Introduction

The degree to which natural and sexual selection respectively affect mating behavior is largely unknown in evolutionary biology, and few examples delineate the problem more clearly than the maintenance of nuptial gift size in Orthoptera. During mating, male bush-crickets (Tettigoniidae) transfer a variable (in size), yet often substantial, spermatophore to the female (for reviews see Gwynne 1990, 2001; Vahed 1998). When transfer is complete the pair uncouple and the female reaches under her abdomen and starts to consume the spermatophore (Boldyrev 1915). As the ejaculate (sperm and seminal fluid) discharges from the ampulla into the female, she consumes the spermatophylax, a large, sperm-free, gelatinous mass. After that, she consumes the ampulla and remaining ejaculate (Boldyrev 1915, Bowen et al. 1984).

Although the function of the ampulla to house the ejaculate is relatively clear, the role the spermatophylax plays in mating is more complicated. Two nonmutually exclusive hypotheses have been suggested for spermatophylax size (for reviews see Vahed 1998, Gwynne 2001). First, the ejaculate-protection hypothesis states that the spermatophylax is sexually selected by preventing the female from removing the ampulla prematurely (Gerhardt 1913, 1914; Boldyrev 1915) and therefore directly increasing a male's assurance in sperm competition in a dose-dependent manner (for reviews see Eberhard 1996, Vahed 1998, Gwynne 2001, Simmons 2001, Arnqvist & Rowe 2005). There may be additional benefits under this hypothesis – consumption of a large spermatophylax may reduce the speed at which a female will remate, thereby indirectly increasing the number of offspring and the number of ova that may be fertilised by the male (Gwynne 1986; Wedell & Arak 1989; Simmons & Gwynne 1991; Wedell 1993a, b; Vahed 2007), or may increase the chance of female survival until oviposition (e.g., Voigt et al. 2005, 2006). Males that produce relatively large spermatophores are also more likely to transfer more ejaculate and therefore succeed in sperm competition (for a review see Simmons 2001). A large ejaculate may also induce longer intermating refractory periods in females (Heller & Helversen 1991, Heller & Reinhold 1994, Lehmann & Lehmann 2000a, Vahed 2007), allowing males to father a greater share of eggs laid in the next oviposition (Gwynne 1986; Wedell & Arak 1989; Simmons & Gwynne 1991; Wedell 1993a, b). Under this hypothesis, spermatophylax size should covary with the size of the ampulla (Reinhold & Heller 1993, Wedell 1993a, Heller & Reinhold 1994) or the number of sperm.

Alternatively, the paternal investment hypothesis suggests that the spermatophylax is under natural selection to provide a positive nutritional effect on the donating male's progeny (Trivers 1972, Gwynne et al. 1984). In this case, spermatophylax size should correspond to a relative increase in fitness and/or quantity of offspring (Trivers 1972; Thornhill 1976; Simmons & Parker 1989; Gwynne 1986, 1988, 1990; Wedell 1991; Reinhold 1999) but is not expected to covary with ampulla size or sperm number (for reviews see Vahed 1998, Gwynne 2001).

Both natural and sexual selection functions of the spermatophore have been observed in tettigoniids, and are reflected in considerable interspecific variation in spermatophore size (Gwynne 1983, Wedell 1993a, Vahed & Gilbert 1996). Spermatophore mass ranges from about 2% of total male body mass (relative mass) (Acripeza reticulata, Wedell 1993a; Anonconotus alpinus, Vahed 2002) to about 40% (Ephippiger ephippiger; Busnel & Dumortier 1955), and sperm numbers range between 38,000 (Phaneroptera nana, Vahed & Gilbert 1996) and 37.3 million sperm (P. thessalicus, McCartney & Heller this issue, p. 227). With respect to spermatophore function it is clear that size variation has significant fitness implications for each sex and species.

Despite the likely benefits to males, producing large spermatophores is expensive, as they represent a loss in future reproductive potential (Simmons 1988a, 1990, 1995a; Heller & von Helversen 1991; Vahed 2007), the costs of which will vary with factors such as local growing conditions and diet (Halliday 1987, Simmons 1988a, Simmons et al. 1993).

