not lost the sclerite in a ductive biology rewards high male investment in one previous mating (see Results and Fig. 5), although all mating and because ecological constraints restrict the other successful matings in this experiment involved the number of viable strategies available to males (Forster loss of a sclerite. Male Nephila clavipes spiders also lose a 1992; Andrade 1996). Although this is an unusual system portion of the palp during copulation (Cohn 1990; Foelix in many ways, there is evidence for similar terminal 1996) and Christenson (1989) concludes that these males investment strategies in a variety of species. Male Tidarren are functionally sterile after their first mating. However, spiders intentionally amputate one palp prior to sexual for Nephila, it is not clear whether palpal damage occurs maturity, are always cannibalized by the female after a in every mating. Interpretation of Christenson’s study is single copulation, and sometimes die before cannibalism problematic because nonvirgin males did not show nor- begins (Branch 1942; Knoflach & van Harten 2000), all of mal copulatory behaviour, and were quite old when which suggest a strategy of maximal investment in one mated (near to the maximum life span in the laboratory mating. Similarly, males of a cannibalistic orb-weaving and past the typical life span in the field, Christenson spider, Argiope aemula, die during their second copulation 1989). with a single female and are frequently consumed by Berendonck & Greven (in press) argue that morphologi- their mates after their death (Sasaki & Iwahashi 1995). In cal damage to the male palp during copulation in these cases, male reproductive effort may partly function L. rivivensis would make sperm reinduction impossible. as paternal effort and add to the benefit of terminal However, in the only other study of male fecundity in investment. However, cannibalism is not a prerequisite Latrodectus, Breene & Sweet (1985) report that nonvirgin for these investment strategies. Male lovebugs (Diptera) L. mactans males are able to fertilize eggs even after the copulate with one female for 50 h and frequently die loss of the palpal sclerite in earlier matings. These contra- while in copula (Hetrick 1970; Thornhill 1976, 1980; dictory results suggest that palpal damage alone may not Hieber & Cohen 1983) and male honeybees die after a explain male sterility in every Latrodectus species. single copulation, which follows an intense competition Another possible explanation for functional sterility is with large numbers of other males for few females that males become sperm depleted because they ejaculate (Thornhill & Alcock 1983). Among mammals, although all available sperm in their first mating and do not refill the number of matings achieved is greater, the general the palps afterwards. In many polygynous spider species, pattern of male investment is similar in Antechinus stuartii males refill their palps with sperm either following (a marsupial mouse). Male A. stuartii expend so much each mating, or immediately preceding each mating energy securing few matings during a single breeding (Robinson & Robinson 1980; Suter 1990; Watson 1991; season that they experience physiological breakdown and Watson & Lighton 1994; Costa 1998), suggesting that senescence by the end of that season (Cockburn 1994; Lee palps are emptied during copulation. Although males & Cockburn 1985). In marine organisms such as deep-sea may modify the amount of sperm ejaculated as a function anglerfish (e.g. Ceratias holbolli, Edriolychnus schmidti), of female reproductive value (Cohn 1990; Bukowski & females are sedentary and difficult to find. In all these Christenson 1997), they are likely to use all the sperm in groups males are essentially reduced to testes that attach each palp when mating with virgin females (Christenson themselves to a single female, and thus inseminate only 1989; Cohn 1990). In our experiment, we did not observe one female (reviewed in Vollrath 1998). These examples male sperm induction following the first mating or pre- suggest that terminal investment strategies may generally ceding subsequent matings with virgin females. This be seen when males rarely have the opportunity to mate suggests that sperm exhaustion in the palps, coupled with with several females (Vollrath & Parker 1992), and/or the absence of postmating sperm induction behaviour, when males have a low likelihood of mating successfully may be the mechanistic cause of male sterility. If this is is any given attempt. the case, the single male that fertilized eggs in a nonvirgin In comparison, despite the cannibalistic reputation of copulation (Fig. 5, dashed line) may not have had suf- the female (see references in Johns & Maxwell 1997), ficient time to ejaculate sperm during his first, brief male praying mantids apparently spread their reproduc- copulation. He would then have had ample sperm tive effort over several mating attempts, and have numer- remaining to fertilize a third female. ous adaptations for avoiding cannibalism (reviewed in Testing these mechanistic hypotheses for male post- Maxwell 1999). In this case, although the average male mating sterility will require detailed manipulations of may have a low encounter rate with females, some males male mating behaviour and morphology, coupled with are able to achieve several copulations (Lawrence 1992; quantitative estimates of sperm transfer (e.g. Bukowski & Maxwell 1998), suggesting that constraints on multiple Christenson 1997). mating are not severe. This variance in mating success 868 ANIMAL BEHAVIOUR, 63, 5

may favour a strategy that includes multiple mating Andrade, M. C .B. 1995. Sexual cannibalism in the redback spider (Rubenstein 1985). (Latrodectus hasselti Thorell): natural selection for female feeding For animals without parental care of young, there are and sexual selection for male sacrifice. M.Sc. thesis, University of few theoretical considerations of the conditions under Toronto. Andrade, M. C. B. 1996. Sexual selection for male sacrifice in the which male mating strategies will include heightened Australian redback spider. Science, 271, 70–72. investment in one mating at the cost of a decreased Andrade, M. C. B. 1998. Female hunger can explain variation in probability of multiple mating. While general insights cannibalistic behaviour despite male sacrifice in redback spiders. from models for the evolution of semelparity and itero- Behavioral Ecology, 9, 33–42. parity could be useful here, such models generally con- Andrade, M. C. B. 2000. Sexual selection and male mating behavior sider reproductive allocation decisions of females only in a cannibalistic spider. Ph.D. thesis, Cornell University, Ithaca, (Roff 1992). In species like redbacks, where males rarely New York. engage in multiple mating due to ecological constraints Arnqvist, G. & Henriksson, S. 1997. Sexual cannibalism in the (Andrade 2000), it may be fruitful to consider the possi- fishing spider and a model for the evolution of sexual cannibalism bility that the male strategy has been tuned by selection based on genetic constraints. Evolutionary Ecology, 11, 253–271. Bauer, R. T. 1992. Repetitive copulation and variable success of to favour maximal investment in few matings (e.g. insemination in the marine shrimp Sicyonia dorsalis (Decapoda: Buskirk et al. 1984; Johns & Maxwell 1997), despite Penaeoidea). Journal of Crustacean Biology, 12, 153–160. intuition to the contrary (Gould 1984). Understanding Berendonck, B. & Greven, H. In press. Morphology of female and the relative importance of the opportunity for multiple male genitalia of Latrodectus revivensis Shulov, 1948 (Araneae, mating, female-determined reproductive payoffs, and ) with regard to sperm priority patterns. In: European intermale competition for male terminal investment Arachnology 2000 (Ed. by S. Toft & N. Scharff). Aarhus: Aarhus strategies will require comparative analyses of represen- University Press. tative systems across taxa, combined with models that Bhatnagar, R. D. S. & Rempel, J. G. 1962. The structure, function, include the fitness effects of reproductive allocation and postembryonic development of the male and female copula- decisions for males. tory organs of the black widow spider Latrodectus curacaviensis (Muller). Canadian Journal of Zoology, 40, 465–510. Birkhead, T. R. & Møller, A. P. 1992. Sperm Competition in Acknowledgments Birds: Evolutionary Causes and Consequences. San Diego: Academic Press. We are grateful for support from the Animal Behavior Birkhead, T. R., Hunter, F. M. & Pellatt, J. E. 1989. Sperm Society, NSERC (1967 Science and Technology Scholar- competition in the zebra finch, Taeniopygia guttata. Animal Behav- ship to M.C.B.A.), NSF (Dissertation Improvement Grant), iour, 38, 935–950. the Marion and Percy A. Leon Fellowship Fund, the Olin Branch, J. H. 1942. A spider which amputates one of its palpi. Foundation, the President’s Council of Cornell Women Bulletin of the South California Academy of Science, 41, 139–140. Breene, R. G. & Sweet, M. H. 1985. Evidence of insemination of and Sigma Xi (Grant-in-Aid of research). We thank D. multiple females by the male black widow spider, Latrodectus Cook, F. Berlandier, and the crew at ‘blowies’ (Agriculture mactans (Araneae, Theridiidae). Journal of Arachnology, 13, 331– Western Australia), W. Bailey, D. Edwards and especially 335. I. Dadour (University of Western Australia) for help with Bukowski, T. C. & Christenson, T. E. 1997. Determinants of sperm permits and/or for research space and equipment in release and storage in a spiny orb weaving spider. Animal Behav- Australia. We are grateful for help rearing spiders in the iour, 53, 381–395. laboratory (L. Cronin, D. Handman, D. Malogna and K. Buskirk, R. E., Frohlich, C. & Ross, K. G. 1984. The natural selection McMunn) and for research space or equipment at Cornell of sexual cannibalism. American Naturalist, 123, 612–624. University provided by K. Adler, R. Booker, T. D. Seeley Cariaso, B. J. 1967. Biology of the black widow spider, Latrodectus and R. R. Hoy. This project and early versions of this hasselti Thorell (Araneidae: Theridiidae). Philippine Agriculturist, 51, 171–180. manuscript were improved by comments from J. D. Christenson, T. E. 1989. Sperm depletion in the orb-weaving spider Calkins, J. Dale, S. Flaxman, M. E. Hauber, S. Kain, J. N. Nephila clavipes (Araneae: Araneidae). Journal of Arachnology, 17, Mager, A. C. Mason, R. Saffron, and other members of the 115–118. ‘shemlens’ behaviour discussion group and behaviour Cockburn, A. 1994. Adaptive sex allocation by brood reduction in ‘lunch bunch’ (Cornell University, 1996–2000). M.C.B.A. antechinuses. Behavioral Ecology and Sociobiology, 35, 53–62. is particularly grateful for helpful suggestions and com- Cohn, J. 1990. Is it the size that counts? Palp morphology, sperm ments from her Ph.D. coadvisors: S. T. Emlen and P. W. storage, and egg hatching frequency in Nephila clavipes (Araneae, Sherman, and committee: B. N. Danforth, R. R. Hoy, and Araneidae). Journal of Arachnology, 18, 59–71. H. K. Reeve. This study was completed in partial fulfill- Costa, F. G. 1998. Copulatory pattern and fertilization success in ment of the requirements for a Ph.D. in the Department male wolf spiders without pre-or post-copulatory sperm induc- tion. Journal of Arachnology, 26, 106–112. of Neurobiology and Behavior at Cornell University. Dewsbury, D. A. 1978. The comparative method in studies of reproductive behaviour. In: Sex and Behaviour: Status and Prospec- References tus (Ed. by T. E. McGill, D. A. Dewsbury & B. D. Sachs), pp. 83–112. New York: Plenum. Alonzo, S. H. & Warner, R. R. 1999. A trade-off generated by sexual Dewsbury, D. A. 1979. Copulatory behaviour of deer mice (Peromy- conflict: Mediterranean wrasse males refuse present mates to scus maniculatus): III. Effects on pregnancy initiation. Journal of increase future success. Behavioral Ecology, 10, 105–111. Comparative and Physiological Psychology, 93, 178–188. Andersson, M. 1994. Sexual Selection. Princeton, New Jersey: Prin- Dewsbury, D. 1982. Ejaculate cost and mate choice. American ceton University Press. Naturalist, 119, 601–610. ANDRADE & BANTA: FUNCTIONAL STERILITY IN SPIDERS 869

Eberhard, W. G. 1985. Sexual Selection and Animal Genitalia. MacDiarmid, A. B. & Butler, M. J. 1999. Sperm economy and Cambridge, Massachusetts: Harvard University Press. limitation in spiny lobsters. Behavioral Ecology and Sociobiology, 46, Elgar, M. A. 1992. Sexual cannibalism in spiders and other inverte- 14–24. brates. In: Cannibalism: Ecology and Evolution Among Diverse Taxa Main, B. Y. 1980. Spiders of Australia. Brisbane: Jacaranda Press. (Ed. by M. A. Elgar & B. J. Crespi), pp. 128–155. Oxford: Oxford Maxwell, M. R. 1998. Lifetime mating opportunities and male University Press. mating behaviour in sexually cannibalistic praying mantids. Ani- Emlen, S. T. & Oring, L. W. 1977. Ecology, sexual selection, and the mal Behaviour, 55, 1011–1028. evolution of mating systems. Science, 197, 215–223. Maxwell, M. R. 1999. The risk of cannibalism and male mating Foelix,R.F.1996. Biology of Spiders. 