Copulation with immature females increases male fitness in cannibalistic widow spiders

Abstract

Copulatory cannibalism of male ‘widow’ spiders (genus Latrodectus) is a model example of the extreme effects of sexual selection, particularly in L. hasselti and L. geometricus where males typically facilitate cannibalism by females and mate only once. We show that these males can increase their reproductive success by copulating with final-instar, immature females after piercing the female's exoskeleton to access her newly developed sperm storage organs. Females retain sperm through their final moult and have similar fecundity to adult-mated females. This is an adaptive male tactic because immature mating increases insemination success relative to adult mating (which predicts higher paternity) and moreover, rarely ends in cannibalism, so males can mate again. Although successful only during a brief period before the female's final moult, males may employ this tactic when they associate with final-instar females in nature. Consistent with this, one-third of L. hasselti females collected as immatures in nature were already mated. Immature mating alters sexual selection on these otherwise monogynous males, and may explain male traits allowing facultative polygyny in Latrodectus. Since male cohabitation with immature females is common among invertebrates, immature mating may be a widespread, previously unrecognized mating tactic, particularly when unmated females are of high reproductive value.

1. Background

Intense sexual selection on competing males has been linked to the independent evolution of monogyny (males mating only once) in many spider taxa [1,2]. In some monogynous ‘widow’ spiders adult males are regularly cannibalized by females during or after mating [3–8]. Latrodectus hasselti and L. geometricus males facilitate monogyny via self-sacrificial mating behaviour [3,4,6], which may increase paternity with polyandrous females [4]. In these spiders, over 2 h of vibratory courtship precedes mating with adult females ([3,6,8], table 1), after which the male inserts one pedipalp (paired secondary copulatory organ) through an opening in the female's genitalia (epigynum, figure 1), and lifts his abdomen onto the female's fangs (copulatory somersault) while in copula [3,4,6]. Males that copulate with both pedipalps inseminate the female's two sperm storage organs (spermathecae, figure 1f, [5,9]). Females may, however, cannibalize males after one copulation (particularly if courtship is brief [5,8]), thus limiting paternity if females remate with a new male [6,8]. Since the tip of each pedipalp can break inside the spermatheca [6,9,10] and block later inseminations, the first mate has the highest paternity [5,10,11]. Breakage does not, however, prevent female nor male remating [4,7,10,12,13]. Since females may mate with more than one male in nature [4], expected paternity is highest if these monogynous males mate with virgins, copulate twice and leave a plug in each spermatheca [4–6].

Table 1.

Pre-copulatory behaviour and female reproductive output (mean ± s.e.) for adult L. hasselti and L. geometricus males that copulate with immature or adult conspecific females (significant comparisons in bold).

variable	L. hasselti females (N)			L. geometricus females (N)
	immatures	adults	statistics	immatures	adults	statistics
web-bound vibratory courtship (proportion of males)	0% (11)	100% (13)	Fisher's exact test p < 0.001	100% (17)	100% (22)	Fisher's exact test, p = 1.0
latency to copulation (min)a	155.2 ± 28.3 (11)	140.4 ± 17.7 (13)	GLMd: Wald p = 0.620	44.4 ± 15.2 (18)	208.9 ± 13.6 (23)	GLM: Wald p < 0.001
fecundityb	192.1 ± 24.6 (10)	180.5 ± 14.0 (11)	GLM: Wald p = 0.69	69.8 ± 9.2 (6)	87.9 ± 9.2 (9)	GLM: Wald p = 0.13
fertilityc	90.9% ± 2.5 (10)	95.3% ± 0.9 (11)	GLM: Wald p = 0.079	82.2% ± 10.3 (6)	93.7% ± 5.0 (9)	GLM: Wald p = 0.98

^aTime from start of trial until start of the first copulation. This included sustained vibratory signalling (all adult females and L. geometricus with immatures) or males approaching without vibrations (L hasselti with immatures).

^bNo. viable eggs in the first egg sac produced.

^c(No. spiderlings/total number of eggs) × 100.

^dGLM: generalized linear models with gamma link functions with female developmental stage as predictor and latency to copulation, fecundity or fertility as outcomes (SPSS v. 22).

