Introduction

Spermatozoa vary considerably in size and shape between species (Franzén 1970; Cohen 1977; Pitnick et al. 2009) and also, albeit at generally lower levels, within species or between ejaculates of individual males (Ward 1998; Morrow and Gage 2001b; Birkhead et al. 2005; Calhim et al. 2007; Harris et al. 2007). However, while the structure of spermatozoa has been well studied, the causes and consequences of the variation in sperm design and the selective pressures driving its evolutionary divergence are poorly understood.

Sperm design is thought to be shaped by the mode of fertilisation and phylogeny (Franzén 1970; Jamieson 1987) and also by post-copulatory sexual selection (PCSS; Sivinski 1980). PCSS consists of sperm competition (Parker 1970) and cryptic female choice (e.g. Thornhill 1983; Eberhard 1991, 1996) and favours sperm characteristics that enhance a male's fertilisation success.

Although theoretical models predict that in internal fertilisers sperm size should remain small regardless of the level of sperm competition (Parker 1993; Parker and Begon 1993), empirical studies have often documented considerable variation in sperm size associated with the intensity of PCSS across species. Most of these studies reported positive co-variation between sperm size and PCSS (e.g. Briskie and Montgomerie 1992; Gage 1994; LaMunyon and Ward 1999; Byrne et al. 2003; but see Gage and Freckleton 2003). In passerine birds, however, this co-variation is positive in some families but negative in others (Immler and Birkhead 2007), and further studies in other taxa are needed for a better understanding of the link between sperm design and PCSS.

In contrast to evolutionary trends across species, we know relatively little about the variation in sperm design within species. Theory predicts that sperm are selected towards specific optima at given levels of sperm competition (e.g. Parker 1993; Parker and Begon 1993). Consequently, the variation of sperm around the optimal design is expected to decline within increasing selection intensity, a phenomenon that has indeed been shown for both intra-specific (Calhim et al. 2007; Kleven et al. 2008) and intra-male variation (Immler et al. 2008; but see Kleven et al. 2008). These three studies on sperm variation were conducted in passerine birds, using species from a large number of different families and with marked variation in sperm competition as reflected by the range of extra-pair paternity levels. One potential reason why Immler et al. (2008) found a negative relationship between intra-male sperm variation and sperm competition but Kleven et al. (2008) did not is that these two studies used different suites of species. Both studies applied rigorous control for phylogenetic effects using the same powerful statistical tools (Freckleton et al. 2002), but most techniques for phylogenetic control currently available rely on the underlying assumption that the traits under examination are exposed to identical modes of evolution across all species in the analysis (Pagel and Harvey 1989; Harvey and Purvis 1991). This assumption may not hold for a broad comparative study on sperm of passerine birds as there is clear evidence that associations between sperm design and sperm competition differ among families (e.g. Immler and Birkhead 2007). Thus, a study within a single family may be less vulnerable to different selection patterns that exist between families.

If the levels of variation in sperm design differ between species in relation to sperm competition, it remains unclear how this affects sperm function. Sperm velocity is an important determinant of male fertilising success (e.g. Holt et al. 1989; Birkhead et al. 1999; Gage et al. 2004; Malo et al. 2005a), and, considering the diversity in sperm size and shape, it seems likely that the design of sperm affects their swimming performance. In fact, theoretical models predict at least three ways through which sperm design may increase sperm velocity: (1) enlarged midpiece size providing elevated energy supply for powering the flagellum (e.g. Cardullo and Baltz 1991; Froman and Feltmann 1998); (2) a longer flagellum increasing propulsive forces (Katz et al. 1989), which have also been assumed to result in greater velocity (e.g. Gomendio and Roldan 1991; Briskie and Montgomerie 1992; Gage 1994; Byrne et al. 2003); and (3) longer flagellum length relative to head size that can better overcome the drag forces acting on the head (e.g. Higdon 1979; Humphries et al. 2008). A recent study across a range of passerine species demonstrated that sperm velocity increases with both absolute and relative dimensions of the sperm midpiece and flagellum (Lüpold et al. 2009a). However, while comparative studies are useful for establishing general evolutionary trends, understanding the basis of these trends requires detailed intra-specific analyses (e.g. Garland and Carter 1994; Arnqvist 1997). Consequently, it remains to be established whether the association between sperm design and sperm velocity reported across passerines also exists within these species.

