Behavioral Ecology The official journal of the ISBE International Society for Behavioral Ecology Behavioral Ecology (2013), 24(5), 1218–1228. doi:10.1093/beheco/art056 Original Article Parental effects on early development: testing for indirect benefits of polyandry Sheri L. Johnson and H. Jane Brockmann Department of Biology, University of Florida, PO Box 118525, Gainesville, FL 32611, USA Received 11 December 2012; revised 2 June 2013; accepted 5 June 2013; Advance Access publication 6 July 2013. Females of many species mate with multiple males even when it is costly. Multiple mating may allow females to exploit postcopulatory mechanisms to ensure that their eggs are fertilized by high quality (good genes) and/or genetically compatible males. We conducted a series of noncompetitive in vitro fertilization experiments to evaluate the benefits of polyandry in naturally occurring pairs of horseshoe crabs, Limulus polyphemus. In this externally fertilizing species, attached pairs migrate to shore to spawn; unpaired males are attracted to spawning pairs by visual and chemical cues and become satellites of some (polyandrous) females while ignoring others (monandrous). When present, satellites may fertilize a high proportion of the female’s eggs, but their presence is costly to female nesting success. We measured developmental success for monandrous and polyandrous females crossed with attached and satellite males. We found that satellite males increased the success of polyandrous but not monandrous females. We then examined the effect of good genes and genetic compatibility on developmental success and offspring size using a North Carolina II breeding design. Results indicate that both incompatibilities between males and females and paternal good genes effects may provide a selective advantage that offsets the costs of multiple mating in this species. Key words: genetic compatibility, good genes, Limulus polyphemus, mate choice. Introduction Females mating with more than 1 male, that is, polyandry, is taxonomically widespread (Birkhead and Møller 1998; Birkhead et al. 2009), but the maintenance of the behavior is puzzling. In most cases, females can fertilize all of their eggs with the sperm from a single male and in many species mating with multiple males can be costly for females (Thornhill and Alcock 2001; Arnqvist and Rowe 2005), suggesting that there must be some compensating benefits. For some species, males provide direct benefits to the female or her offspring, such as resources or paternal care; for others, females mate with multiple males to ensure that all their eggs are fertilized (Ridley 1988), considered to be a direct benefit for internal fertilizers (Jennions and Petrie 2000). But for many others, including most externally fertilizing species (Byrne and Roberts 2004), males provide no benefits to females other than their genes (Jennions and Petrie 2000; Tregenza and Wedell 2000; Hosken and Stockley 2003; Zeh and Zeh 2003; Neff and Pitcher 2005; Simmons 2005; Kempenaers 2007; Slatyer et al. 2012). These females may gain from increasing offspring diversity (Yasui 1998; Johnson and Yund 2008), improving male quality (“good genes”; Kempenaers Address correspondence to S.L. Johnson, who is now at Department of Zoology and Department of Anatomy, University of Otago, PO Box 913, Dunedin 9054, New Zealand. E-mail: [email protected]. © The Author 2013. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: [email protected] et al. 1992; Wedell and Tregenza 1999) or increasing genetic compatibility (Wedekind 1994; Zeh and Zeh 1996, 1997; Newcomer et al. 1999; Ivy 2007). Benefits from good genes depend solely on male genetic quality, and directional selection favors intrinsically high-quality males through sperm competition or cryptic female choice (Eberhard 1996; Hosken and Stockley 2003; García-González and Simmons 2005, 2007; García-González and Evans 2011). In contrast, benefits from compatible genes depend on the interacting effects of male and female genotypes (either at fertilization or subsequently during embryo or offspring development), and nondirectional selection favors compatible male by female crosses (Neff and Pitcher 2005; Kempenaers 2007; Neff and Pitcher 2008). Although many studies have suggested that genetic compatibility may be important in the evolution of multiple mating, until recently few studies have explicitly tested the importance of genetic compatibility in a way that separates the effects of good genes from compatible genes (but see Wedekind et al. 2001, 2008; Evans and Marshall 2005; Marshall and Evans 2005; Rudolfsen et al. 2005; Trippel et al. 2005; Evans et al. 2007; Ivy 2007; Pitcher and Neff 2007; Dziminski et al. 2008; Rodriguez-Munoz and Tregenza 2009). To understand the benefits of polyandry, we take advantage of the unique mating system of the American horseshoe crab, Limulus polyphemus. In this species, females that are ready to lay eggs migrate toward shore during a limited period of time when high tides are exceptionally high due to a new or full moon (Brockmann Johnson and Brockmann • Indirect benefits of polyandry in horseshoe crabs 1219 and Johnson 2011). Males are attracted to these females visually (Barlow and Powers 2003), and as the female swims by, a male simply attaches to her posterior spines with his modified pedipalps and the attached pair swims to shore where they spawn. The male fertilizes the female’s eggs externally with free-swimming sperm as the eggs are being laid in the sand (Brockmann 2003b). Little is known about the behavior of pairs prior to arriving on the beach, but there is no evidence and seemingly little opportunity for female mate choice prior to mating (Brockmann 2003a), particularly given the time constraints under which females are nesting (at our site <3 h at the time of the maximum high tide). Unattached males join some spawning pairs (polyandrous), whereas other nearby pairs spawn without satellites (monandrous). When present, satellites fertilize 1–80% of the female’s eggs depending on their position (Brockmann et al. 1994, 2000). Recent research, using the same study population, has established that nesting with satellites (i.e., polyandrous nesting) is costly to females in this system: when satellite males were experimentally removed from around polyandrous pairs, they continued to nest longer and the females laid more eggs than control females that continued to nest with their satellites in place (but were otherwise treated the same; Johnson and Brockmann 2010). In spite of these costs in nesting time and egg-laying success, the frequency of polyandrous females is high in this system, particularly when nesting densities are high (20–80% when more than 50 pairs are present; Johnson and Brockmann 2012). It could be that this costly nesting with satellites is due to competitive male behavior, but there is also some evidence that polyandrous females (or pairs) attract satellites using chemical and visual cues (Hassler and Brockmann 2001; Schwab and Brockmann 2007; Saunders et al. 2010). What makes this system particularly puzzling is that some females that are nesting on the same beach and at the same time as the polyandrous females do not have satellites (monandrous) and do not attract nearby males even when densities are high (Johnson and Brockmann 2012). These monandrous females do not pay the nesting costs associated with being surrounded by competing satellite males, so why are not all females monandrous? In some externally fertilizing species, polyandrous mating increases the number of fertilized eggs, but in this system nearly all eggs develop whether satellites are present or not (96.4% for monandrous and 96.8% for polyandrous manipulated pairs; Johnson and Brockmann 2010). Taken together, our previous studies suggest the hypothesis that there are postmating, indirect genetic benefits accruing to polyandrous females that compensate for the costs of nesting with satellites. Using eggs and sperm collected from wild, naturally occurring pairs, we conducted laboratory crosses to examine the hypothesis that polyandrous females benefit from multiple mating. Using a quantitative genetics breeding design, we further evaluated whether those benefits came from good genes or genetic compatibility. In Experiment 1, we conducted crosses (in vitro fertilization from collected eggs and sperm) between females and their attached males (i.e., the male that was attached to that female when she arrived on the beach) or females and a random selection of satellite males. We used both kinds of females, those observed to be polyandrous in the field and those that were collected at the same time but were monandrous. Under our hypothesis, we predicted that polyandrous females would show greater success (measured as developmental success of the embryos and larvae) when their eggs were fertilized by sperm from satellite males compared with their attached males, whereas monandrous females would show greater success when fertilized by their attached males. Previous work (Johnson and Brockmann 2010) demonstrated