The variation found in spermatophore size among species may be, at least partly, a consequence of phylogenetic relatedness (Gwynne 1995, Vahed & Gilbert 1996). Nevertheless, in an analysis of 19 bush-cricket genera, Wedell (1993a, 1994a) showed that interspecific differences in spermatophore size, spermatophylax mass and ampulla mass are largely influenced by diet. Controlling for phylogeny in 43 tettigoniid species, Vahed & Gilbert (1996) found that there was also a large residual variation in sperm number and spermatophore size. Vahed & Gilbert (1996) however, did not control for diet, and used laboratory-reared bush-crickets (Vahed 1994) — a condition that may affect sperm number (e.g., Reinhold 1994) and spermatophylax size (e.g., Heller & von Helversen 1991).

Comparisons among species within a genus can be particularly informative because many variables that are shared by congeners are held constant (Ridley 1983, Felsenstein 1985, Harvey 1991, Harvey & Pagel 1991). The aim of this study was to compare spermatophore and body-mass data from field observations within the diverse bush-cricket genus Poecilimon. Poecilimon species share a similar diet and morphology, and while we recognise that this genus does not represent the full diversity found in bush-crickets, we show here that variation in spermatophore size approximates family-wide variation, so variations in diet and relatedness are, to a certain extent, controlled for. In this paper, we test the ejaculate-protection and paternal-investment hypotheses in Poecilimon by examining the correlations between the spermatophore components: spermatophylax mass, ampulla mass and sperm number.

Methods

Poecilimon

Poecilimon Fischer, 1853, (Fig. 8) is a genus of barbistine bush-crickets (Phaneropterinae, Tribe Barbistini) (Orthoptera: Ensifera: Tettigoniidae). There are 128 currently recognized species and subspecies (Otte et al. 2005), with about 65 European species, mostly situated in the east Mediterranean (Heller 2004). While the current position of species within the Poecilimon clade is under constant review (e.g., Heller 2004, Heller & Lehmann 2004, Heller et al 2004, Heller 2006), the status of Poecilimon at the genus level is well supported (Ramme 1933, Bey-Bienko 1954, Heller 1984). Since the description of the genus in 1853 there has been no dispute about the homogeneity of this group (see references in Otte 1997). The nomenclature used here follows that of Otte et al. (2005), with additional species P. gerlindae (Lehmann et al. 2006), P. ege (Ünal 2005), and P. ukrainicus (Bey-Bienko 1951).

The genus Poecilimon is quite uniform in terms of behavior and life-history patterns. Notable exceptions include differences in how females consume the spermatophore, and timing of the active mating phase. Most Poecilimon species consume the spermatophylax directly from underneath the abdomen, where it remains attached to the ampulla. However, at least one species, P. erimanthos, detaches the spermatophylax from the ampulla before consumption. Most species used are nocturnal. Notable exceptions are P. erimanthos, P. mytilenensis, and P. werneri, which are predominantly active during the day. P. nobilis, P. affinis, and P. gracilis seem to be active both night and day (Heller & von Helversen 1993). All species are semelparous, have obligate diapause and most have a univoltine lifecycle. All the Poecilimon species employed eat flowers and leaves, so are foliovores when ordered into gross feeding categories, such as those given by Wedell (1994a): 1) omnivorous-predaceous, 2) seed eaters, and 3) foliovores.

Collection.—Previously published and unpublished data were compiled from a range of sources for 33 species (36 taxa, 62 independent observations) of Poecilimon to supplement the data we collected ourselves. All were found in Greece, Turkey, Italy, Slovenia or the Ukraine (see Appendix 1 for the location of each population). The data for several species were obtained from the paper by Vahed & Gilbert (1996). Although these authors did not present relative spermatophore, spermatophylax and ampulla mass, we calculated these percentages directly from the table in their paper (see below for calculations of relative mass). The sources for all novel data included here are appended to Table 1; the locations where they were observed are listed in Appendix 1. For 11 species, two (or more) independent measurements from different populations or different years were included (designated by Roman numerals), and two species were sampled at the subspecific level: P. veluchianus veluchianus, P. veluchianus minor, and P. jonicus jonicus, P. jonicus superbus, P. jonicus tessellatus. In all, 62 taxa-site-year combinations were collated from 36 taxa (Table 1, Appendix 1).