We used the New World blackbirds (Icteridae) as a study taxon to test the hypotheses that (1) sperm length and the size of individual sperm components are influenced by the level of sperm competition; (2) the intra-specific variation in sperm design decreases with increasing sperm competition; and (3) sperm design affects sperm velocity within species.

Material and methods

Sperm collection and analyses

In collaboration with numerous ornithologists, museum collectors and bird banders in North America and South America, we collected sperm samples from wild populations of 38 different species of Icteridae, using any of the four different methods: (1) cloacal massage (e.g. Burrows and Quinn 1937; Samour et al. 1986), (2) natural ejaculations into the false cloaca of model females (Pellatt and Birkhead 1994), (3) from faecal samples (Immler and Birkhead 2005) or (4) by dissection of the distal end of the seminal glomera (i.e. sperm-storage organ at the end of the deferent duct) in species where specimens were collected for other research projects, museums or management programmes. Sperm collected through these different techniques do not differ in their morphological measurements (Immler and Birkhead 2005; Immler et al. 2008). All samples were fixed in 5–10% formalin. A full list of species with their phylogeny and sample sizes for the different analyses is given in the online Supplementary Table S1.

Sperm morphometric analyses

For each sperm sample, we analysed five to ten morphologically normal and undamaged sperms, which is a sufficient sample size to capture the within-male variation for comparative studies in passerine birds (Calhim et al. 2007; Immler and Birkhead 2007). We measured samples from digital images taken at magnifications of ×250 or ×400 (depending upon sperm size). Using computer-assisted image analysis, we measured the following traits of each sperm cell to the nearest 0.1 µm: (1) head length; (2) helical midpiece length, which we used to calculate straight midpiece length using the formula in Birkhead et al. (2005); (3) flagellum length; and (4) total sperm length. We used the within-sample mean of each sperm trait for all further analyses.

Sperm velocity analyses

We measured sperm velocity in sperm samples that we collected using model females, cloacal massage or immediate dissection of collected birds. After diluting a freshly collected sperm sample in Dulbecco's modified Eagle medium (DMEM, Invitrogen Ltd.) to a final concentration of approximately 20 × 106 sperm per millilitre, we immediately placed 15 µl under a phase-contrast microscope at 35 ± 0.1°C and videotaped at a magnification of ×200 . We then analysed all video recordings using computer-aided sperm analysis (Hobson Tracking Systems Ltd., U.K.) and removed non-typical trajectories following the principles of Mossman (2008).

In the intra-specific comparisons of sperm morphology and velocity, we focused on the four species for which we had full datasets for ≥20 males (range 20–120 males; see online Supplementary Table S1): Agelaius phoeniceus, Xanthocephalus xanthocephalus, Quiscalus quiscula and Molothrus ater. Within the Icteridae studied so far, these four species span the entire range of sperm lengths (61–145 µm) and also most of the range of relative testes size (see Figs. 1 and 2).