Figure 1.

(a) Adult female Latrodectus hasselti (centre) with sexually cannibalized male (anterior). Square and arrow indicate the epigynum (EP, external genitalia), magnified in (b). The same area is intact but raised in immature females (c–e) and the paired spermathecae (SP) are usually visible through the exoskeleton late in the final juvenile instar (d). There are no external openings unless immatures have copulated with males (e). Transverse-posterior view of dissected, fully developed spermathecae (SP), copulatory ducts (CD) and the exoskeleton (EX) covering the epigynum of a late-stage immature female (f). Images (b–f) taken with a Nikon DXM 1200, and Zeiss Stemi 2000C microscope (scale bars, 0.55 mm).

As predicted by first-male sperm precedence, Latrodectus males frequently cohabit on the webs of final-instar juvenile females (immatures) in nature ([13] and references therein). Here we document a fitness-enhancing alternative tactic of cohabiting male L. hasselti and L. geometricus whereby males copulate with immature females by accessing their concealed genitalia (immature mating). These females moult normally and produce offspring despite not having mated as adults.

2. Methods and results

Internal genitalia of female spiders develop during the final juvenile instar, but are covered by exoskeleton without external openings (e.g. [9], figure 1). Nevertheless, L. hasselti males approach immature females in the laboratory [3], and mount them in nature (M.C.B.A. 1996, 1997, personal observation); 21% (n = 74) of cohabiting L. hasselti males were found with final-instar females and 35% of L. hasselti females collected within five days prior to their adult moult (n = 78) had already mated (electronic supplementary material). To evaluate immature mating as an adaptive mating tactic, we examined whether temporal patterns of male sexual behaviour corresponded with female spermathecal development, and whether immature matings could increase male fitness.

The duration of the final instar of laboratory-reared females (electronic supplementary material) was used to estimate the expected moult date (EMD) of L. hasselti females dissected at different points in their final instar (n = 29). While spermathecae were absent at the start of the final instar, internal genitalia were present in more than half of females by EMD-3 (n = 6), and were fully developed in all females by EMD-2 (n = 6). Developed spermathecae were often visible through the intact exoskeleton (figure 1d, n = 17) of the protruding epigynal area [9]. L. geometricus immatures had similar signs of spermathecal development by EMD-4 (n = 31). As predicted, males paired with final-instar females mounted and copulated only within 2 (L. hasselti, n = 35) or 4 days of the final moult (L. geometricus, n = 32; electronic supplementary material, figure S1). Mated immature females had precise openings in the cuticle (cf. figure 1b,c,e) exposing the underlying epigynal copulatory openings. Observed contact between the males' mouthparts and the raised epigynal area during trials suggests males opened the female's cuticle with their fangs and accessed the copulatory openings.

We paired males with immatures within 2 (Latrodectus hasselti) or 4 (L. geometricus) days of EMD and compared courtship and mating with males paired with adults. Male courtship effort with immatures was significantly reduced in both species ([14], table 1). All males that mated adults produced web-based vibratory signals throughout the period leading to mating. L. hasselti immatures, however, were approached on the web without vibratory signals (although latency to copulation was similar, table 1). Although L. geometricus males approached immatures with vibratory signals, the duration of courtship prior to mating was reduced by a factor of 4 (table 1). Despite reduced courtship, males frequently copulated with immatures (L. hasselti: 64.7%, n = 17; L. geometricus: 58%, n = 32), although at lower rates than with adults (L. hasselti: 89.5%, n = 38; Fisher's exact p = 0.05; L. geometricus: 96%, n = 25, Fisher's exact p = 0.001). Males that mated immatures, however, were more likely to inseminate both spermathecae, and males were equally (L. hasselti) or more likely (L. geometricus) to plug both spermathecae of immatures compared with adults (figure 2). Immature-mated females moulted normally and had similar fecundity and fertility to adult-mated females (table 1).

Figure 2.

Mating outcomes for L. hasselti (grey, grey-striped) or L. geometricus (black, black-striped) males paired with adult (striped) or immature (solid) females (sample sizes above). Significant comparisons are indicated (Fisher's exact tests, **p ≤ 0.002; *p = 0.04).