Table 1.

Mean male body mass and sperm number with relative and actual mean spermatophore, spermatophylax, ampulla masses and sperm number of 33 Poecilimon spp. (36 taxa, 62 independent observations) (n = number of individuals). Each species is listed with the describer and with reference to the collectors or source of publication (see key at bottom for reference). Some species with more than one independent observation are distinguished by Roman numerals. Status of observations: field observations (F); exlarvae specimens (EL) that were field-obtained, but allowed to mature in large cages in the location of the natural population; purely lab-reared (L) individuals. Relative sperm number (rel#) = sperm number / male body weight (μg). Dashes (-) indicate a lack of gathered information and on occasion data have been published more than once, so we refer to original publications.

Determination of male body mass, spermatophore size, and sperm number.—We separated field-caught juveniles (ex-field larvae) and field-caught adults (EL and F respectively, Table 1) into cages of each sex. Field-caught juveniles were separated until at least seven days after their imaginal moult, in order to ensure sexual maturity (Heller & Reinhold 1994). Field-caught adults were separated for at least three days prior to pairing, in order to ensure full receptivity (Heller & von Helversen 1991, Lehmann & Lehmann 2000b). Two exceptions to this were P. thessalicus I and P. v. minor III (taken from independent mating experiments) where individuals were paired immediately after they were collected. Some data were used from individuals that were reared in the laboratory (for example P. elegans, P. gracilis, Table 1). While their treatment and the experimental procedures were otherwise the same as those in the field, they are not included in final interspecific analyses.

For mating, pairs were typically placed in 500-ml containers and observed every 15 min or less until the female bore a spermatophore, which we then carefully removed with forceps for weighing. All weights were measured to the nearest 1 mg. In some cases, the measurements were made in the field from wild matings. Where possible, the spermatophore, spermatophylax and ampulla masses were measured immediately after mating. When this was not possible (for example, P. laevissimus IV), male weight loss and female weight gain (with the spermatophore attached) before and after mating were compared (Reinhold & Helversen 1997). If the difference between the male weight loss and female weight gain was larger than 20%, that datum was excluded (following the procedure of Heller & Reinhold 1994). On occasion, either the spermatophylax or the ampulla mass was not measured; in these cases the missing component was calculated as the difference between the full spermatophore mass and the mass from the known component.

Relative spermatophore mass was calculated as the percentage of male body mass for each individual, and then the mean for all individuals taken to calculate a species average. On occasion, the spermatophore mass and male body mass were taken from different males, so the average spermatophore mass was divided by the average male mass to give relative spermatophore mass.

After weighing, the ampulla was cut from the spermatophylax, added to a known quantity of water (between 1 and 5 ml depending on the organ size), and sliced with a scalpel. We further mixed the solution by passing it repeatedly through a syringe until the sperm had been suspended in the water and fully homogenised. A subsample was taken and the sperm counted on a field haemocytometer (Swift: Neubauer improved). Normally three subsamples were taken and the solution remixed before taking each new subsample. If there was a large variation between subsamples or the sperm was not evenly distributed over the slide, the solution was remixed and further subsamples taken. Sperm from a known volume (50 μl – 200 μl) were counted and multiplied by the appropriate dilution factor to give the total number of sperm for the entire ampulla. For P. mariannae a Coulter counter was used (for details of the method see Lehmann & Festing 1998). Relative sperm number was calculated as the number of sperm per mg of mean male body mass and expressed as sperm number ×10³.

Analysis.—Using data from multiple populations or seasons means that some species are over-represented and may inflate the contribution of those taxa in the analyses. However, full data sets with multiple species may give a better understanding of how the environment affects spermatophore size. Therefore, we restricted our use of the full data set to descriptive comparisons, and only performed analyses on reduced data sets that included only one of each taxa. Priority for removal was first given to observation location (i.e., field observations were preferred over lab observations) and then to sample size (Table 1). Unless otherwise stated, statistics with multiple observations removed are presented in text and figures.