Fig. 1
figure 1

Relationships of sperm head (grey circles), midpiece (solid black circles) and flagellum length (open circles) with total sperm length. The arrows at the bottom of the graph refer to the four species used for intra-specific analyses of sperm velocity (from left to right: M. ater, sperm length = 61 µm; Q. quiscula, 98 µm; X. xanthocephalus, 114 µm; A. phoeniceus, 145 µm). All slopes deviate significantly from zero and from one another (for statistics, see text)

Full size image
Fig. 2
figure 2

Partial residual plot showing the inter-specific association of total sperm length with combined testes mass (both corrected for body mass) across 38 species of Icteridae (r = 0.42, P = 0.01). Each point indicates a species, and the four open circles depict the species used for intra-specific analyses of sperm velocity (from top to bottom: A. phoeniceus, X. xanthocephalus, Q. quiscula, M. ater)

Full size image

To calculate a sperm velocity index, we performed a principal components analysis (PCA) on five motility parameters: (1) curvilinear velocity (total sperm trajectory, VCL), (2) average path velocity (smoothed and averaged trajectory, VAP), (3) straight-line velocity (shortest distance between the start and end point of the trajectory, VSL), (4) path linearity (departure of actual sperm track from straight line, VSL/VCL) and (5) path straightness (departure of average sperm path from straight line, VSL/VAP). Following a two-factor ‘varimax’ rotation in the PCA, the speed parameters (1–3) were maximally loaded on the first principal component (PC1) and the path-shape parameters (4–5) on the second (PC2; Table 1). We used PC1 for our analyses.

Table 1 Results of the principal components analysis on the sperm motility measures for A. phoeniceus
Full size table

Statistical analyses

We conducted all statistical analyses using the statistical package R v.2.7.1 and normalised all non-normal data distributions by log or square-root transformations to meet the parametric requirements of the statistical models.

To control for phylogenetic effects in all comparative analyses, we constructed a phylogenetic tree (see online Supplementary Table S1) and accounted for statistical non-independence of data points by shared ancestry of species (Felsenstein 1985; Harvey and Pagel 1991), using a generalised least-squares (GLS) approach in a phylogenetic framework (Pagel 1999; Freckleton et al. 2002). This approach allows the estimation of the phylogenetic scaling parameter λ, with values of λ close to 0 indicating phylogenetic independence and values of λ close to 1 indicating a complete phylogenetic association of the traits. We used likelihood ratio tests to establish whether the model with the maximum-likelihood value of λ differed from models with values of λ = 1 or 0, respectively. We indicate the significance levels of these likelihood ratio tests by superscripts following the λ estimates (e.g. λ 0.1;1.0; first superscript: against λ = 1; second superscript: against λ = 0).

We used the intra-specific coefficients of variation (CV), all log-transformed, as indices of intra-specific variation in sperm dimensions or velocity. For intra-specific CV estimates, Calhim et al. (2007) estimated the minimum adequate sample size to be around ten males. Although we attempted to collect samples from at least ten males for each species, this was not possible for most species due to limited access to reproductive males or because, for some species, many faecal samples contained no sperm. However, the simulations of Calhim et al. (2007) also showed that even five males yield intra-specific CV estimates within about 5% of that obtained from 50 males. Including species with ≥6 males increased our sample size from nine to 15 different species and formed a wider basis for our comparative analyses. We thus performed our analyses of intra-specific variation on all 15 species and confirmed the results using the more conservative, albeit considerably smaller, dataset of nine species (for species and sample sizes, see online Supplementary Table S1). For intra-specific variation in sperm velocity, we had data from six species with more than five males (one of these species with less than ten males).

To account for the variation in sample sizes across species (six to 351 males per species; online Supplementary Table S1), we maximised the efficiency of parameter estimation by weighting our analyses by the corresponding sample sizes (N). In brief, similar to the phylogeny matrix (V) used in the above GLS approach, we constructed a matrix (W) with 1/log(N) in the diagonal and 0 in all other positions and combined the two matrices with Q = V + cW, where c is a constant (Martins and Hansen 1997; Garamszegi and Møller 2007). By varying c, we determined the model with the highest maximum-likelihood score. We will report the weighted and unweighted results of our final models.