that although monandrous females do not attract males, some would tolerate the presence of a satellite male and continue nesting with a satellite present (“tolerant” females) if one arrived, whereas other females would not lay eggs and would leave the beach as soon as a satellite arrived at the nesting pair (“intolerant” females). We hypothesized that these 2 kinds of monandrous females would differ in the indirect benefits they derived if their eggs were fertilized by sperm from satellite males. In Experiment 2, we compared the success of eggs from “tolerant” and “intolerant” monandrous females when they were crossed using sperm from their attached males compared with random satellite males. We predicted that tolerant females should have the same or greater success when mated with satellite males and intolerant females should show decreased success when mated with satellites compared with their attached males. To determine the source of postmating, genetic benefits from multiple mating (i.e., good genes or compatibility effects), we conducted a third experiment. We used a cross-classified breeding design (North Carolina II [NCII]) and quantitative genetics analysis to partition phenotypic sources of variance in offspring fitness among additive (male and female intrinsic quality) and nonadditive (male × female interaction) effects (Evans and Marshall 2005; Evans et al. 2007; Pitcher and Neff 2007; Dziminski et al. 2008). We measured fitness traits for eggs from monandrous and polyandrous females crossed with sperm from random satellite males from the population. Our results indicate that the attached males of monandrous and polyandrous females differ in quality because polyandrous females have lower success with their attached males than monandrous females have with their attached males. Further, we find evidence for both good genes and incompatibility effects, which may offset the costs of multiple mating in this system. We add to the diverse list of animals that are known to show good genes and incompatibility effects (Evans and Marshall 2005; Rudolfsen et al. 2005; Bang et al. 2006; Evans et al. 2007; Marshall and Evans 2007; Pitcher and Neff 2007; Rodriguez-Munoz and Tregenza 2009), but our study is novel in that we used naturally occurring monandrous and polyandrous females to evaluate the role of indirect genetic benefits in this horseshoe crab mating system. Material and Methods Study organism We studied a wild population of horseshoe crabs, L. polyphemus, at Seahorse Key, Florida (SHK), an island along the northern Gulf of Mexico coast near Cedar Key, Levy County, Florida (Brockmann et al. 1994). During the highest tides in March, April, and May 2008 and 2009, we observed and collected males and females along the island’s sandy south beach. Each female arrived at the beach with 1 male attached to her posterior opisthosomal spines by his modified pedipalps. She dug into the sand where she usually laid eggs in a series of batches. We defined polyandrous females as those pairs that were joined by unattached, satellite males when the satellite was in contact with the female, her attached male, or another satellite. Monandrous females were defined as females that were dug into the sand with their attached male (i.e., nesting) but had no unattached, satellite males in contact with the pair. Eggs are fertilized externally as they are being laid in the sand by free-swimming sperm from the attached male or, if satellites are Behavioral Ecology 1220 present, also by satellites (Brockmann et al. 1994, 2000). Fertilized eggs remain in the sand for 2–4 weeks where embryos develop and hatch into free-swimming (nonfeeding) trilobite larvae that metamorphose into juveniles 2–4 weeks later in inshore areas near the breeding beach. The juveniles molt several times over the next 6 months and take 9–10 years to become adults (Shuster and Sekiguchi 2003). In vitro fertilization Monandrous and polyandrous females were picked up from the nesting beach at SHK, along with their attached males and random satellite males; they were marked uniquely with tags and taken to the University of Florida Marine Laboratory at SHK where they were maintained in flowing seawater tables for 1–2 days before being returned to the same nesting beach. Prior to release, all individuals were measured for size (width of carapace at the widest point) and condition. The condition index (ranging from 4 to 20, with the highest condition individuals receiving a 20) included evaluations of carapace color, the surface condition of the carapace, the condition of the eyes, and the condition of the posterior spines (see Johnson and Brockmann 2010) for specifics. Sperm and eggs were collected using an electrical stimulation protocol adapted from Brown and Clapper (1980) and Sasson et al. (2012). The animals were removed from the water and placed on their dorsal surface on a rubber mat, and the genital operculum was lifted to expose the gonopores. Excess water was wiped away, and an electrical stimulus (males: 10 V; females: 15–20 V) was applied below and slightly lateral to the gonopores. After stimulation, eggs were collected using a plastic spoon, and approximately 30 eggs were counted out and placed in each plastic dish (11.3 cm diameter × 3.9 cm height) containing 40 mL of clean seawater (average salinity 34 ppt). The University of Florida obtains seawater from Whitney Marine Laboratory for use in the laboratory at the Gainesville campus, where we reared the eggs, so we also used this water for the fertilizations that were conducted in the laboratory at SHK. After stimulation, 5–30 µL of concentrated sperm was collected using a pipette, placed in a microcentrifuge tube, and kept on ice until needed (1 h maximum). When we were ready to conduct the noncompetitive in vitro fertilizations, we diluted the sperm using filtered, 34 ppt seawater at room temperature (seawater does not activate sperm motility in Limulus; Clapper and Brown 1980a, b). In each experiment, eggs and sperm were combined within 5–10 min of egg collection by gently pipetting 60 µL of diluted sperm suspension (see below for specifics) onto the eggs in each dish and gently and briefly swirling the dishes. After 60 min, excess sperm was washed away. The in vitro fertilized eggs were taken from SHK to the laboratory at the University of Florida in Gainesville, FL, where they were held at room temperature (22–25 °C). The water in each dish was changed daily for the first 3–4 days and then weekly for the remainder of the developmental period. We had previously determined that the eggs of monandrous and polyandrous females in this population did not differ in size (Johnson and Brockmann 2012); hence, we did not measure egg size for our experiments. Further, the amount of oviducal fluid was not controlled for in our study as little measureable oviducal fluid is released with the eggs during collection. Although horseshoe crabs do not fall under animal ethics regulations in the United States, we were always careful to ensure animal welfare throughout all experiments. All animals were returned to the beach, and many were seen spawning again at a later date. Sperm concentration measures In all experiments, sperm were diluted using serial dilutions. Male horseshoe crabs from our population release sperm at a concentration of 2–9.5 × 109 sperm mL−1 (Sasson et al. 2012). In Experiment 1, sperm were diluted to ~108 sperm mL−1 (10−1 dilution), but for Experiments 2 and 3, sperm were diluted to ~107 sperm mL−1 (10−2 dilution). The rationale behind switching to a more dilute sperm concentration resulted from a serial dilution experiment that we performed after starting Experiment 1. We compared the developmental success of eggs (N = 10 pairs of monandrous females with their respective attached males, mean sperm concentration of 6.57 × 109 sperm mL−1 ± 4.50 × 108 standard error [SE]) fertilized with 5 dilutions of sperm (108 to 101 sperm mL−1). We found that 107 sperm mL−1 resulted in asymptotic rates of development, and there was no significant difference in development rates among the 108, 107, and 105 sperm mL−1 concentration (Figure S1). For Experiments 2 and 3, we opted to use the 107 sperm mL−1 concentration because it ensured that we had plenty of sperm, particularly for Experiment 3 where we had to allocate sperm from a single male across 10 sets of eggs. We also calculated exact sperm counts for each male by diluting sperm to ~106 sperm mL−1 (10−3 dilution) and fixing with a 2:1 ratio of 2% gluteraldehyde. Sperm concentrations (sperm mL−1) were calculated using an improved Neubauer hemocytometer (see Table 1 for descriptive statistics). Because sperm concentrations varied somewhat from our expected 108 sperm mL−1 or 107 sperm mL−1, we included sperm concentration as a covariate in all analyses. Measure of developmental success It was not possible to assay true fertilization success (whether a zygote had been formed or not) in our in vitro fertilization experiments because horseshoe crab eggs have a tough, opaque chorion surrounding the egg and