P. mytilenensis is unusual as it has a greatly enlarged ampulla and a large variation in sperm number (between 6.3 and 15.8 million sperm, Heller et al. 2004). Data for the current paper were from laboratory-reared individuals for this species, although observations from the field show that this variation in size approximates that found in its natural environment. Our intention in this paper was to compare among field-observed animals, avoiding any confounds imposed by lab-reared species. However, in terms of taxonomy, P. mytilenensis is quite typical for Poecilimon and large variations in spermatophore components are likely to represent realistic variations within the genus. Preliminary analysis that included data from P. mytilenensis also indicated that its impact on our understanding of mating systems within Poecilimon required further exploration. We therefore duplicated all analyses a second time, with the inclusion of P. mytilenensis, in order to directly compare this with variations found in the rest of the genus.

To normalize the data, all variables were log₁₀ transformed prior to analysis unless otherwise stated. Two types of analysis were performed. First, the correlation coefficients between male body mass and each of spermatophore mass, spermatophylax mass, ampulla mass, and sperm number were calculated. Second, the overall effect of male body mass (MBM) was estimated for each parameter using least-squares regressions and the residuals for each population examined, to reveal cases where male investments were over or under expectation based on the overall allometric relationships. All data were analysed using SAS 9.1.3.

Results

Comparisons between Poecilimon and other Tettigoniidae.—The wide range in each spermatophore component within the genus Poecilimon approximates that occurring among the Tettigoniidae as a whole (Fig. 1., Poecilimon dataset not reduced). However, the smallest relative spermatophore size in Poecilimon is around 6.1% (P. laevissimus IV, Table 1), while some other tettigoniids have spermatophores that are even smaller than this: Mecopoda elongata and Meconema thalassinum, for example, have spermatophores that are barely 1% of male body mass, with little or no spermatophylax. Poecilimon have relatively large spermatophores (always >5% relative mass) and nearly always have a larger spermatophylax than an ampulla. Poecilimon mytilenensis (Fig. 1), however, is an exception with an unusually large ampulla (14.7 % relative mass) and a relatively small spermatophylax (8.2 % relative mass; see Heller et al. 2004 for details). The upper limits of spermatophylax size are similar between Poecilimon and tettigoniids in general, with P. thessalicus, P. ornatus and P. pergamicus, for example, and Steropleurus stali, producing spermatophylaces that represent between 25% to 28% of male body mass (Fig. 1).

Fig. 1.

Male body, spermatophylax and ampulla mass as proportions of combined mass in 29 Poecilimon species (solid circles, 31 taxa; n = 37) and 40 other tettigoniid species (open circles, see Vahed & Gilbert 1996 for details), showing that variation in Poecilimon approximates family-wide variation. The solid arrow points to P. mytilenensis, a species that has a remarkably large ampulla (Heller et al. 2004).

There is also a very large range in sperm number within Poecilimon, which could not be accounted for simply by body size (y = 1.11× − 2.73, F_{1, 26}= 7.706, p = 0.011, r² = 0.22; Fig. 2). In most tettigoniids sperm number follows body size quite closely (y = 1.12× − 3.11, F_1,29 = − 60.45, p<0.001, r² = 0.68), but in Poecilimon, sperm number ranged between about 200,000 sperm per spermatophore (P. hamatus, P. ikariensis, P. jonicus and P. werneri) to about 28 million (P. zimmeri), although P. thessalicus can reach 37.3 million sperm (McCartney & Heller unpub. data). Within other tettigoniids, sperm number ranges between 38,000 for Phaneroptera nana, to about 10 million for Pycnogaster inermis. Many species of Poecilimon had far more sperm than would be expected for their body size, based on the overall pattern within the tettigoniids (e.g., P. thessalicus, P. zimmeri and P. ege, Table 1), though there are also a few species with unusually low sperm counts for their size (e.g., P. jonicus and P. werneri).

Fig. 2.

Male body mass and sperm number within Orthoptera (open circles, Vahed & Gilbert (1996) and Poecilimon (solid circles). Male body mass explains little of the variation in sperm number within Poecilimon. In contrast, 68% of the sperm number is explained by male body mass in other Orthoptera.