2nd edn. Cambridge, Massa- behaviour in the Mediterranean praying mantid, Iris oratoria. chusetts: Harvard University Press. Behaviour, 136, 205–219. Forster, L. M. 1984. The Australian redback spider (Latrodectus Newman, J. A. & Elgar, M. A. 1991. Sexual cannibalism in orb- hasselti): its introduction and potential for establishment and weaving spiders: an economic model. American Naturalist, 138, distribution in New Zealand. In: Commerce and the Spread of Pests 1372–1395. and Disease Vectors (Ed. by M. Laird), pp. 273–289. New York: Parker, G. A. 1998. Sperm competition and the evolution of Praeger. ejaculates: towards a theory base. In: Sperm Competition and Forster, L. M. 1992. The stereotyped behaviour of sexual cannibal- Sexual Selection (Ed. by T. R. Birkhead & A. P. Møller), pp. 3–54. ism in Latrodectus hasselti Thorell (Araneae: Theridiidae), the London: Academic Press. Australian redback spider. Australian Journal of Zoology, 40, 1–11. Paul, A. J. 1984. Mating frequency and viability of stored sperm in Forster, L. M. 1995. The behavioural ecology of Latrodectus hasselti the tanner crab Chinoecetes bairdi (Decapoda, Majidae). Journal of (Thorell), the Australian redback spider (Araneae: Theridiidae): a Crustacean Biology, 4, 375–381. review. Records of the Western Australia Museum, Supplement, 52, Petrie, M., Hall, M., Halliday, T., Budgey, H. & Pierpoint, C. 1992. 13–24. Multiple mating in a lekking bird: why do peahens mate with Gage, A. R. & Barnard, C. J. 1996. Male crickets increase sperm more than one male and with the same male more than once? number in relation to competition and female size. Behavioral Behavioral Ecology and Sociobiology, 31, 349–358. Ecology and Sociobiology, 38, 349–353. Polis, G. A. 1981. The evolution and dynamics of intraspecific Gould, S. J. 1984. Only his wings remained. Natural History, 9, predation. Annual Review of Ecology and Systematics, 12, 225–251. 10–18. Prenter, J., Elwood, R. W. & Montgomery, W. I. 1994. Male Hetrick, L. 1970. Biology of the ‘love-bug’ Plecia nearctica (Diptera: exploitation of female predatory behaviour reduces sexual canni- Bibiondae). Florida Entomologist, 53, 23–26. balism in male autumn spiders, Metellina segmentata. Animal Hieber, C. S. & Cohen, J. A. 1983. Sexual selection in the Behaviour, 47, 235–236. lovebug, Plecia nearctica: the role of male choice. Evolution, 37, Robinson, M. H. & Robinson, B. 1980. Comparative studies of the 987–992. courtship and mating behaviour of tropical araneid spiders. Pacific Hunter, F. M., Petrie, M., Otronen, M., Birkhead, T. R. & Møller, Insects Monographs, 36, 1–218. A. P. 1993. Why do females copulate repeatedly with one male? Roff, D. A. 1992. The Evolution of Life Histories: Theory and Analysis. Trends in Ecology and Evolution, 8, 21–26. New York: Chapman & Hall. Jakob, E. M., Marshall, S. D. & Uetz, G. W. 1996. Estimating fitness: Rubenstein, D. I. 1985. Risk, uncertainty and evolutionary strate- a comparison of body condition indices. Oikos, 77, 61–67. gies. In: Current Problems in Sociobiology (Ed. by King’s College Jackson, R. R. 1980. The mating strategy of Phidippus johnsoni Sociobiology Group), pp. 91–111. Cambridge: Cambridge (Araneae, Salticidae): II. Sperm competition and the function of University Press. copulation. Journal of Arachnology, 8, 217–240. Sasaki, T. & Iwahashi, O. 1995. Sexual cannibalism in an orb- Jackson, R. R. & Pollard, S. D. 1997. mating weaving spider Argiope aemula. Animal Behaviour, 49, 1119–1121. strategies: sex among cannibals in and out of webs. In: The Schneider, J. M. & Lubin, Y. 1998. Intersexual conflict in spiders. Evolution of Mating Systems in Insects and Arachnids (Ed.byJ.C. Oikos, 83, 406–506. Choe & B. J. Crespi), pp. 340–351. Cambridge: Cambridge Schneider, J. M., Herberstein, M. E., de Crespigny, F. C., Rama- University Press. murthy,S.