early embryo. This outer membrane ruptures by the 7th day at embryonic stage 19-2, and the developing embryo is then visible inside a balloon-like inner embryonic membrane (Shuster and Sekiguchi 2003). Because of this limitation, we measured the proportion of eggs that developed by day 20 (day 20 was chosen to ensure that we counted all eggs that were going to develop). Note that our serial dilution experiment (Figure S1) indicates that our sperm:egg ratios were not too high, as we see higher fertilization success at higher concentrations, but it is unclear whether we were using a sufficient amount of sperm to achieve fertilization of 100% of the eggs. The highest development achieved was 88% at 107 sperm mL−1, with a mean of 64% (a result similar Table 1 Mean sperm concentration (sperm mL−1) of the males used in Experiments 1–3 Variable Mean SE Sample size, N (missing) Attached males monandrous Satellite males monandrous Attached males polyandrous Satellite males polyandrous Attached males intolerant Attached males tolerant Satellite malesa Satellite malesb 9.2 × 109 1.0 × 1010 9.2 × 109 8.1 × 109 1.7 × 1010 1.3 × 1010 1.9 × 1010 6.56 × 109 9.0 × 108 1.0 × 109 7.4 × 108 8.9 × 108 5.0 × 109 4.0 × 109 5.5 × 109 1.21 × 108 18 (2) 14 20 (1) 14 25 25 24 (1) 176 aSatellite bSatellite males used in Experiment 2. males used in Experiment 3. Johnson and Brockmann • Indirect benefits of polyandry in horseshoe crabs 1221 to that reported in previous studies using similar methodology; Brown and Knouse 1973). Whether our measure of “developmental success” represents viability (i.e., all eggs were fertilized but only a proportion of those developed) or true fertilization success is unclear. Therefore, we take the conservative approach and refer to this measure as developmental success, recognizing that the variance in “developmental success” could be due to factors that affect zygote formation during the fertilization phase or genetic elements during the early developmental phase (García-González 2008). with 4 randomly chosen satellite males in 16 pairwise combinations, with replicate crosses performed for each (i.e., a total of 32 crosses per block). Eleven of these blocks were completed yielding 176 families for our genetic analysis. The eggs from each female were divided into 4 pairs of containers with ~30 eggs (mean = 30.05 ± 0.18 SE; median = 30; range 18–38; N = 352) in each. The eggs were fertilized with 60 µL of ~107 sperm mL−1 of ejaculate from each male. The sperm from each male were divided between each of the 4 females (2 replicates each). Response variables included 1) egg development (defined above) and 2) “early metamorphosis” (i.e., the proportion of eggs that developed and that then reached the juvenile stage (metamorphosis) by 45 days (from the time the cross was made). Most fertilized eggs take longer than 45 days to reach metamorphosis, so this is a measure of early metamorphosis, and 3) juvenile size at 60 days. Juvenile size (prosoma width [mm] measured at the widest point) was measured with calipers for 3 random juveniles from each cross. Note that for all experiments described above, if a female or male did not produce enough gametes for a complete block, they were not used in the experiment. Experiment 1: postmating benefits of multiple mating for monandrous and polyandrous females We compared egg development for both female types (monandrous and polyandrous) when crossed with their respective attached males to their developmental success when crossed with 2 randomly chosen satellite males (Figure S2-A; conducted 3–8 May 2008 and 12–26 March 2009). The eggs from each female were divided into replicate containers with ~30 eggs (mean = 27.5 ± 0.26 SE; median = 29; range 16–37; N = 315) in each. The eggs were fertilized with 60 µL of ~108 sperm mL−1 from each male. Three females were included in each block, and a total of 7 blocks were completed for each female type. We made 2 (2008) or 3 (2009) replicates for each of the 3 crosses with the different males (attached male, 2 satellite males). We measured the mean egg development (as defined above) for each female with each male. Experiment 2: postmating benefits of multiple mating for tolerant and intolerant monandrous females To identify monandrous females as tolerant or intolerant, we located a monandrous pair nesting on the beach (i.e., the female was well buried in the sand) and allowed them to nest for an additional 2 min to confirm that they were nesting and that no unattached males were going to spontaneously approach the pair. We then guided a nearby, unattached male toward the nesting pair from 1–10 m away. Once this male joined the nesting pair as a satellite, we observed the behavior of the nesting female. If she left the beach in less than 8 min after the addition of the satellite then she was considered “intolerant” and if she stayed past 8 min she was considered “tolerant.” We chose the 8-min threshold because a previous study at the same site demonstrated that 8 min was the median time for a monandrous female to leave the beach when a satellite approached (Johnson and Brockmann 2010). We compared egg development (defined above) from each intolerant and tolerant monandrous female when crossed with her attached male compared with a random satellite male (Figure S2-B; conducted 13 March to 25 April 2009). The eggs from each female were divided into replicate containers with ~30 eggs (mean = 29.6 ± 0.18 SE; median = 30; range 19–38; N = 200) in each. The eggs were fertilized with 60 µL of ~107 sperm mL−1 from each male. A total of 25 blocks were conducted, with 2 replicate fertilizations performed for each cross. Experiment 3: breeding design—good genes and/or compatible genes The breeding design (Figure S2-C; conducted 9–26 March 2009) involved blocks of 4 × 4 factorial crosses (North Carolina Type II breeding design; Lynch and Walsh 1998). In each block, 4 randomly chosen females (2 monandrous; 2 polyandrous) were crossed Statistical analysis All analyses were conducted using R. v. 2.15.2 (R Core Development Team 2013) unless otherwise specified. To investigate how female status and male status influenced developmental success in Experiments 1 and 2, we used a generalized linear mixed-effects model (GLMM) with a logit-link function and binomial error structure. GLMMs were implemented in the lme4 package using the lmer function (Bates et al. 2009) in R. The response variable was a binomial vector where the binomial numerator was the number of developed eggs and the denominator the number of undeveloped eggs. We fitted an interaction between female status (monandrous or polyandrous) and male status (attached or satellite) as fixed effects, along with sperm concentration as a covariate (centered to facilitate the interpretation of interaction terms; Schielzeth 2010). We included female ID, male ID, year, block, and an additive dispersion parameter (to account for overdispersion in our data; Nakagawa and Schielzeth 2010) as random factors in both models. We looked for differences in male and female carapace size or condition using linear models and Wilcoxon tests. Note that we were missing sperm concentration data or size/condition data for a few individuals. For Experiment 3, we fitted a GLMM with a logit-link function to partition variance of each response variable (eggs developed, early metamorphosis, and mean juvenile size) into male, female, and their interaction. The “eggs developed” response variable was a binomial vector where the binomial numerator was the number of developed eggs and the denominator the number of undeveloped eggs. The “early metamorphosis” response variable was a binomial vector where the binomial numerator was the number of juveniles and the denominator the number of developed eggs. We included sperm concentration (centered by scaling) as a covariate and female status as a fixed effect. We fitted male, female, male × female, and block as random factors in the model and estimated variance components using a Laplace approximation (binomial response variables) or restricted maximum likelihood procedure. We used log likelihood ratio testing to evaluate the significance of the random effects. Standard errors of the variance components were estimated using ASReml v. 3.0.5 (VSN International, Hemel Hempstead, UK). We further examined relationships between the proportion of eggs developed, proportion of early metamorphosis, and juvenile size using correlation coefficients (following a similar method to Evans Behavioral Ecology 1222 et al. 2007). Correlations of the entire data set are subject to pseudoreplication (because there were only 4 uniquely independent male × female combinations per block). We averaged response variables across the 2 replicate samples for each male × female combination and conducted 20 000 randomizations in which we randomly drew 4 uniquely independent male × female combinations from the 11 blocks (a total of 4 194 304 possible combinations, n = 44 for each combination). We calculated the mean and 95% confidence limits for the correlation coefficients