Variation within Poecilimon.—Within Poecilimon there is a large range in both body mass and spermatophore size. P. hoelzeli, for example, is more than fifteen times the weight of P. pergamicus (Table 1) and produces an accordingly large spermatophore of up to 454 mg, compared to 18.1 mg in P. pergamicus. Within the genus, spermatophore mass is closely correlated with male body mass (y = 0.7545× + 1.24, F_1,35 = 59.255, p = 0.000, r² = 0.64, Table 2). Similarly, male body mass is closely correlated with spermatophylax mass (y = 0.86× − 0.44, F_1,30 = 72.20, p<0.001, r² = 0.71), and is significantly correlated with ampulla mass (y = 0.67× − 0.60, F_1,30 = 12.91, p = 0.001, r² = 0.31; Fig. 3).

Table 2.

Regressions between male body mass (MBM), spermatophore mass, spermatophylax mass, ampulla mass and sperm number among 33 species of Poecilimon (36 taxa, n=62). *=significant.

Fig. 3.

Spermatophore components are largely dictated by male body mass: the relationships between male body mass and both spermatophylax mass (solid circles) and ampulla mass (open circles) in 31 Poecilimon taxa.

Intraspecific variation.—In four taxa, P. erimanthos, P. hamatus, P. j. jonicus, and P. laevissimus, spermatophore size varied two-fold within populations among seasons, while the numbers of sperm per spermatophore remained relatively constant (Table 1). Most of this variation is attributable to spermatophylax mass rather than ampulla mass, apart from P. laevissimus, where the ampulla mass (actual and relative) also varied two-fold among seasons. P. affinis showed only a small range in relative spermatophore size (13 to 15%) among years and populations in field conditions, but there was a remarkable difference in sperm number between field and laboratory-reared individuals (4.4 million sperm and 21.6 million sperm respectively). In P. thessalicus, while body size varied between 442 and 610 mg over four seasons, spermatophore mass varied from 102 mg (23% relative mass) to 224 mg (36.7% relative mass), and sperm number showed a four-fold range from 3.9 ×10⁶ to 16.5 × 10⁶ sperm over the same period. The two subspecies of P. veluchianus have been sampled repeatedly from both laboratory and field-reared bush-crickets. In P. v. veluchianus spermatophore size varied a little from 150 mg to 212 mg (23% to 26% relative mass), but sperm number varied from 6.3 million sperm in laboratory-reared bush-crickets (Reinhold & von Helversen 1997) to 10.5 million sperm in the field (Reinhold 1994). Similarly, the relative spermatophore mass of P. v. minor varied from 17% − 20% of body mass but sperm number ranged from 3.4 million to 7.6 million.

Spermatophore components.—The previous sections demonstrate that there is a tendency for relative spermatophore size to increase with an increase in body size. However, there is considerable variation among the species in spermatophore investment that is independent of this general pattern. No significant relationship was found between male body mass and relative spermatophylax mass (y = −0.0004× + 16.22, F_1,29 = 0.06, p = 0.81, r² = 0.002), relative ampulla mass (y = −0.0009× + 4.35, F_1,29 = 1.48, p = 0.23, r² = 0.048), and relative number of sperm (x10³ per 1mg of male body mass, y = −0.0192× + 86.71, F_1,26 = 0.17, p = 0.68, r² = 0.007). Allowing for body size reveals that some species invest relatively much more heavily in some spermatophore components than other species (Fig. 4, Table 1). Spermatophore components show considerable variation with some small males producing large spermatophylaces, ampullae or sperm numbers, and some large males producing small spermatophylaces, ampullae or sperm numbers.

Fig. 4.

A large variation in the relative investment in spermatophore components: no relationships between male body mass and relative spermatophylax mass (black/solid circles), relative ampulla mass (grey/solid circles) and relative sperm number (open circles).