As an index of sperm competition, we used relative testes size by including both log-transformed combined testes mass and body mass as independent variables in our analyses, which is preferable to the use of residuals from a regression between the two variables (e.g. García-Berthou 2001). We obtained testes and body mass data from the published literature, our own measurements and from museum databases (i.e. Field Museum of Natural History Chicago and Smithsonian National Museum of Natural History), using data only from birds that were likely to be in breeding condition and to have fully developed testes given their geographic location, date of collection and the range of testes sizes in the databases (see Calhim and Birkhead 2007). Although relative testes size as a measure of sperm competition tends to slightly underestimate the sperm production rate and thus potentially the level of sperm competition, for species under intense sperm competition (Lüpold et al. 2009b), it was the best index available for the entire collection of species in our study.

In the intra-specific comparisons of sperm morphology and velocity, we used four morphological traits: (1) midpiece and (2) flagellum length, (3) flagellum to head (F:H) ratio and (4) midpiece to flagellum (M:F) ratio to account for both absolute and relative sizes of sperm components. For A. phoeniceus, we obtained 148 natural ejaculates from 120 different males, using model females. For the 16 males with ≥2 samples, we used intra-male means to avoid pseudoreplication. In the other three species, we obtained a single sample from each male after dissection. The processing time of samples (from collection in the marsh to start of video recording for motility analysis) varied between males due to field conditions, but it had no significant effect on any of our analyses (all P > 0.21). We thus omitted control for the elapsed time and performed simple regressions in all four species.

Results

We found considerable variation in the size of sperm dimensions across the 38 species of Icteridae, with mean sperm length ranging between 61 and 145 µm. In nested analyses of variance (ANOVAs), about 98% of the total variation in midpiece, flagellum or total sperm length was explained by the inter-specific variation, whereas the intra-specific and intra-male variations contributed only around 1.8% and 0.2%, respectively. In contrast, head length varied relatively little amongst species (range 12.3–15.0 µm), such that the inter- and intra-specific variance components shared similar percentages of the total variation (46.0% and 44.2%, respectively), with the remaining 9.8% being explained by the variation within males.

Relationships between sperm traits

Using the means for each of the 38 species, midpiece length was positively correlated with flagellum length (r = 0.97, P < 0.0001, λ < 0.0001<0.001;1.0) and with total sperm length (r = 0.96, P < 0.0001, λ < 0.0001<0.001;1.0). Head length was also positively related to all other sperm measures (midpiece: r = 0.46, P = 0.007, λ < 0.00010.15;1.0; flagellum: r = 0.50, P = 0.003, λ < 0.00010.04;1.0; total length: r = 0.54, P = 0.002, λ < 0.00010.03;1.0).

The relative increase of the three sperm components (head, midpiece and flagellum length) with total sperm length may indicate metabolic and/or functional regulation of the sperm (e.g. Cardullo and Baltz 1991; Humphries et al. 2008). We thus compared the slopes between the three sperm traits relative to total sperm length in an analysis of co-variance (ANCOVA) with trait size as the response variable and trait and sperm length as the explanatory variables. All interactions between slopes were statistically significant: flagellum length (slope = 1.14) increased significantly more rapidly with total sperm length than did head length (slope = 0.15; partial r = −0.89, P < 0.0001), and the slope for midpiece length (slope = 1.48) was significantly steeper than that of flagellum length (partial r = −0.44, P < 0.0001; see Fig. 1). Consequently, any elongation of the sperm head would result in a disproportionate increase in flagellum length, and this in turn would entail an even greater increase in midpiece length.

Associations between sperm morphology, variation and sperm competition

Across the 38 species, all morphometric sperm traits except head length increased significantly with relative testes size (Table 2; Fig. 2). The λ values were all very small and not significantly different from 0, indicating that the relationships between these traits are independent of the underlying phylogeny (Freckleton et al. 2002).

