Facultative pupal mating in Heliconius erato: Implications for mate choice, female preference, and speciation

Abstract Mating systems have broad impacts on how sexual selection and mate choice operate within a species, but studies of mating behavior in the laboratory may not reflect how these processes occur in the wild. Here, we examined the mating behavior of the neotropical butterfly Heliconius erato in the field by releasing larvae and virgin females and observing how they mated. H. erato is considered a pupal‐mating species (i.e., males mate with females as they emerge from the pupal case). However, we observed only two teneral mating events, and experimentally released virgins were almost all mated upon recapture. Our study confirms the presence of some pupal‐mating behavior in H. erato, but suggests that adult mating is likely the prevalent mating strategy in this species. These findings have important implications for the role of color pattern and female mate choice in the generation of reproductive isolation in this diverse genus.


| INTRODUCTION
Animal mating systems establish which sex holds more power in mate choice (Shuster, 2009). This balance influences which traits are under sexual selection and arbitrates conflicts between the sexes. By observing mating behavior, we can begin to understand the evolutionary and ecological processes that generated the current mating system and predict how mating systems could influence evolution.
The colorful, mimetic butterflies in the genus Heliconius are an excellent system in which to explore mating strategies and the role they play in sexual selection, sexual conflict, and speciation. In terms of mating system, Heliconius butterflies are traditionally classified into two groups (Figure 1). About half of Heliconius species are considered adult mating, the prevalent mode of mating in butterflies: males approach and court adult females, who either reject or copulate with the male (Rutowski, 1984;Scott, 1972;Walters, Stafford, Hardcastle, & Jiggins, 2012).
The other half of Heliconius species exhibit a mating system known as pupal mating. First described in the late 1800s, pupal mating occurs when male butterflies copulate with females as they emerge from the pupal case (Deinert, Longino, & Gilbert, 1994;Edwards, 1881;Gilbert, 1975Gilbert, , 1976Gilbert, , 1991. This is an example of sexual coercion (sensu Clutton-Brock & Parker, 1995), as females are unable to reject male courtship attempts. Males may spend days waiting on the pupae and may compete with other males for space on and access to the female (Deinert et al., 1994;Edwards, 1881). In some cases, males even break the pupal case and insert their abdomen to begin mating before the female has fully emerged (Deinert et al., 1994;Gilbert, 1975;Sourakov, 2008). In this article, we distinguish between matings in which the male inserts his abdomen into the pupal case ("pharate matings") and those in which males do not insert their abdomen, but still mate with the female during or immediately after emergence when she is unable to resist ("teneral matings"), as some may consider only pharate matings "true" pupal mating (Sourakov, 2008;Walters et al., 2012). Pupal-mating species are monophyletic within the Heliconius phylogeny (Beltrán, Jiggins, Brower, Bermingham, & Mallet, 2007).
Only one other species of butterfly, the Lycaenid Jalmenus evagoras, is known to engage in pupal mating (Elgar & Pierce, 1988), though similar pupal guarding and mating behaviors are seen in other insect orders (Thornhill & Alcock, 1983).