&Elgar,M.A.2000. Sperm competition and small Johns, P. M. & Maxwell, M. R. 1997. Sexual cannibalism: who size advantage for males of the golden orb-web spider Nephila benefits? Trends in Ecology and Evolution, 12, 127–128. edulis. Journal of Evolutionary Biology, 13, 939–946. Johnson, J. C. 2001. Sexual cannibalism in fishing spiders Simmons, L. W., Teale, R. J., Maier, M., Standish, R. J., Bailey, W. (Dolomedes triton): an evaluation of two explanations for female J. & Withers, P. C. 1992. Some costs of reproduction for male aggression towards potential mates. Animal Behaviour, 61, 905– bushcrickets, Requena verticalis (Orthoptera: Tettigoniidae): allo- 914. cating resources to mate attraction and nuptial feeding. Behavioral Kaston, B. J. 1970. Comparative biology of American black widow Ecology and Sociobiology, 31, 57–62. spiders. San Diego Society of Natural History Transactions, 16, Smith,H.G.1995. Experimental demonstration of a trade-off 33–82. between mate attraction and paternal care. Proceedings of the Kavale, J. 1986. The comparative biology of two Latrodectus species. Royal Society of London, Series B, 260, 45–51. M.Sc. thesis, University of Otago. Spence, J. R., Zimmermann, M. & Wojcicik, J. P. 1996. Effects of Knoflach, B. & van Harten, A. 2000. Palpal loss, single palp food limitation and sexual cannibalism on reproductive output of copulation and obligatory mate consumption in Tidarren cuneola- the Dolomedes triton (Araneae: Pisauridae). tum (Tullgren, 1910) (Araneae, Theridiidae). Journal of Natural Oikos, 75, 373–382. History, 34, 1639–1659. Suter, R. B. 1990. Courtship and assessment of virginity by male Lake, P. E. 1975. Gamete production and the fertile period bowl and doily spiders. Animal Behaviour, 39, 307–313. with particular reference to domesticated birds. Symposia of the Suter, R. B. & Parkhill, V. S. 1990. Fitness consequences of Zoological Society of London, 35, 225–244. prolonged copulation in the bowl and doily spider. Behavioral Lawrence, S. E. 1992. Sexual cannibalism in the praying mantid, Ecology and Sociobiology, 26, 369–373. Mantis religiosa: a field study. Animal Behaviour, 43, 569–583. Thornhill, R. 1976. Reproductive behaviour of the lovebug, Plecia Lee, A. K. & Cockburn, A. 1985. Evolutionary Ecology of Marsupials. nearctica (Diptera: Bibionidae). Annals of the Entomological Society Cambridge: Cambridge University Press. of America, 69, 843. 870 ANIMAL BEHAVIOUR, 63, 5

Thornhill, R. 1980. Sexual selection within mating swarms of the Watson, P. J. 1986. Transmission of female sex pheromone lovebug, Plecia nearctica (Diptera: Bibionidae). Animal Behaviour, thwarted by males in the spider Linyphia litigiosa. Science, 233, 28, 405–412. 219–221. Thornhill, R. 1983. Cryptic female choice and its implications in the Watson, P. J. 1990. Female-enhanced male competition deter- scorpionfly Harpobittacus nigriceps. American Naturalist, 122, 765– mines the first mate and principal sire in the spider Linyphia 788. litigiosa (). Behavioral Ecology and Sociobiology, 26, Thornhill, R. & Alcock, J. 1983. The Evolution of Insect Mating 77–90. Systems. Cambridge, Massachusetts: Harvard University Press. Watson, P. J. 1991. Multiple paternity and first mate sperm prec- Trivers, R. L. 1972. Parental investment and sexual selection. In: edence in the sierra dome spider, Linyphia litigiosa Keyserling Sexual Selection and the Descent of Man (Ed. by B. Campbell), pp. (Linyphiidae). Animal Behaviour, 41, 135–148. 136–179. Chicago: Aldine. Watson, P. J. & Lighton, J. R. B. 1994. Sexual selection and the Uhl, G. 1998. Mating behaviour in the cellar spider, Pholcus phalan- energetics of copulatory courtship in the Sierra dome spider, gioides, indicates sperm mixing. Animal Behaviour, 56, 1155–1159. Linyphia litigiosa. Animal Behaviour, 48, 615–626. Vollrath, F. 1998. Dwarf males. Trends in Ecology and Evolution, 13, Wilkinson, L., Hill, M., Welna, J. P. & Birkenbeuel, G. K. 1997. 159–163. SYSTAT 7.0 for Windows. Chicago: SPSS. Vollrath, F. & Parker, G. A. 1992. Sexual dimorphism and distorted sex ratios in spiders. Nature, 360, 156–159. Williams, G. C. 1966. Natural selection, the costs of reproduction, Warner, R. R., Shapiro, D. Y., Marcanato, A. & Petersen, C. W. and a refinement of Lack’s principle. American Naturalist, 100, 1995. Males with highest mating success convey the lowest 687–690. fertilization benefits to females. Proceedings of the Royal Society of Zar, J. H. 1984. Biostatistical Analysis. Englewood Cliffs, New Jersey: London, Series B, 262, 135–139. Prentice Hall.