Table 2 Statistics of the inter-specific associations between morphological sperm traits and relative testes size (i.e. combined testes mass [CTM] corrected for body mass); all variables are log-transformed; N = 38 species
Full size table

The intra-specific CV estimates of sperm components varied sixfold to 13-fold across the 15 species (e.g. CV(flagellum) = 1.1–6.7%, CV(midpiece) = 1.0–13.2%) and were negatively correlated with relative testes size, particularly after weighting for sample sizes within species (Table 3; Fig. 3). Here, the λ value for the variation in head length was much higher than in any of the other analyses but not significantly different from 0, suggesting again independence of the phylogeny. The same relationships were also significantly negative using only the nine species with ≥10 males and thus more than the minimal adequate sample sizes within species (all P ≤ 0.04; Fig. 3).

Table 3 Statistics of the unweighted and weighted inter-specific associations between the variation (expressed by intra-specific coefficients of variation, CV) in morphological sperm traits and relative testes size (combined testes mass [CTM] corrected for body mass); N = 15 species
Full size table
Fig. 3
figure 3

Partial residual plot showing the negative association between the degree of intra-specific variation (CV) in total sperm length and combined testes mass, both corrected for body mass. Each point represents a species (N = 15; black points: species with ≥ 10 males; grey points: species with six to nine males). The thick lines are the regression lines across all 15 species (solid: weighted for sample size, r = −0.59, P = 0.032; dashed: unweighted, r = −0.55, P = 0.051), and the thin line depicts the (weighted) regression line of the analysis restricted to the minimum adequate sample sizes within species (i.e. black points only: N = 9, r = −0.88, P = 0.007). All statistics are controlled for phylogeny, and the numbers refer to the species in the online Supplementary Table S1

Full size image

Correlations between sperm morphology and velocity

Within the four species with samples from ≥20 males, sperm velocity (PC1) was not significantly correlated with any of the four morphological sperm measures that are predicted to affect sperm velocity.

In A. phoeniceus, the sperm velocity index (PC1) was not significantly correlated with any morphological sperm trait (Fig. 4; N = 120 males; midpiece: r = 0.17, P = 0.06; flagellum: r = 0.08, P = 0.39; F:H ratio: r = −0.01, P = 0.96; M:F ratio: r = 0.14, P = 0.12). However, due to the result with midpiece length, we also tested VAP and VSL separately, which are both frequently used as univariate measures of sperm velocity (e.g. Moore and Akhondi 1996; Burness et al. 2004; Birkhead et al. 2005; Malo et al. 2005a; Cornwallis and Birkhead 2007). VAP increased significantly with midpiece (r = 0.18, P = 0.05) and flagellum length (r = 0.21, P = 0.03), but VSL did not (midpiece: r = 0.14, P = 0.12; flagellum: r = 0.11, P = 0.72). Both velocity measures were not significantly correlated with the F:H or M:F ratios (all r < 0.14, all P > 0.11). Overall, we found no consistent pattern in A. phoeniceus but some indication for a potential link between sperm morphology and velocity, although any effects were at best weak and not significant after accounting for multiple testing.

Fig. 4
figure 4

Two examples showing no significant relationship between the sperm velocity (PC1 of sperm velocity) and (a) midpiece length (r = 0.17, P = 0.06) and (b) flagellum length (r = 0.08, P = 0.39) in 120 different male red-winged blackbirds (A. phoeniceus)

Full size image

In X. xanthocephalus, we also found no association between the size of sperm components and sperm velocity (N = 24 males; midpiece: r = −0.05, P = 0.82; flagellum: r = −0.09, P = 0.68; F:H ratio: r = 0.18, P = 0.40; M:F ratio: r = 0.04, P = 0.84). Likewise, sperm velocity did not co-vary with sperm dimensions in Q. quiscula (N = 20 males; midpiece: r = 0.05, P = 0.84; flagellum: r = −0.29, P = 0.23; F:H ratio: r = −0.27, P = 0.26; M:F ratio: r = 0.24, P = 0.32) and in M. ater (N = 21 males; midpiece: r = 0.11, P = 0.64; flagellum: r = 0.25, P = 0.30; F:H ratio: r = −0.12, P = 0.61; M:F ratio: r = 0.03, P = 0.91). Using individual parameters of sperm velocity did not change these results qualitatively.