generated by the possible combinations. Finally, we tested whether there were differences in carapace size or condition between the different male and female types. Results Experiment 1. Postmating benefits for monandrous and polyandrous females The difference in developmental success between attached and satellite males was greater for polyandrous females than for monandrous females (GLMM logit estimate ± SE = 0.759 ± 0.191, z = 3.977, P < 0.0001; Figure 1). This difference is driven by the significant difference in developmental success for eggs from polyandrous females that were fertilized by sperm from their attached males or from satellite males (GLMM logit estimate ± SE = −0.804 ± 0.141, z = −5.719, P < 0.0001; Figure 1). In contrast, there was no significant difference in developmental success of eggs from monandrous females that were fertilized by sperm from their attached male or from satellite males (GLMM logit estimate ± SE = 0.045 ± 0.129, z = 0.352, P = 0.72; Figure 1). Hence, we found a marginally nonsignificant difference between monandrous and polyandrous females when crossed to their attached males (GLMM logit estimate ± SE = −0.679 ± 0.361, z = −1.880, P = 0.06; Figure 1) but no significant difference when crossed to satellite males (GLMM logit estimate ± SE = −0.080 ± 0.343, z = −0.233, P = 0.816; Figure 1). There was no significant effect of sperm concentration on developmental success in our model (GLMM logit estimate ± SE = 0.007 ± 0.058, z = 0.124, P = 0.902). See Table S1 for estimates of random effects that were included in the model. Figure 1 Experiment 1: Mean (± SE) developmental success with attached (black bars) and satellite (white bars) males for monandrous (left) and polyandrous (right) females. A total of 21 females of each type were crossed to their respective attached males and to 2 random satellite males. The difference in carapace size between satellite males and the attached males of monandrous females was not significantly different than the size difference between satellites and the attached males of polyandrous females (generalized linear model [GLM] estimate ± SE = 0.42 ± 0.53, t = 0.802, P = 0.42, df = 67). Likewise, there was no significant difference in condition between satellites and the attached males of monandrous females (Wilcoxon Test Statistic = 205.5, P = 0.12, N = 34) or satellites and the attached males of polyandrous females (Wilcoxon Test Statistic = 196, P = 0.21, N = 35). There was also no significant difference in carapace size (GLM estimate ± SE = 0.34 ± 0.55, t = 0.63, P = 0.54, df = 41) or condition (Wilcoxon Test Statistic = 232, P = 0.77, N = 42) between monandrous and polyandrous females. See Table 2 for descriptive statistics. Experiment 2. Postmating benefits for tolerant and intolerant females The developmental success of eggs from intolerant females was not significantly different when using sperm from their attached males (45.2 ± 3.43% SE, N = 25) compared with sperm from random satellite males (48.1 ± 4.39% SE, N = 25; GLMM logit estimate ± SE = −0.143 ± 0.120, z = −1.197, P = 0.231). Likewise, there was no significant difference when eggs from tolerant females were fertilized by their attached males (48.1 ± 3.42% SE, N = 25) compared with random satellite males (43.4 ± 3.34% SE, N = 25; GLMM logit estimate ± SE = −0.188 ± 0.122, z = −1.534, P = 0.125). The difference in developmental success between eggs fertilized by attached versus satellite males was marginally greater (though nonsignificantly) for tolerant females than for intolerant females (GLMM logit estimate ± SE = 0.330 ± 0.170, z = 1.939, P = 0.052). Overall, we found no significant difference between tolerant and intolerant females when their eggs were fertilized by attached males (GLMM logit estimate ± SE = −0.144 ± 0.238, z = −0.603, P = 0.546) or satellite males (GLMM logit estimate ± SE = −0.188 ± 0.241, z = −0.775, P = 0.438). There was a Table 2 Mean (± SE) carapace and median condition of horseshoe crabs used in each experiment Variable Carapace width (cm), Mean ± SE Condition indexa, Median N (missing) Monandrous attached males Monandrous satellite malesb Polyandrous attached males Polyandrous satellite males Monandrous females Polyandrous females Intolerant attached males Tolerant attached males Satellite malesc Tolerant females Intolerant females Monandrous females Polyandrous females 16.25 ± 0.24 15.75 ± 0.31 16.45 ± 0.23 16.37 ± 0.27 21.62 ± 0.30 21.96 ± 0.45 15.77 ± 0.26 16.09 ± 0.17 16.43 ± 0.17 21.2 ± 0.22 21.57 ± 0.28 21.80 ± 0.30 21.8 ± 0.29 17.5 16 17 16 15 16 18 16 15 16 14 16 17 20 (1) 14 (1) 21 14 21 21 24 (1) 25 22 (3) 24 (1) 25 22 22 The sample size, N, for each descriptive statistic and the number of missing observations (missing) are provided. aCondition index ranging from 4 to 20, with the highest condition individuals receiving a 20. bSatellite males that were compared with monandrous females in this experiment. cSatellite males that were used in Experiment 2. Johnson and Brockmann • Indirect benefits of polyandry in horseshoe crabs 1223 marginally nonsignificant effect of sperm concentration in our model (GLMM logit estimate ± SE = 0.114 ± 0.067, z = 1.703, P = 0.089), suggesting that developmental success tended to increase with increasing sperm concentration, though the mean difference in sperm concentration between male types was minor (Table 1). See Table S2 for estimates of random effects that were included in the model. There was no significant difference in carapace size (GLM estimate ± SE = 0.37 ± 0.36, t = 1.03, P = 0.31, df = 48) or condition (Wilcoxon Test Statistic = 367, P = 0.18, N = 42) between tolerant and intolerant females. In contrast, we detected a marginally significant difference in carapace size between the attached males of intolerant females and the random satellite males used in this design (Student’s t-test, P = 0.03) but no significant difference in carapace size between these same satellite males and attached males of tolerant females (P = 0.26). Similarly, we found a significant difference in condition between the attached males of intolerant females and the random satellite males we used (Wilcoxon test, Z = −2.74, P = 0.006) but no significant difference between the attached males of tolerant females and these same satellite males (Z = 1.30, P = 0.19). Finally, as in our previous work (Johnson and Brockmann 2012), we found a significant difference in condition between the attached males of tolerant and intolerant females (Z = −2.00, P = 0.046). See Table 2 for descriptive statistics. success (measured at 20 days) ranged from 0% to 84.5% (mean = 42.8 ± 1.42 SE, N = 176). Metamorphosis by 45 days ranged from 0% to 63.6% (mean = 6.45 ± 0.76 SE, N = 171). Juvenile size (carapace width in millimeter) ranged from 4.51 to 6.10 mm (mean = 5.05 ± 1.49 SE, N = 166). Sperm concentration of the males used in the experiment ranged from 2.84 × 109 to 9.41 × 109 sperm mL−1 (mean = 6.56 × 109 ± 1.21 × 108 SE, N = 176). There was no significant effect of sperm concentration on developmental success or juvenile size, but we did detect a marginally nonsignificant, positive effect on early metamorphosis (Table 3). There was no significant effect of female status on developmental success or juvenile size, but we did detect a marginally nonsignificant effect on early metamorphosis (Table 3), with a mean of 8.02 ± 1.24% of offspring from monandrous females (N = 85) offspring reaching early metamorphosis compared with only 4.70 ± 0.72% of the offspring of polyandrous females (N = 86). For developmental success, Male, Female, and Male × Female effects explained 24.22% of the phenotypic variation (Table 3). The factorial analysis revealed a particularly strong and significant Female effect on developmental success (Table 3), accounting for 20.85% of the variance for this trait, which includes both potential additive genetic effects and maternal effects such as egg nutrients (Pitcher and Neff 2007). The Male variance component was nonsignificant, indicating that there were no additive genetic effects on developmental success. The Male × Female variance component was significant, accounting for 2.68% of the phenotypic variance in developmental success, suggesting there were nonadditive genetic effects on developmental success. Experiment 3. Breeding design: good genes and/ or compatible genes In Experiment 3, we had 3 measures of developmental success for the 10 578 eggs assayed from 176 crosses. Developmental Table 3 Experiment 3: sources of variation in 3 measures of reproductive success from in vitro fertilization experiments: fertilization (and early developmental) success, proportion of embryos that metamorphosed into juveniles within 45 days, and juvenile size measured at 60 days Developmental success (eggs developed by day 20) Early metamorphosis (juveniles by day 45) Mean juvenile size (mm) Variance components Likelihood ratio test Effect Variance LogLik Sperm concentration Female status Block (11) Male (44) Female (44) Male × Female (176) Error Sperm concentration Female Status Block (11) Male (44) Female (44) Male × Female (171) Error Sperm concentration Female status Block (11) Male (44) Female (42) Male × Female (166) Error GLMM estimate = 0.062 ± 0.052, z = 1.219, P = 0.233 GLMM estimate = 0.298 ± 0.295, z = 1.008, P = 0.313 0.020 0.142 −430.17 0.026 0.030 0.021 −431.69 3.05 0.908 0.245 −493.25 126.19 0.117 0.028 −458.12 55.926 π2/3 GLMM estimate = 0.272 ± 0.142, z = 1.920, P = 0.055 GLMM estimate = −0.580 ± 0.296, z = −1.959, P = 0.050 0.046 0.287 −290.73 0.059 0.425 0.240 −297.28 13.146 0.676 0.333 −304.08 26.744 0.310 0.162 −299.58 17.758 π2/3 0 GLMM estimate = −0.008 ± 0.033, t = −0.63, P = 0.533 GLMM estimate = 0.0137 ± 0.0375, t = 0.37, P = 0.717 0.0042 0.0038 72.784 2.012 0.0008 0.0014 73.652 0.276 0.0105 0.0038 63.476 20.629 0.0017 0.0031 73.733 0.116 0.0283 0.0034 SE χ2 P 0.872 0.081 <0.0001 <0.0001 0.808 0.0003 <0.0001 <0.0001 0.156 0.599 <0.0001 0.734 Phenotypic variation (%) 0.69 20.85 2.68 75.77 9.02 14.35 6.58 70.05 1.94 25.42 4.12 65.52 Developmental success and early metamorphosis were modeled as binomial data. Male, female, and block were treated as random effects in the model. Variance components were estimated using Laplace approximation (developmental success and early metamorphosis) or restricted maximum likelihood (juvenile size). Standard errors (SE) are square roots of the large samples’ variances calculated using ASReml. Significance of variance components was tested using likelihood ratio tests. Phenotypic variance (%) was based on the total phenotypic variance excluding block as this was considered a nongenetic/design factor. Note that π2/3 ≈ 3.3 is the underlying residual variance of a binomial model with logit link function (Nakagawa and Schielzeth 2010). Values in boldface represent significant effects. Behavioral Ecology 1224 For early metamorphosis (the number of individuals that metamorphosed by 45 days), Male, female, and Male × Female component explained 29.95% of the phenotypic variance (Table 3). The Female and Male and Male × Female components were all significant, representing 14.35%, 9.02%, and 6.58% of phenotypic variance, respectively, in early metamorphosis. Finally, for juvenile size, Male, Female, and Male × Female explained 31.47% of the phenotypic variance (Table 3). The Male component and the Male × Female component were nonsignificant, indicating no or little additive or nonadditive genetic effects on juvenile size. Female effects accounted for ~26% of the phenotypic variance in juvenile size. We also examined the relationships between developmental success, early metamorphosis, and juvenile size. We found a positive relationship between developmental success and early metamorphosis (Table 4). In contrast, we found no significant relationship between developmental success and juvenile size or between early metamorphosis and juvenile size (confidence interval at α = 0.05 included zero; Table 4). There was no significant difference between monandrous and polyandrous females in carapace size (GLM estimate ± SE = 0.26 ± 0.42, t = 0.63, P = 0.53, df = 43) or condition (Wilcoxon Test Statistic = 215.5; P = 0.54, N = 44). See Table 2 for descriptive statistics. Discussion Experiment 1 demonstrated that polyandrous females had lower developmental success with their attached male compared with random satellite males, whereas monandrous females had equal success with their attached male and random satellite males (Figure 1). This result supports our hypothesis that there are postmating, indirect genetic benefits accruing to polyandrous females from multiple mating through improvements in offspring development, whereas monandrous females would have no such gains were they to mate with satellite males. This result helps to explain why polyandrous females attract satellite males when the presence of those males is costly, that is, when females nest longer and lay more eggs per nesting attempt without satellites (Johnson and Brockmann 2010). The previous study also demonstrated experimentally that multiple mating would be costly for monandrous females if they were to attract Table 4 Experiment 3: mean correlation coefficients for relationships between proportion of eggs developed, proportion of early metamorphosis, and mean juvenile size from NCII breeding design Variables Proportion eggs developed and proportion early metamorphosis Proportion eggs developed and juvenile size Proportion early metamorphosis and juvenile size Mean correlation coefficient (95% confidence limitsa) 0.279 (0.055, 0.444) −0.086 (−0.374, 0.199) 0.088 (−0.126, 0.307) Mean correlation coefficients from 20 000 randomizations of unique male × female crosses. Each correlation coefficient has a sample size of 44, the number of independent values that can be extracted from the 4 × 4 factorial units across 11 blocks. aConfidence intervals of mean correlation coefficients that include zero are nonsignificant. satellites and here we show that they derive no additional, compensatory benefits through improved developmental success of their offspring as do the polyandrous females when nesting with satellites. We conducted Experiment 2 because we thought that monandrous females might be a heterogeneous grouping in which tolerant monandrous females might be polyandrous females that had not as yet attracted satellite males. We found, however, that tolerant and intolerant monandrous females did not differ in success when their eggs were fertilized by their attached male compared with random satellite males. If anything, the trend was in the opposite direction: developmental success tended to be higher when eggs were fertilized by satellite males compared with attached males of intolerant females, whereas developmental success tended to be lower for satellite males compared with the attached males of tolerant females. Taken together then, this may mean that only truly polyandrous females (those that have attracted satellites) gain from satellite male fertilizations and that intolerant females are somehow better able to resist sexual harassment during nesting than tolerant females (Johnson and Brockmann 2010). Alternatively, there may be some other compensating benefit, other than developmental success for tolerant females. Our results from Experiment 3 demonstrated that both genetic compatibility and good genes may be important factors in the benefits that polyandrous females derive from mating with multiple males. This is particularly interesting as our breeding design was conducted with random satellite males and not with the female’s attached males (NCII design assumes individuals are randomly selected from the population; Lynch and Walsh 1998). Experiment 3 focused on 3 potential fitness traits in the development of horseshoe crab offspring: egg development by day 20 (i.e., developmental success), successful metamorphosis by 45 days (i.e., early metamorphosis), and early juvenile size. Our analyses revealed highly significant Female variance components for all 3 measures, with the largest variance observed for developmental success (Table 3). The strong influence of female effects observed for developmental success is consistent with a large body of work on other marine broadcast spawners (Marshall and Evans 2005; Evans et al. 2007; Levitan 2008) and may be due to consistent differences in egg size, egg quality, or egg maturity. In turn, such differences may affect both the likelihood that an egg will be fertilized and the chance that a fertilized egg will complete the early stages of development. There is no evidence that the eggs from monandrous and polyandrous females differ in size (Johnson and Brockmann 2012), and we have no reason to believe that the eggs used in our experiments were not mature. Females were collected while nesting; hence, the small subset of eggs collected from their gonopores was presumably the next batch ready to be spawned by these females. It is likely that these large maternal effects are due to differences in egg quality between females. Variation in egg quality may have several proximate explanations, such as differences in resources invested in egg production between females (Rudolfsen et al. 2005), but nothing is known about the effects of such variation on fertilization and/or developmental success in Limulus. As in other studies (Wedekind 1994; Evans et al. 2007; Ivy 2007), the influence of nonadditive and additive sources of variation depended on which trait was measured. We found a significant Male × Female interaction and a lack of a Male effect for developmental success, suggesting that no single male is best for all females; rather, there is individual compatibility resulting in enhanced developmental success. This finding is consistent with the genetic compatibility hypothesis because it suggests that postmating sexual selection favors compatible male–female crosses (García-González Johnson and Brockmann • Indirect benefits of polyandry in horseshoe crabs 1225 and Simmons 2005; Evans et al. 2007). That said, only 2.68% of phenotypic variation was attributed to nonadditive variation in this trait