Copulatory somersaults and sexual cannibalism were rare (L. hasselti) or absent (L. geometricus) in immature matings ([14], figure 2). Immature-mating males later paired with virgin adult females frequently courted (L. hasselti: 59%, n = 12; L. geometricus: 100%; n = 15), and sometimes re-mated and fathered additional offspring (L. hasselti: 8%, n = 12; L. geometricus: 80%, n = 15).

3. Discussion

Latrodectus hasselti and L. geometricus males successfully mated with sexually immature females by opening the exoskeleton that covers the female's spermathecae. Moreover, L. hasselti females appear to frequently mate in this way in nature. This tactic will increase male fitness in two ways. First, immature-mating males are more likely to achieve sperm precedence as males had higher success at inseminating and plugging both spermathecae of immature compared with mature females (figure 2). Second, immature-mating males can copulate with a second female as they usually survive their first mating (figure 2). These gains come at no cost to offspring production (table 1). Thus, the intense sexual selection argued to lead to monogyny [1,2] may also have given rise to facultative polygyny in L. hasselti and L. geometricus. More broadly, this previously unrecognized alternative mating tactic may be common in other invertebrate taxa where males interact with immature females [15–17].

Although advantageous, the frequency of the immature-mating tactic in nature depends on the male's ability to identify immature females during a brief developmental window (electronic supplementary material, figure S1). Approaching a younger female may be fatal. Immature females without developed spermathecae have been observed cannibalizing adult males in the field (Latrodectus hasselti, M.C.B.A. 1996, 1997, personal observation) and in our trials (L. hasselti n = 1; L. geometricus n = 2). Despite this risk, our data show that L. hasselti males successfully find and mate one-third of immature females in nature. Frequent reports of cohabitation of Latrodectus males with immature females [15,16] suggest the opportunity for immature mating may be common in the genus.

Latrodectus hasselti and L. geometricus males are fascinating because of their extreme adaptations for monogyny [3,6,10]. Nevertheless, this study predicts the potential for polygyny should also shape the evolution of male traits in these species. This is consistent with demonstrations of sustained sperm production by adult male L. hasselti [18] and L. geometricus (P. Michalik and I.S. 2010, unpublished data), in strong contrast with several other monogynist spider species where spermiogenesis ceases in adults, and males become permanently sperm-depleted after mating [19].

Males of some invertebrates mate with adult females that are still inside their pupal cases [20], or that have recently moulted [21–24], and others court [17] or pseudocopulate [25] with immature females. These could be evolutionary precursors to immature mating. Although to our knowledge, successful insemination of immature females has not been previously reported, we suggest immature-mating tactics could be widespread among invertebrates where cohabitation of adult males and immature females occurs in nature (e.g. [17]) particularly when males are under selection to increase otherwise rare opportunities to mate or secure fertilizations. In widow spiders, extreme sexual selection is thought to be responsible for the evolution of male self-sacrifice [1,2,4]. Our work suggests strong sexual selection may be concurrently linked to this opportunistic tactic that simultaneously escalates inter-male competition over mating, and circumvents monogyny.

Ethics

Spider collection and exportation followed regulations in Israel and Australia, and importation regulations for Canada, treatment of spiders followed guidelines at the University of Toronto.

Data accessibility

Raw data are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.t3j5n [14].

Authors' contributions

Data collection for L. hasselti was by M.D.B. (laboratory) and M.C.B.A. (field), and for L. geometricus by I.S. (laboratory). M.C.B.A. analysed the data and wrote the paper with input from A.R.H. and Y.L. Study design, interpretation and manuscript editing involved all authors. All authors agree to be held accountable for the content herein and approved the final manuscript.

Competing interests

The authors declare no competing interests.

Funding

Israel Association for Canadian Studies to I.S., and NSERC Canada (discovery grants 229029-04 and 229029-12), Canada Research Chairs (950-228362), Canadian Foundation for Innovation and Research & Innovation Ontario (203764) to M.C.B.A.