Having found no significant relationships between sperm morphology and velocity within species, we combined the datasets from all four species and performed an ANCOVA for each sperm trait to establish whether a general pattern existed across species. For each morphometric trait, the interactions between sperm trait and species were not statistically significant (all P > 0.36) and thus omitted from the analyses. Furthermore, in the analyses with midpiece and flagellum length, respectively, as explanatory variables, the coefficients did not differ significantly between A. phoeniceus, X. xanthocephalus and Q. quiscula. Collapsing these three factor levels did not change the explanatory power of the models (ANOVAs between models: P > 0.53) but lowered Aikake's Information Criterion (AIC) by more than three units, thus indicating a better fit of the simplified models. Using the same criteria, X. xanthocephalus and Q. quiscula could be combined in the analyses with the ratios between sperm components. All minimal adequate models showed a positive effect of trait size (all partial r > 0.37, all P < 0.001) on sperm velocity and significantly different intercepts between species groups (all partial r > 0.17, all P < 0.03), but the explanatory power of all models was relatively low (all multiple r 2 = 0.19 to 0.21). In other words, after combining the data amongst all four species, there is some indication that sperm morphology may influence sperm velocity as predicted by theory.

Finally, we compared the intra-specific variance of sperm velocity with that of sperm morphology and with relative testes size. Across the six species with a full dataset for more than five males, we combined the log-transformed intra-specific CVs of the five motility parameters in a PCA with a ‘varimax’ rotation and used PC1 as an index of intra-specific variation in sperm velocity. This index was inversely correlated with the CVs of sperm components (Fig. 5; midpiece: r = −0.93, P = 0.008; flagellum: r = −0.94, P = 0.009; total sperm length: r = −0.96, P = 0.005; all λ < 0.0001<0.03;1.0), and it increased with relative testes size (testes: r = 0.96, P = 0.02; body: r = −0.89, P = 0.07; λ < 0.00010.03;1.0). However, using the intra-specific variation in VSL or VAP (measured as CV's) instead of PC1 (above), the association between the variation in sperm velocity and relative testes size was no longer significant (all r < 0.75, all P > 0.11). In all analyses, the unweighted analyses yielded qualitatively the same results as the weighted ones.

Fig. 5
figure 5

Relationship between the index of overall variation in sperm velocity (intra-specific CVs of the five motility parameters combined in PCA) and the intra-specific coefficient of variation of total sperm length across six species of Icteridae with data from over five different males. The solid line is weighted for sample size (r = −0.96, P = 0.004); the dashed line is unweighted (r = −0.97, P = 0.003), and both statistics are controlled for phylogeny. The numbers refer to the species in the online Supplementary Table S1

Full size image

Discussion

Our main results were that across species, (1) all major sperm components (except head length) increased in size with the level of PCSS, indicating directional selection for longer sperm; (2) the intra-specific variation in all sperm components decreased with increasing sperm competition, suggesting stabilising selection around an optimal sperm design within species; (3) within four species varying in their level of sperm competition, sperm velocity was not significantly correlated with sperm design, but in A. phoeniceus we found interesting patterns with midpiece and flagellum length that deserve further investigation; and (4) the intra-specific variation in sperm velocity was inversely related to the variation in sperm design, while there was no clear pattern for its association with sperm competition.

Sperm design and sperm competition

We found positive relationships between sperm competition and all sperm components (except head size), consistent with previous studies on other taxa (e.g. Gage 1994; Balshine et al. 2001; Anderson and Dixson 2002; Byrne et al. 2003). However, while longer sperm may be advantageous and selected for by PCSS in these taxa, in other taxa, shorter sperm appear to be favoured (Immler and Birkhead 2007). The reasons for these inverse evolutionary trends are currently poorly understood, and further insight into the links between sperm design, sperm function and male and female reproductive organs may help us understand the selective mechanisms and processes.

