Spider behaviors include oral sexual encounters

Several clades of spiders whose females evolved giant sizes are known for extreme sexual behaviors such as sexual cannibalism, opportunistic mating, mate-binding, genital mutilation, plugging, and emasculation. However, these behaviors have only been tested in a handful of size dimorphic spiders. Here, we bring another lineage into the picture by reporting on sexual behavior of Darwin’s bark spider, Caerostris darwini. This sexually size dimorphic Madagascan species is known for extreme web gigantism and for producing the world’s toughest biomaterial. Our field and laboratory study uncovers a rich sexual repertoire that predictably involves cannibalism, genital mutilation, male preference for teneral females, and emasculation. Surprisingly, C. darwini males engage in oral sexual encounters, rarely reported outside mammals. Irrespective of female’s age or mating status males salivate onto female genitalia pre-, during, and post-copulation. While its adaptive significance is elusive, oral sexual contact in spiders may signal male quality or reduce sperm competition.

(all matings in nature happened off-web), letting her adjust for at least 2 hours, and then gently introducing a virgin male to the mating arena. We recorded the male courting behaviors (i.e. mate binding: yes/no; latency from the male first touching the female to first palpal insertion), the number of oral sexual contacts (see Results for description), number of insertions with each palp, duration of each insertion, the occurrence of palpal damage, and the occurrence of female aggression and cannibalism. We terminated a mating trial after the couple had permanently separated or the female had attacked the male. After mating trials, we anesthetized females using CO2 and inspected them under a stereomicroscope for externally visible genital plugs. We checked males after 24 hours for emasculation. After the experiments, we euthanized the females, stored them in 70% ethanol, and inspected them for plugs within internal genitals by cutting open their spermathecae 2 . Four of the 17 mated and plugged females were subjected to further mating trials (up to three mating trials per female) to test for remating ability, and for persistence and effectiveness of genital plugs.

Statistical analysis
We checked for data normality using Kolmogorov-Smirnov test. We report mean values (A) ± standard deviations for normally distributed parameters and medians (µ1/2) ± interquartile ranges (IQR) for non-normally distributed data. To test for male preference for subadult versus adult female webs in the field, and for how long males guarded these females, we used Mann-Whitney U-test. To test for differences in courting duration between males mating with older virgin and previously mated females, we used Mann-Whitney U-test. To test for differences in insertion durations between older virgin and previously mated females, we used analysis of variance (ANOVA). We used the χ² test to test whether the number of previously used palps affects the male intensity of oral sexual contact, and Cramer's V test to test for differences in binomial variables (female aggressiveness towards males, sexual cannibalism).
We used logistic regressions to test whether female and male size measures (independent factors: length of first patella tibia and live mass) relate to female aggressiveness and sexual cannibalism occurrences (dependent factors). Because live mass data was not available for several spiders, we measured the abdomen size (length x width) as an estimate of live mass, following 3 who showed that abdomen size explains up to 94% of mass variability. Besides using only female and male size measures, we also explored a possible effect of ratios of female:male size and mass. We performed all the above analyses in PASW 18 4 .
To estimate the operational sex ratio, we used the mark-release-recapture data of the 14 days transect. To estimate the sizes of the female and male subpopulations in our transect, we used the constrained linear model (CLM) methodology in the MARK 5.1 software, as used in Čelik 5 .

SSD and operational sex ratio
On the first transect day, we found 40 penultimate or adult females and 18 males of C. darwini. During the subsequent two-week long monitoring of the transect, we found an additional four females and 22 males, for a total of 44 females and 40 males. Of the 44 females, 31 were adult and 13 were penultimate. Of the 40 males, 36 were intact while three had one palp, and one had both palps severed. The mark-release-recapture method estimated the female subpopulation being 37 ± 0.96 • 10 -4 individuals, and the male subpopulation being 52.69 ± 3.54 individuals. In our transect, males outnumbered females by ~ 1.4 (1.33-1.52) times.

Mating observations
In the field we observed five matings. In four of these matings, a male mated with a teneral virgin female (Fig. 1C), and once a male mated with an older virgin female two days after her final molt (Fig.1 D). In the laboratory, we subjected 17 older virgin females to mating trials, of which 14 mated in the first trial, while three mated in the second. We subjected four of these 17 females to further two or three mating trials for a total of nine trials, all of which ended in copulation.

Mate guarding
The five females observed maturing during the field transect ceased web building four to seven days prior to maturation to rest on vegetation. During this time, the guarding male was in contact with the female (Fig. 1A). Males associated with web-building females were always on the periphery of the web, only occasionally walking onto the web, when both subadult and adult females typically responded by aggressively shaking the web.
In laboratory trials, males courted (latency from first touching the female to first palpal insertion) older virgin females for 485-3254 s (µ1/2 ± IQR = 1181 ± 694 s, N = 17), and previously mated females for 804-2613 s (µ1/2 ± IQR = 1208 ± 893 s, N = 9). Male courting duration did not differ between older virgin and previously mated females (Mann-Whitney U = 71, p = 0.767, N = 26). In the field, we did not measure the courting duration of males mating with teneral females, because such males spent two to five days on vegetation in contact with subadult females that were about to molt.