for females crossed to satellite males, and the main Male effect was marginally nonsignificant. These effects may have been stronger had we included attached males in our breeding design. Our analysis of the number of offspring that reached metamorphosis in 45 days (i.e., those that showed rapid development and early metamorphosis) revealed that the component of variance associated with Males was significant, accounting for ~9% of the observed phenotypic variance for this trait (Table 3). Likewise, we identified a significant component of variance associated with Male × Female interaction, accounting for 6.58% of the variation in this trait. We also found that early metamorphosis (by 45 days) and developmental success (at 20 days) were positively correlated; thus, male × female combinations that were most likely to result in successful egg development provided offspring with the fastest development through to metamorphosis (Table 3). This positive relationship suggests further support for postcopulatory female choice for compatible males and/or high-quality males (good genes). If our measure of developmental success represents true fertilization success, then this would mirror similar findings in other broadcast spawning species (Dziminski et al. 2008; Evans et al. 2007). It is important to note that although it is not known whether early metamorphosis has fitness implications for horseshoe crabs, it is reasonable to assume that the heavily armored juveniles that feed are more likely to survive than the nonfeeding, unarmed larvae. Consequently, variation in settlement rates is likely to be an important target for natural selection in horseshoe crabs. The significant Male effect on early metamorphosis suggests that postmating sexual selection is acting on females when they are paired with a low-quality male. This result is particularly interesting as we only used satellite males in our crosses; hence, we would expect to see even larger male effects had we conducted crosses with attached males as in Experiment 1. Unfortunately, we have little information about what constitutes a “good quality” male horseshoe crab. We did observe a marginally nonsignificant effect of sperm concentration on early metamorphosis, such that early metamorphosis tended to increase with increasing sperm concentration. Sperm concentration could be an indication of quality in horseshoe crabs but could also reflect the degree of sperm competition a male encounters (Simmons et al. 1999; Neff et al. 2003; Sasson et al. 2012). Interestingly, we also found a marginally significant effect of female status on early metamorphosis (P = 0.05), with polyandrous females having a somewhat lower proportion of their offspring achieving early metamorphosis than monandrous females (8% vs. 4.7%). This difference is consistent with our previous research that demonstrated condition-dependent differences between monandrous and polyandrous females and context-dependent differences depending on the male that was attached to the female (Johnson and Brockmann 2012). Finally, our analysis of juvenile size revealed no significant Male or Male × Female interaction effects, suggesting that neither good genes nor compatible genes effects are important for this trait. However, we only measured 3 individuals per male–female cross, which may have obscured our ability to detect variation in this trait. It is unclear why our mean developmental rate is only ~50% in our in vitro fertilization experiments, but this value is consistent with other studies using similar methods (Brown and Knouse 1973). If all eggs were fertilized, then developmental success reflects offspring viability, and we presume that genetic elements during the early developmental phase are responsible for the variation that we observed in developmental success (García-González 2008). If on the other hand, developmental success represents true fertilization success, then gamete quality or gamete interactions at zygote formation are likely responsible for the variation in egg development that we observed. Moreover, differential viability and fertilization success are not mutually exclusive; hence, our measure of developmental success could be due to genetic elements during the early developmental phase and to factors that affect zygote formation during the fertilization phase (García-González 2008). If developmental success does reflect fertilization success, then possible explanations for a low fertilization rate include sperm– egg ratio, maturity of eggs or sperm, lack of oviducal fluids, the electrical stimulation used to obtain sperm and eggs, timing of fertilizations, or salinity. We used sperm concentrations that were optimum based on our initial serial dilution experiment (Figure S1). There is no evidence for polyspermy in this species (Brown and Knouse 1973), and fertilization increased with increasing sperm concentrations in our serial dilution experiment (Figure S1). Sperm may have been limiting, though we did observe development between 80% and 90% in a number of crosses. Further, we would have expected to see a stronger influence of sperm concentration on developmental success if sperm were limiting in our experiments. There is no reason to think that the small amount of eggs and sperm obtained for our in vitro fertilization experiments would be immature as they are nearest to the gonopore and presumably would have been naturally spawned if the animals had been allowed to continue nesting. It is possible that the electrodes may have damaged eggs during collection, but there is no evidence to suggest that this is an issue (Clapper and Brown 1980a, b). The timing of fertilizations could lead to lower fertilization success if there were a narrow window for egg receptivity (Brown and Clapper 1980), but our experiments were conducted within the established range for fertilizing eggs (Brown and Knouse 1973). Salinity is known to influence development in horseshoe crabs (Ehlinger and Tankersley 2004), with the optimal range of 25 and 30 ppt (Ehlinger and Tankersley 2003). Our experiments were conducted with 33–35 ppt seawater, so this may have contributed to lower fertilization success. Regardless, our protocol was the same across all experiments, so this could not be an explanation for the differences we found. Our in vitro fertilization methods result in oviducal fluids being removed or diluted, though we did not observe any appreciable amount of fluid, except in a few older females that were in poor condition (personal observation). Studies have previously demonstrated that oviducal fluids affect sperm motility in both competitive and noncompetitive situations, primarily in fish (Rosengrave et al. 2008; Gasparini and Pilastro 2011). Horseshoe crab sperm are not activated by seawater (Clapper and Brown 1980a), but rather likely by a low molecular weight molecule that diffuses from the egg (Clapper and Brown 1980b); hence, it is unclear whether oviducal fluids play a role in sperm dynamics in this system. It would, however, be difficult to measure the importance of ovarian fluid on sperm dynamics and fertilization success in a biologically meaningful way as horseshoe eggs are fertilized externally in a sand-water slurry under the nesting pair. A potential issue with the interpretation of indirect genetic benefits in our experiments is that we were unable to measure fertilization success at the very early stages of egg development; hence, our presumed indirect genetic benefits might actually reflect direct fertilization benefits (Evans and Marshall 2005; García-González and Simmons 2007). Horseshoe crab eggs are surrounded by opaque egg Behavioral Ecology 1226 membranes making it impossible to see the earliest stages of development without the use of dyes or dissection (Sekiguchi et al. 1982); hence, we can only infer fertilization success based on whether the eggs develop or not. We have several reasons for favoring indirect genetic effects of multiple mating over direct fertilization success. First, our field data suggest that the presence or addition of multiple males does not influence the number of eggs that develop (Johnson and Brockmann 2010). Second, there was no effect of sperm concentration on developmental success in any of our experiments. Third, we have conducted experiments where we fertilize eggs with 1 versus 2 male’s sperm and have found no difference in the proportion of eggs that developed (unpublished data), suggesting that there is no evidence that multiple mating alone directly increases fertilization success in our study system. Nonetheless, it is possible that some of the differences in our measures of developmental success may be due to differences in the direct benefit, fertilization success. Implications for evolution of polyandry in horseshoe crabs It is not clear what if any precopulatory mate choice occurs at the time of amplexus (attachment of male to female, which occurs offshore). Amplexus could simply be driven by male choice, and because females are limited in the time available for spawning and because they have to