We found a significant correlation between the residuals of spermatophylax and ampulla mass (y = 1.1116×, F_1,30 = 23.79, p<0.001, r² = 0.46), although a substantial portion of the variance in residuals did not co-vary (Fig. 5). Including data from P. mytilenensis predictably decreased the relationship further (y = 1.10× + 0.023, F_1,31 = 16.26, p<0.001, r² = 0.36; Table 2). Surprisingly, residual spermatophylax mass did not correlate with residual sperm number across Poecilimon species (y = 0.38×, F_1,21 = 1.76, p = 0.2, r² = 0.08; Fig. 6), and was largely unaffected by the inclusion of P. mytilenensis, y = 0.39×, F_1,22 = 1.48, p = 0.2, r² = 0.06; Table 2).

Fig. 5.

Positive relationship between residual ampulla mass and residual spermatophylax mass across 31 Poecilimon species.

Fig. 6.

No relationship exists between residual sperm number and residual spermatophylax mass across 22 species of Poecilimon.

A significant correlation was found between the residuals of ampulla mass and sperm number (y = 0.4817× − 0.0032, F_1,21 = 9.426, p = 0.006, r² = 0.32; Fig. 7, Table 2), although a substantial portion of the variance in residuals, about 68%, could not be explained by the model. Including P. mytilenensis in this model strengthened this association so that 57% of the variation could not be accounted for (y = 0.54×, F_1,22 = 15.71, p<0.000, r² = 0.43).

Fig. 7.

The relationship between residual sperm number and residual ampulla size across 22 Poecilimon species.

Discussion

Spermatophore variation, ejaculate protection and paternal investment.—The positive correlation we found between residual spermatophylax mass and residual ampulla mass is consistent with other research supporting the ejaculate protection hypothesis (Reinhold & Heller 1993; Wedell 1993a, 1994b; Heller & Reinhold 1994; Vahed & Gilbert 1996). Vahed & Gilbert (1996) also found a strong correlation between residual spermatophylax mass and residual ampulla mass within 43 species from nine subfamilies of mostly European bush-crickets. Similarly, Wedell (1993a, 1994b) found a positive correlation between spermatophylax mass and ampulla mass in 19 genera of mostly Australian bush-crickets. While the correlation found between these components within Poecilimon was moderate, the relationship was strengthened by the removal of P. mytilenensis — a species known to have an inordinately large ampulla, but a modestly sized spermatophylax (Heller et al. 2004).

Fig. 8.

Poecilimon veluchianus minor with attached spermatophore. From a population at Makrakomi, mainland Greece, near the village of Tsouka, 1998. Photo by J. McCartney. See Plate III.

While our findings are consistent with the ejaculate protection hypothesis, they are not inconsistent with the paternal investment hypothesis.

The spermatophylax of P. veluchianus, for example, is approximately the size required to allow for an optimum amount of sperm to enter into the female (Reinhold & Heller 1993, Heller & Reinhold 1994), although the spermatophore of the last male to mate will have a positive effect on the dry weight of his own offspring (Reinhold 1999). The paternal investment hypothesis assumes selection acts on the spermatophylax through a direct nutritional benefit to the offspring (Trivers 1972, Gwynne et al. 1984). Yet compared to the spermatophylax, the ejaculate may be produced relatively inexpensively (e.g., Bateman 1948, Trivers 1972, but see Dewsbury 1982, Reinhold & Helversen 1997; Wedell et al. 2002 provide a review) and is critical to male fertilization success. Ampullae size (ejaculate volume) may still modulate spermatophylax size through influences of ejaculate protection, while the primary factors influencing spermatophylax size itself are paternal investment. Males that primarily invest heavily in spermatophylaces and as a result, provide a significant nutrient investment to their offspring, may also produce greater than normal quantities of sperm in order to ‘hedge their bets’ and maintain paternity shares in the face of sperm competition (Reinhold & von Helversen 1997, Lehmann & Lehmann 2000b). The ejaculate and/or spermatophylax mass may also have flow-on effects in females by influencing female intermating refractory period (Heller & Helversen 1991, Heller & Reinhold 1994, Lehmann & Lehmann 2000b, Vahed 2007), female lifespan (Brown 1997), the timing of oviposition (Wedell & Arak 1989), and the share of eggs that are laid with the donating males' nutritional investment (Simmons 1990, Vahed 2003).