As a coercive mating strategy, pupal mating appears to impose serious costs on females, potentially leading to strong sexual conflict. Fiercely competing males have been observed to injure females and even knock them from the pupal case to the ground (Edwards, 1881;Gilbert, 2003). Beyond the increased risk of injury or death, pupal mating seems to eliminate females' ability to actively select the "best" mate. Lack of choice could be costly if female reproductive output is based on male quality and there is variation in male quality. This is likely in Heliconius, as males (of both adult-and pupal-mating species) transfer a nutrient-rich spermatophore to the female during mating that females use for egg production (Boggs & Gilbert, 1983). However, they also transfer an anti-aphrodisiac pheromone to females to discourage remating (Estrada, Schulz, Yildizhan, & Gilbert, 2011;Gilbert, 1976;Schulz, Estrada, Yildizhan, Boppré, & Gilbert, 2008). Males can mate multiply, while female remating rates are estimated to be about 25% in adult-mating species and pupal-mating females are generally but not exclusively monandrous (Pliske, 1973;Walters et al., 2012).
Pupal mating may also be costly for males. Searching for and guarding pupae preclude foraging for food, and males have been observed to exhaust and starve themselves waiting on pupae (Deinert et al., 1994). Pupal mating could also influence what cues are most important for attraction and mate choice. Color pattern has long been considered an important part of species recognition and attraction in Heliconius butterflies (Crane, 1955). However, if males locate females as pupae when color pattern is absent or obscured, chemical or pheromonal cues from the host plant and pupae may be more important to mate choice Estrada, Yildizhan, Schulz, & Gilbert, 2010). Pupal mating is also likely to intensify male-male competition, with corresponding sexual selection on traits (e.g., wing size, olfaction, spatial memory) which would increase mating success.
Though many Heliconius are considered pupal-mating species (e.g., in Brown, 1981;Beltrán et al., 2007), formal study of this behavior is mostly limited to two species. Studies of wild populations of H. hewitsoni and H. charithonia have described a suite of searching and mate-guarding behaviors associated with pupal mating (Deinert, 2003;Deinert et al., 1994;Mendoza-Cuenca & Macías-Ordóñez, 2010). Insectary studies have also shown that H. charithonia males use host plants to find immatures and distinguish male and female pupae using chemical cues . However, Mendoza-Cuenca and Macías-Ordóñez (2010) also found evidence of adult mating in a population of H. charithonia with highly asynchronous female pupal emergence. Males with smaller wings, likely to be unsuccessful competing for female pupae, instead patrolled territories and were observed to mate with experimentally released adult virgin females (Mendoza-Cuenca & Macías-Ordóñez, 2010). This shows that, even in pupal-mating species, other modes of reproduction may occur.
Here, we study mating behavior in a Panamanian population of the red postman butterfly, Heliconius erato (L. 1758, Lepidoptera: Nymphalidae). Though nominally a pupal-mating species, H. erato F I G U R E 1 Simplified phylogeny of Heliconius butterflies. All members of the pupal-mating clade are shown (in orange), but for simplicity, only a subset of adult-mating species are presented here. Phylogenetic relationships following Kozak et al. (2015). Branch lengths are not scaled. Ecological data summarized from Brown (1981), though we note that classification of Heliconius as host-plant specialists and generalists at the species level is an oversimplification (see Section 4)  differs from other pupal-mating species across multiple behavioral, life history, and biogeographic axes (Brown, 1981;Beltrán et al., 2007;Walters et al., 2012; Figure 1). The published literature contains conflicting evidence about the extent to which pupal mating occurs in H. erato. Mating behavior has been best studied in insectaries, where some researchers have reported pupal mating (Gilbert, 1976