have an attached male in order to spawn (Brockmann 2003b), they may have little opportunity to engage in premating mate choice. If this is the case, as seems likely, then the best way for females to control mate choice may be after pairing through their control of polyandry. Because the eggs of polyandrous females have greater developmental success when fertilized by satellites rather than by their attached males (Experiment 1) and because we found significant evidence for potential indirect benefits (Experiment 3), this suggests that females may actively discriminate and choose to be monandrous or polyandrous based on the quality of the male that is attached to them or based on their own genotype. If so, how is this achieved? How could females acquire information about the quality of their attached males? We did not detect size or condition differences between the attached males of monandrous and polyandrous females in this study or in our previous work (Johnson and Brockmann 2012), but the attached males of intolerant females were in significantly better condition than attached males of tolerant females (Johnson and Brockmann 2012), although we found no difference in the developmental success of eggs fertilized by these males (Experiment 2). Perhaps females are able to assess quality based on a male’s sperm traits. Although we found no significant differences in sperm concentration in this study, recent work using a larger sample size of all males from the same population has found that the attached males of polyandrous females have a lower sperm concentration on average than the attached males of monandrous females (Sasson et al. unpublished data). Is it possible that females know the quality of their attached male’s sperm and based on that information she attracts satellites or not? Schwab and Brockmann (2007) showed that satellites often join polyandrous females before the first eggs were laid (and therefore before sperm were ejaculated), but these could have been pairs that had nested together previously. Competitive fertilization experiments to assess whether sperm quality and/or quantity influence developmental success for monandrous and polyandrous females would be informative and provide a means to test whether a good sperm process is important in this system (Evans et al. 2007). Can females discriminate among attached males based on their own or the male’s genotype? Mechanisms underlying compatible gene effects are largely unknown but may include postcopulatory mate choice to prevent inbreeding or selfing, to increase heterozygosity or major histocompatibility complex compatibility, or to avoid selfish genetic elements, such as Wolbachia (Jennions and Petrie 2000; Tregenza and Wedell 2000; Zeh and Zeh 2003). Limulus polyphemus populations are highly heterozygous, and there is very little evidence of inbreeding (King et al. 2005), suggesting that compatible gene effects might be operating to limit inbreeding in this system. Additional studies assaying the relatedness of monandrous and polyandrous females compared with their attached males would shed further light on this hypothesis. Conclusion We have shown that polyandrous females crossed with their attached males have lower developmental success of their eggs than monandrous females crossed with their attached males. This suggests that the attached males of polyandrous females are of lower quality than the attached males of monandrous females, which may include intrinsic quality and/or compatibility. Our quantitative genetic breeding design identified significant nonadditive (compatibility) variation in developmental success and in early metamorphosis. We further identified significant additive (good genes) variation in early metamorphosis, suggesting that male intrinsic quality (good genes) may be an important source of variation in this trait. Because females come ashore with an attached male, and because females appear to be consistently monandrous or polyandrous across multiple nesting periods (suggesting females have some control over whether they are polyandrous or monandrous; Johnson and Brockmann 2012), we were able to conduct our experiments with naturally occurring monandrous and polyandrous females and natural mating pairs. If females have knowledge about the quality or compatibility of their attached males and control over polyandry, then we would expect that a female should solicit additional mates when her attached male’s sperm are not compatible with her eggs or when he is of low quality. Our results suggest that incompatibilities between males and females and good genes may provide a potential selective advantage that offsets the costs of multiple mating (Johnson and Brockmann 2010) in this system. Whether a mating system is dominated by mate choice for good genes or by compatibility will have a profound effect on the course of sexual selection and the maintenance of genetic variation in the population. Like other studies (Wedekind et al. 2001, 2008; Bang et al. 2006; Evans et al. 2007; Ivy 2007; Pitcher and Neff 2007), we have found that both good genes and compatibility play a role in the horseshoe crab mating system, but we know little about how the 2 interact over time (Neff and Pitcher 2005; Puurtinen et al. 2009). Additional in vitro studies assaying sperm performance under competitive conditions combined with estimating male fertilization success would be useful in evaluating the contribution of both additive and nonadditive genetic variation to offspring success and the maintenance of polyandry in this system. Finally, the considerable female effects on development that we observed in Experiment 3 deserve further consideration. SUPPLEMENTARY MATERIAL Supplementary Figures S1 and S2 and Table S1 can be found at http://www.beheco.oxfordjournals.org/. Johnson and Brockmann • Indirect benefits of polyandry in horseshoe crabs Funding This research was supported by the National Science Foundation (IOB 06-41750), the University of Florida Division of Sponsored Research, the Department of Biology, and the UF Marine Laboratory at Seahorse Key. We thank our field assistants: N. Montes, N. Williams, S. Kromrey, N. Eliazar, L. Riegler, S. Steininger, and T. Sentner. We also thank D. Sasson, M. Smith, and K. Saunders for their scholarly input and assistance in the field. Numerous undergraduate volunteers assisted with field work and rearing. We thank K. Dodds, S. Nakagawa, and A. Senior for help with statistical analyses. We also thank our editor and two anonymous reviewers whose comments considerably improved this manuscript. Our research was conducted under a permit from the Lower Suwannee National Wildlife Refuge, J. Kasbohm, refuge manager. We thank H. Lillywhite, H. Coulter, and Al Dinsmore of the UF Marine Laboratory at Seahorse Key for their help with field logistics. Handling editor: Paco Garcia-Gonzalez References Arnqvist G, Rowe L. 2005. Sexual conflict. Princeton (NJ): Princeton University Press. Bang A, Grønkjær P, Clemmesen C, Høie H. 2006. Parental effects on early life history traits of atlantic herring (Clupea harengus L.) larvae. J Exp Mar Biol Ecol. 334:51–63. Barlow R, Powers M. 2003. Seeing at night and finding mates: the role of vision. In: Shuster CN, Barlow RB, Brockmann HJ, editors. The American horseshoe crab. Cambridge (MA): Harvard University Press. p. 83–102. Bates D, Maechler M, Bolker B. 2009. lme4: linear mixed-effects models using S4 classes R package version 0999999-2. Available from: http:// cran.r-project.org/web/packages/lme4/. Birkhead TR, Hosken DJ, Pitnick S. 2009. Sperm biology: an evolutionary perspective. Burlington (MA): Academic Press. Birkhead TR, Møller AP. 1998. Sperm competition and sexual selection. New York (NY): Academic Press. Brockmann HJ. 2003a. Nesting behavior: a shoreline phenomenon. In: Shuster CN, Barlow RB, Brockmann HJ, editors. The American horseshoe crab. Cambridge (MA): Harvard University Press. p. 33–49. Brockmann HJ. 2003b. Male competition and satellite behavior. In: Shuster CN, Barlow RB, Brockmann HJ, editors. The American horseshoe crab. Cambridge (MA): Harvard University Press. p. 50–82. Brockmann HJ, Colson T, Potts W. 1994. Sperm competition in horseshoe crabs (Limulus polyphemus). Behav Ecol Sociobiol. 35:153–160. Brockmann HJ, Johnson SL. 2011. A long-term study of spawning activity in a florida gulf coast population of horseshoe crabs (Limulus polyphemus). Estuar Coasts. 34:1–19. Brockmann HJ, Nguyen C, Potts W. 2000. Paternity in horseshoe crabs when spawning in multiple-male groups. Anim Behav. 60:837–849. Brown GG, Clapper DL. 1980. Cortical reaction in inseminated eggs of the horseshoe crab, Limulus polyphemus. Dev Biol. 76:410–417. Brown GG, Knouse JR. 1973. Effects of sperm concentration, sperm aging, and other variables on fertilisation in the horseshoe crab, Limulus polyphemus. Biol Bull. 144:462–470. Byrne PG, Roberts JD. 2004. Intrasexual selection and group spawning in quacking frogs (crinia georgiana). Behav Ecol. 15:872–882. Clapper DL, Brown GG. 1980a. Sperm motility in the horseshoe crab, Limulus polyphemus: I. Sperm behavior near eggs and motility initiation by egg extracts. Dev Biol. 76:341–349. Clapper DL, Brown GG. 1980b. Sperm motility in the horseshoe crab, Limulus polyphemus L. II. Partial characterization of a motility initiating factor from eggs and the effects of inorganic cations on motility initiation. Dev Biol. 76:350–357. Dziminski MA, Roberts JD, Simmons LW. 2008. Fitness consequences of parental compatibility in the frog Crinia georgiana. Evolution. 