Under the ejaculate-protection hypothesis, the cost of extra sperm or ejaculate fluid is assumed to be negligible in comparison to the gain in paternity (Simmons 1995b). Evidence showing sperm to be less costly than the production of the spermatophylax has been observed in P. mariannae: parasitized males lose their ability to replenish their spermatophylax, but not their sperm (Lehmann & Lehmann 2000b). Similarly, Reinhold & von Helversen (1997) found that spermatophore replenishment rather than sperm number limits intermating interval in male P. veluchianus.

In contrast to predictions of the ejaculate-protection hypothesis, we did not observe a relationship between sperm number and spermatophylax size in Poecilimon. This runs counter to findings from other studies where a positive relationship existed across taxa (e.g., Wedell 1994b, Vahed & Gilbert 1996). Sperm number has also been found to be independent of spermatophylax mass in an Australian bush-cricket, R. verticalis (Simmons et al. 1993), and in P. veluchianus, (Reinhold & von Helversen 1997). Reinhold & von Helversen (1997) further predicted that this lack of relationship may represent a general trend in bush-crickets. However, sperm number and spermatophylax mass are adjusted in concert in parasitized P. mariannae (Lehmann & Lehmann 2000b), so the situation appears to be more complicated in Poecilimon.

While our results confirm the prediction of Reinhold & von Helversen (1997), the validity of the ejaculate-protection hypothesis relies more specifically on the relationship between spermatophore consumption time and sperm discharge time, rather than covariance of spermatophylax mass and sperm number (see for example Reinhold & Heller 1993, McCartney & Heller submitted ms.). An association between spermatophylax consumption time and sperm drainage has been observed in all bush-cricket studies thus far: R. verticalis (Gwynne 1984a, 1986, 1997, but see Simmons 1995a, Vahed 1998 for different interpretations), Decticus verrucivorus (Wedell & Arak, 1989), Kawanaphila nartee (Simmons & Gwynne, 1991), and Leptophyes laticauda (Vahed 1994), as well as Poecilimon hoelzeli (Achmann 1996), and two subspecies of Poecilimon veluchianus (Reinhold & Heller 1993, Heller & Reinhold 1994). However, the spermatophore consumption time and sperm discharge do not correspond in two further Poecilimon species (P. laevissimus and P. thessalicus, McCartney & Heller submitted ms.). This, combined with our detection of a large intraspecific variation in spermatophylax mass and sperm numbers between individuals, populations and years (e.g., P. thessalicus and P. veluchianus, Table 1) is likely to explain the lack of association we found within the genus.

Under the ejaculate-protection hypothesis, the spermatophylax may be viewed as a sperm-protection device, allowing the transfer of a maximum number of sperm, and being primarily influenced by sperm competition. However, chemicals in the ejaculate itself can increase male fitness by functioning to increase onset of egg-laying, increase total number of eggs laid and to prolong the female intermating period (e.g., Reinhold & von Helversen 1997; Vahed 1998, ; 20032006, 2007; Arnqvist & Rowe 2005). Our study indicates that discharge of the ejaculate may be more important in terms of spermatophylax function than the discharge of sperm per se. While we found a significant association, ampulla mass only explained a small amount of variation in sperm number within Poecilimon.

Only one other comparative study seems to have measured the association between ampulla mass and sperm number among bush-cricket species and no relationship was found (Vahed 2006). This, in combination with our finding that ampulla mass, but not sperm number, correlates with spermatophylax mass, indicates that the spermatophylax, in terms of mating effort, has an ejaculate-protection function, but not a primary sperm-drainage function in Poecilimon. Our results lead us to believe that sperm number itself should not be used as an assessment of the ejaculate protection function, nor should ejaculate volume (ampulla size) be used to assess sperm protection or competition (e.g., Wedell 1993a, Wedell 1997) when making interspecific comparisons.