| Experiment 1: Pupal-mating observations
All experiments were carried out between February and May 2014 in Gamboa, Panamá, and nearby Soberanía National Park. To improve chances that pupae were discovered by males, we placed larvae, instead of pupae, on experimental plants at five sites and tracked them through pupation (see Supporting information for GPS coordinates).
Second, we chose experimental sites that were within one meter of the larval host-plant Passiflora biflora and near an adult food source, usually Lantana camara, in the hopes that our experimental sites would be quickly incorporated into the traplines, or daily routes, of adult individuals (Gilbert, 1991). Sites were between 250 m and 2.5 km apart.
At each site we hung potted P. biflora from a metal frame and applied grease and Tanglefoot (Tree Tanglefoot Co., Grand Rapids, MI) to the legs of the frame. This was an attempt to prevent ants, an important cause of larval mortality in Heliconius, from accessing the experimental plants (Smiley, 1985(Smiley, , 1986. After placing the plants, we waited 1 week before placing out larvae to allow discovery of the experimental plants.  (Gilbert, 1976;Walters et al., 2012). Recaptured females were deemed mated if they smelled strongly of the male-transferred odor and/or contained a palpable spermatophore. Mated females were removed from the population, while unmated females were released to be potentially recaptured.

| Experiment 1: Observations of mating behavior
Over the course of almost one hundred field days, we caught and marked 231 wild individuals (128 males and 103 females) in the study area. We recaptured 53 males (females were not individually marked), and the vast majority (50/53) of our recaptures were at the same study area, suggesting that movement of males between our experimental sites was rare. Male forewing length was not bimodally distributed (Figure 2), which could be interpreted

| Experiment 2: Release and recapture of virgin females
In addition to the nine females that went unmated at our experimental sites, we released 52 insectary-reared virgin females, for a total of 61 experimentally released virgin females. We recaptured 20 of these females, 19 of which had been mated. The sole unmated female was part of experiment 1 and was visited and courted by males upon emergence, but the males did not initiate copulation. We recaptured this female multiple times, and she was unmated on her final recapture 2 weeks after emerging.
Given the central place of H. erato in these studies, it is important to clarify how H. erato mate in natural populations. Ours is the first direct study of H. erato pupal-mating behavior in the wild. Although H. erato is part of the pupal-mating clade, we find that pupal mating is not obligate.
Males performed some behaviors associated with pupal mating (e.g., searching host plants, visiting, and perching on pupae); however, we observed no instances of pharate mating and only two instances of teneral mating in the 11 cases in which females successfully pupated and emerged. This was not because experimental pupae went undiscovered, but instead reflects a high rate of adult mating in the population.
Most experimentally released females were mated upon recapture, and we observed an adult mating at one of our experimental sites.
When we combine the evidence from studies of both captive and wild Heliconius pupal-maters, there is clear variation in the propensity for pupal-mating across the clade. In some situations, pupal mating seems dominant (Deinert et al., 1994;Gilbert, 1976;Mendoza-Cuenca & Macías-Ordóñez, 2010), while in others, adult mating is prevalent (McMillan et al., 1997;Walters et al., 2012, this study). What factors might promote pupal mating in some species and populations, or constrain it in others, to generate this variation?
Our behavioral observations can provide some insights into the factors which might influence pupal mating. Our experimental sites varied in the quality and abundance of both adult food sources and larval host plants. Sites with higher quality resources had both higher butterfly densities and more observations of behaviors associated with pupal mating (Table 1). Indeed, the two teneral matings we witnessed, both by the same male, occurred at a high-density site where adult food plant was abundant throughout the experimental period. High butterfly density may promote pupal mating, as individual males may be more likely to guard pupae and attempt pupal mating as a method to outcompete other males. Abundant food, similarly, might decrease the costs of guarding pupae by making it easier for pupal-guarding males to intermittently forage.
On the other hand, the high rates of larval and pupal mortality we found may constrain pupal mating. Only ~7% of the third and fourth instar larvae we placed out survived to adulthood, even with our attempts to control ant predation. This is similar to rates of mortality seen in experiments with adult-mating Heliconius, in which only ~15%-30% of larvae survived 2 days when transplanted to their natural host plants in the wild (Merrill, Naisbit, Mallet, & Jiggins, 2013;Smiley, 1985Smiley, , 1986. Such high mortality rates could make it unprofitable for males to repeatedly monitor larvae and guard pupae, any one of which has a very low probability of survival.  (Beltrán et al., 2007;Brown, 1981). Second, H. erato is a host-plant generalist, while most other pupal-mating species specialize on a single, or perhaps multiple closely related, species of Passiflora host plant (Benson, Brown, & Gilbert, 1975;Brown, 1981;Merrill et al., 2013). The other butterfly species that engages in pupal mating, J. evagoras, is also a host-plant specialist with gregarious larvae (Elgar & Pierce, 1988). specialists or generalists at the species level, as we do in Figure 1, is an oversimplification. Host-plant usage can vary across populations within a species and may be especially dependent on abundance and diversity of local Passiflora or the presence of competitor species. For example, H. melpomeme is a generalist at the species level (Brown, 1981), but in Panama, it feeds almost exclusively on P. menispermifolia (Merrill et al., 2013). At our study sites, H. erato feeds on at least three species of Passiflora (Merrill et al., 2013 (Beltrán et al., 2007), and color pattern has been shown to be an important cue for male mate choice in some of these species (Jiggins, Naisbit, Coe, & Mallet, 2001;Merrill et al., 2012).  Gilbert, 1976) would be particularly useful in helping us understand the origin and persistence of this strange and rare mating system.

ACKNOWLEDGMENTS
We thank the government of Panama for permits for research and collection and the Smithsonian Tropical Research Institution for support. We are grateful to Natasha Hinosa for assistance in the field.