Female sexual behavior
In all field and laboratory observed matings, females interrupted male courtship and copulation several times. No teneral females attacked the males (N = 4). Older females behaved aggressively towards their mates in 38.5% matings (N = 26) and cannibalized them in 30.8% cases. Of these cannibalized males (N = 8), 50% were cannibalized after using both palps, while the other 50% were cannibalized after using one palp. Older virgin and previously mated females did not differ in aggression towards males (Cramer's V = 0.077, p = 0.696, N = 26, Table 2) or in sexual cannibalism (Cramer's V = 0.135, p = 0.492, N = 26, Table 2).

385-1422
A ± SD = 799.9 ± 375.6 N = 9 Females were more likely to be aggressive towards and to cannibalize males of a higher mass relative to them (aggressiveness:

Male sexual behavior
While all males (100%) used both palps when mating with teneral females, only 52.9% (N = 17) and 66.67% (N = 9) males used both palps when mating with older virgin and previously mated females, respectively.
The duration of palpal insertions did not differ between males that mated with older virgin and previously mated females (F25,24 = 1.09, p = 0.307, Tab. 2, Fig. 4). We were able to measure the duration of only three insertions with teneral females; they lasted 691 s, 1524 s and 1802 s, all considerably longer compared to non-teneral females (Fig. 4). In field and laboratory observations, all males employed oral sexual behavior where they salivate onto female genitalia, independent of the females' mating status (see main text). Males mating with an older female always engaged in extensive mate binding, while males that mated with a teneral female never bound her in silk (see main text).
In laboratory trials, we noted externally visible genital plugs after 58.8% matings of virgin older females (N = 17, Fig. 2C). We found genital plugs logged inside female spermathecae (Fig. 2D) in 58.3% of females mating once (N = 12), and in all females that mated at least two times (N = 4). In females (N = 4) from remating trials, we never found more than one external or internal plug per copulatory opening. In 75% of these females, the presence of external plugs was changing with subsequent matings (N = 9), indicating plug removal by subsequent males.

Further discussion
With a mean female/male body size ratio of 2.35, C. darwini exhibits a moderate SSD ratio (following 6 ). Highly sexually size dimorphic species have a male-biased operational sex ratio due to an asynchronous development of the sexes, where large females undergo more molts to reach adulthood [7][8][9] . Consistently with this prediction, we found a ~ 1.4 male-biased operational sex ratio in the transect. A male-biased sex ratio leads to male accumulation around females and thus strong male-male competition [10][11][12] . Additionally, mortality of searching males can be high 13,14 . In theory, this leads to monogyny through male adaptations to monopolize females 10,15 . Here, we report on several C. darwini sexual behaviors that indicate a mono-or at most bigynous and a polyandrous mating system.
In response to intense sperm competition males of sexually size dimorphic orb weavers evolved strategies to monopolize females, thereby securing paternity, i.e. mate guarding [16][17][18] , opportunistic mating 18,19 , plugging of female genitalia [20][21][22] , genital self-mutilation [23][24][25] , and remote copulation 26 . Most such male strategies are present in C. darwini. In this species, males prefer subadult (and thus virgin) over older females, and pre-copulatory guard subadult females. C. darwini males obligatorily damage their palps by leaving copulatory plugs (embolic leftovers) inside female copulatory openings. These males then chew-off the remaining palpal bulbs to become eunuchs 6,27 . Mate-plugging is hypothesized to be adaptive because genital plugs might prevent subsequent males to copulate with the same female (plugging hypothesis), while removal of entire palps might render the eunuch male to become a better fighter in male-male contests, either through increased aggression (better-fighter hypothesis) or increased agility (gloves-off hypothesis) 17,25,27,28 . However, in C. darwini, genital plugs are likely removed by subsequent males, enabling females to remate into previously used copulatory openings. The described mating patterns indicate that C. darwini males invest their whole paternity potential into one female, while females may be polyandrous. An analogous combination of mating strategies is known from some SSD spider lineages, e.g. some species of Argiope, Latrodectus and Nephila [28][29][30][31][32][33]  Sexual aggression and cannibalism are female mechanisms to counter male monopolization, to increase pre-copulatory mate choice and to manipulate male paternity by controlling copulation duration [37][38][39][40][41][42][43] . To counter female aggression, males of sexually size dimorphic spiders have evolved strategies such as opportunistic mating and mate binding. C. darwini males preferentially guard subadult females and then engage in lengthy copulations with teneral females which are unable to cannibalize them. Males always use both palps when mating with teneral females and copulations are longer compared to mating with older females (Fig. 4). In contrast, older females terminate 42% matings prior to the second palpal insertion, and cannibalize 31% of their mates. Additionally, older females are not increasingly aggressive towards subsequent suitors, indicating female preference for polyandry.
Finally, among the plethora of sexual behaviors in C. darwini, it seems that the newly described oral sexual contact is particularly noteworthy (see Discussion in the main text).