Spermatophore size variation within Poecilimon.—Spermatophore size within the genus Poecilimon approximates that found within the entire family Tettigoniidae (c.f. Wedell 1993a, Vahed & Gilbert 1996, Wedell 1997, Vahed 2007), indicating that variation in spermatophore size is unlikely due to relatedness or diet alone. This large variation between species is likely to reflect within-species adjustments that male bush-crickets make to specific spermatophore components as a conditional strategy — apparently in order to maximise reproductive output (e.g., P. affinis, P. erimanthos, P. hamatus, P. jonicus, P. laevissimus, P. thessalicus, P. veluchianus). We found that all spermatophore components in Poecilimon scale approximately with male body mass, but large variations are apparent in relative investment when body mass is taken into consideration.

Preferential investment in spermatophore components suggests that variations in environment and available energy or nutrients are directed to whichever spermatophore component is more effective at increasing reproductive fitness (see for example Voigt et al. 2005 and references cited therein). Examples of this have been found in a variety of bush-crickets. Male Requena verticalis, for example, increase the number of sperm when mating with older females, or when exposed to a high female sex ratio, effectively increasing their chances of paternity, given the likely increase in sperm competition (Simmons et al. 1993, Simmons 1995a). Similarly, R. verticalis males disproportionately adjust the ampulla mass over the spermatophylax mass in relation to their remating frequency (Simmons 1995b) or mating potential (Simmons 1995c). Males of another species, Decticus verrucivorus, adjust the size of the offered spermatophore depending on whether or not a mate is virgin (Wedell 1992).

Considerable variation in the size of Poecilimon spermatophore components was found between and within populations (e.g., P. erimanthos, P. hamatus, P. jonicus, P. laevissimus, P. thessalicus, P. veluchianus). The foundation for this variation is likely the availability of environmental resources (e.g., Hubbell & Johnson 1987, Gwynne & Simmons 1990, Adamo & Hoy 1994) yet, while related, more proximal causes associated with life histories and mating behavior, including population density, operational sex ratio, and sexual size dimorphism, influence the relative pay-offs in spermatophore production (e.g., Gwynne 1981, 1984a, b; Gwynne & Simmons 1990; Heller & von Helversen 1991; Allen 1995; Bateman 1997). There is little published information on intraspecific variation in spermatophore component size among bush-cricket populations, and evidence presented here suggests that further research on Poecilimon is needed to help clarify how environmental factors affect male investment in spermatophore components.

Spermatophore differences between field and laboratory-raised individuals.—Importantly, we found large differences between laboratory-reared individuals and those from the field. For example, P. v. minor males reared in the laboratory had a larger body mass and over twice as many sperm per spermatophore, compared to those in the field. The converse was true for P. v. veluchianus, which had a larger number of sperm in individuals collected in the field. A large range in ampulla mass was also seen in this subspecies (5.3 to 25.4 mg) and previous studies show that spermatophore consumption time also varies greatly between conditions (Reinhold & Heller 1993). Similarly, P. affinis differs considerably in sperm number in laboratory and field observations, with nearly five times more sperm in laboratory-reared individuals; however it is difficult to assess whether this reflects environment differences or bias due to small sample size. Laboratory-reared animals it seems, often show extreme variations in spermatophore component size. This may provide important information in some circumstances; however, given the highly variable nature of spermatophore production, we recommend caution when interpreting spermatophore function using laboratory-reared animals, small sample sizes, or means from short-term observations.

Conclusions

Detailed analyses of spermatophore size with respect to phylogeny and diet will be important to developing a more complete understanding of the evolutionary significance of variation in spermatophore size. Spermatophore component size in Poecilimon appears to be evolutionarily labile and a general lack of association within Poecilimon between relative spermatophore-component size and male body mass, reflects differences related to mating strategy. This, combined with a lack of association between spermatophore component size, indicates that effective ejaculate transfer, not sperm drainage per se, is a significant influence in the evolution of spermatophore size. Mating effort and paternal investment are not mutually exclusive and further analysis within Poecilimon on the direct association between the amount of sperm that drains into the female and its relationship to spermatophore-consumption time is needed for a full understanding of the relative influences of ejaculate protection and paternal investment on spermatophore size. Given the significance of sperm competition in evolutionary biology, studies within and between closely related species in natural populations are necessary to improve knowledge of the processes that influence the evolution of nuptial feeding in insects.