62:879–886. Eberhard W. 1996. Female control: sexual selection by cryptic female choice. Princeton (NJ): Princeton University Press. Ehlinger GS, Tankersley RA. 2003. Larval hatching in the horseshoe crab, Limulus polyphemus: facilitation by environmental cues. J Exp Mar Biol Ecol. 292:199–212. 1227 Ehlinger GS, Tankersley RA. 2004. Survival and development of horseshoe crab (Limulus polyphemus) embryos and larvae in hypersaline conditions. Biol Bull. 206:87–94. Evans JP, García-González F, Marshall DJ. 2007. Sources of genetic and phenotypic variance in fertilisation rates and larval traits in a sea urchin. Evolution. 61:2832–2838. Evans JP, Marshall DJ. 2005. Male-by-female interactions influence fertilisation success and mediate the benefits of polyandry in the sea urchin Heliocidaris erythrogramma. Evolution. 59:106–112. García‐González F. 2008. Male genetic quality and the inequality between paternity success and fertilisation success: consequences for studies of sperm competition and the evolution of polyandry. Evolution 62:1653–1665. García-González F, Evans JP. 2011. Fertilisation success and the estimation of genetic variance in sperm competitiveness. Evolution. 65:746–756. García-González F, Simmons LW. 2005. The evolution of polyandry: intrinsic sire effects contribute to embryo viability. J Evol Biol. 18:1097–1103. García-González F, Simmons LW. 2007. Paternal indirect genetic effects on offspring viability and the benefits of polyandry. Curr Biol. 17:32–36. Gasparini C, Pilastro A. 2011. Cryptic female preference for genetically unrelated males is mediated by ovarian fluid in the guppy. Proc Roy Soc B. 278:2495–2501 Hassler C, Brockmann HJ. 2001. Evidence for use of chemical cues by male horseshoe crabs when locating nesting females (Limulus polyphemus). J Chem Ecol. 27:2319–2335. Hosken DJ, Stockley P. 2003. Benefits of polyandry: a life history perspective. Evol Biol. 33:173–194. Ivy TM. 2007. Good genes, genetic compatibility and the evolution of polyandry: use of the diallel cross to address competing hypotheses. J Evol Biol. 20:479–487. Jennions MD, Petrie M. 2000. Why do females mate multiply? A review of the genetic benefits. Biol Rev. 75:21–64. Johnson SL, Brockmann HJ. 2010. Costs of multiple mates: an experimental study in horseshoe crabs. Anim Behav. 80:773–782. Johnson SL, Brockmann HJ. 2012. Alternative reproductive tactics in female horseshoe crabs. Ecol. 23:999–1008. Johnson SL, Yund PO. 2008. Multiple paternity and subsequent fusionrejection interactions in a kin-structured population. Mar Ecol Prog Ser. 364:129–134. Kempenaers B, 2007. Mate choice and genetic quality: a review of the heterozygosity theory. Adv Study Behav. 37:189–278. Kempenaers B, Verheyen GR, Vandenbroeck M, Burke T, Vanbroeckhoven C, Dhondt AA. 1992. Extra-pair paternity results from female preference for high quality males in the blue tit. Nature. 357:494–496. King TL, Eackles MS, Spidle AP, Brockmann HJ. 2005. Regional differentiation and sex-biased dispersal among populations of the horseshoe crab Limulus polyphemus. Trans Am Fish Soc. 134:441–465. Levitan DR. 2008. Gamete traits influence the variance in reproductive success, the intensity of sexual selection, and the outcome of sexual conflict among congeneric sea urchins. Evolution. 62:1305–1316. Lynch M, Walsh B. 1998. Genetics and analysis of quantitative traits. Sunderland (MA): Sinauer. Marshall DJ, Evans JP. 2005. The benefits of polyandry in the free-spawning polychaete Galeolaria caespitosa. J Evol Biol. 18:735–741. Marshall DJ, Evans JP. 2007. Context-dependent genetic benefits of polyandry in a marine hermaphrodite. Biol Lett. 3:685–688. Nakagawa S, Schielzeth H. 2010. Repeatability for Gaussian and non‐ Gaussian data: a practical guide for biologists. Biol Rev. 85:935–956. Neff BD, Fu P, Gross MR, 2003. Sperm investment and alternative mating tactics in bluegill sunfish (Lepomis macrochirus). Behav Ecol 14:634–641. Neff BD, Pitcher TE. 2005. Genetic quality and sexual selection: an integrated framework for good genes and compatible genes. Mol Ecol. 14:19–38. Neff BD, Pitcher TE. 2008. Mate choice for non-additive genetic benefits: a resolution to the lek paradox. J Theor Biol. 254:147–155. Newcomer SD, Zeh JA, Zeh DW. 1999. Genetic benefits enhance the reproductive success of polyandrous females. Proc Natl Acad Sci USA. 96:10236–10241. Pitcher TE, Neff BD. 2007. Genetic quality and offspring performance in chinook salmon: implications for supportive breeding. Conserv Genet. 8:607–616. Puurtinen M, Ketola T, Kotiaho JS. 2009. The good genes and compatible genes benefits of mate choice. Am Nat. 174:741–752. 1228 R Core Development Team. 2013. R: a language and environment for statistical computing. Vienna (Austria): R Foundation for Statistical Computing. Available from: http://www.R-project.org. Ridley M. 1988. Mating frequency and fecundity in insects. Biol Rev Camb Philos Soc. 63:509–549. Rodriguez-Munoz R, Tregenza T. 2009. Genetic compatibility and hatching success in the sea lamprey (Petromyzon marinus). Biol Lett. 5:286–288. Rosengrave P, Gemmell NJ, Metcalf V, McBride K, Montgomerie R. 2008. A mechanism for cryptic female choice in chinook salmon. Behav Ecol 19:1179–1185. Rudolfsen G, Figenschou L, Folstad I, Nordeide J, Søreng E. 2005. Potential fitness benefits from mate selection in the atlantic cod (Gadus morhua). J Evol Biol. 18:172–179. Sasson DA, Johnson SL, Brockmann HJ. 2012. The role of age on sperm traits in the american horseshoe crab, Limulus polyphemus. Anim Behav. 84:975–981. Saunders KM, Brockmann HJ, Watson W, Jury SH. 2010. Male horseshoe crabs Limulus polyphemus use multiple sensory cues to locate mates. Curr Zool. 56:485–498. Schielzeth H. 2010. Simple means to improve the interpretability of regression coefficients. Meth Ecol Evol 1:103–113. Schwab RL, Brockmann HJ. 2007. The role of visual and chemical cues in the mating decisions of satellite male horseshoe crabs, Limulus polyphemus. Anim Behav. 74:837–846. Sekiguchi K, Yamamichi Y, Costlow JD. 1982. Horseshoe crab developmental studies I. Normal embryonic development of Limulus polyphemus compared with Tachypleus tridentatus. In: Bonaventura J, Bonaventura C, Tesh S, editors. Physiology and biology of horseshoe crabs: studies on normal and environmentally stressed animals. New York (NY): Alan R. Liss, Inc. p. 53–73. Shuster CN Jr, Sekiguchi K. 2003. Growing up takes about ten years and eighteen stages. The American horseshoe crab. In: Shuster CN, Barlow RB, Brockmann HJ, editors. The American horseshoe crab. Cambridge (MA): Harvard University Press. Behavioral Ecology Simmons LW. 2005. The evolution of polyandry: sperm competition, sperm selection, and offspring viability. Ann Rev Ecol Ecol Syst. 36:125–146. Simmons LW, Tomkins J, Hunt J. 1999. Sperm competition games played by dimorphic male beetles. Proc Roy Soc Lond B 266:145–150. Slatyer RA, Mautz BS, Backwell PRY, Jennions MD. 2012. Estimating genetic benefits of polyandry from experimental studies: a meta-analysis. Biol Rev. 87:1–33. Thornhill R, Alcock J. 2001. The evolution of insect mating systems. Cambridge (MA): Harvard University Press. Tregenza T, Wedell N. 2000. Genetic compatibility, mate choice and patterns of parentage: invited review. Mol Ecol. 9:1013–1027. Trippel EA, Kraus G, Köster FW. 2005. Maternal and paternal influences on early life history traits and processes of baltic cod Gadus morhua. Mar Ecol Prog Ser. 303:259–267. Wedekind C. 1994. Mate choice and maternal selection for specific parasite resistances before, during and after fertilisation. Philos Trans R Soc Lond B Biol Sci. 346:303–311. Wedekind C, Evanno G, Urbach D, Jacob A, Müller R. 2008. ‘Good-genes’ and ‘compatible-genes’ effects in an alpine whitefish and the information content of breeding tubercles over the course of the spawning season. Genetica. 132:199–208. Wedekind C, Müller R, Spicher H. 2001. Potential genetic benefits of mate selection in whitefish. J Evol Biol. 14:980–986. Wedell N, Tregenza T. 1999. Successful fathers sire successful sons. Evolution. 53:620–625. Yasui Y. 1998. The ‘genetic benefits’ of female multiple mating reconsidered. Trends Ecol Evol. 13:246–250. Zeh JA, Zeh DW. 1996. The evolution of polyandry I: intragenomic conflict and genetic incompatibility. Proc Roy Soc B. 263:1711–1717. Zeh JA, Zeh DW. 1997. The evolution of polyandry II. Post-copulatory defences against genetic incompatibility. Proc Roy Soc B. 264:69–75. Zeh JA, Zeh DW. 2003. Toward a new sexual selection paradigm: polyandry, conflict and incompatibility (invited article). Ethology. 109:929–950.
© Copyright 2026 Paperzz