It also remains to be established why PCSS may favour longer sperm, but at least three potential mechanisms have been proposed: first, sperm size can be selected for by co-evolution with sperm-storage structures within the female reproductive tract, as evinced in various taxa (e.g. Dybas and Dybas 1981; Briskie and Montgomerie 1992; Miller and Pitnick 2002). Second, if a larger midpiece can produce more ATP, longer sperm (with a longer midpiece) may also be more competitive because they are longer lived (Parker 1993, 1998, but see Levitan 2000; Gage et al. 2002; Immler et al. 2007) or swim faster (discussed below, “Sperm design and velocity”). Finally, larger sperm may also displace smaller rival sperm from the female reproductive tract (LaMunyon and Ward 1998). Which of these mechanisms apply to the Icteridae has not been established, but a combination of different factors seems likely given the positive association of sperm design with sperm velocity across the Icteridae (Lüpold et al. 2009a) or with the size of female sperm-storage tubules across passerine birds in general (Briskie and Montgomerie 1992; Briskie et al. 1997).

Variation in sperm design and sperm competition

In addition to selection for longer sperm, PCSS also appears to reduce the intra-specific variation in sperm design in the Icteridae, probably by stabilising selection around a species-specific optimum (Morrow and Gage 2001b; Calhim et al. 2007). That we found the same pattern across closely related species of a single family as in the sample of species used in the studies of Calhim et al. (2007) or Kleven et al. (2008), where the range of PCSS levels is much greater, suggests that stabilising sexual selection may be particularly strong within icterid species. However, the depletion of sperm variation between males in species under intense PCSS raises the question of how much sperm design ultimately contributes to the variation in reproductive success. It would also be interesting to know more about the heritability and selection potential of sperm design in species under intense selection. To date, such information in passerine birds is, to our knowledge, available only for the zebra finch (Taeniopygia guttata), a species under low sperm competition, in which sperm design is highly heritable and varies considerably between males (Birkhead et al. 2005).

Sperm design and velocity

Within each of four icterid species, we found no significant association between the size of sperm components and the sperm velocity index, and only very weak positive relationships for individual velocity parameters. We thus found at best little support of theoretical models that have predicted different ways through which sperm design may influence sperm velocity (Higdon 1979; Katz et al. 1989; Cardullo and Baltz 1991; Humphries et al. 2008).

Our results contrast with a recent empirical study across 40 passerine species, including 13 species of Icteridae, where strong positive associations have been observed between sperm velocity and absolute or relative sperm dimensions (Lüpold et al. 2009a), or with Mossman's (2008) study in the zebra finch, where sperm velocity is positively related to and genetically linked with sperm design. Overall, however, the majority of intra-specific studies have found no significant links between sperm design and velocity, in either internal or external fertilisers (e.g. Gage et al. 2002; Burness et al. 2004; Minoretti and Baur 2006; Stoltz and Neff 2006; Fitzpatrick et al. 2007; Locatello et al. 2007; Pitcher et al. 2007), including another passerine species, the red-billed quelea (Quelea quelea; S. Calhim and T. R. Birkhead, unpublished data). Finally, as in the Icteridae, sperm velocity in cichlid fish is also independent of sperm length in most species but strongly associated across species (Fitzpatrick et al. 2009).

The reasons for the difference between the intra- and inter-specific studies in passerine birds remain unclear, but the discrepancy may at least in part be attributed to the variation in both sperm design and velocity. For example, both A. phoeniceus and Q. quelea are under intense sperm competition and thus show very low variation in sperm design, whereas zebra finches exhibit much greater variation (Birkhead et al. 2005; Calhim et al. 2007). Therefore, the low variation of the first two species may compromise the detectability of a potential relationship between sperm design and velocity despite large sample sizes (120 and 113 males, respectively), compared to the zebra finch or the inter-specific comparison. Due to the inverse relationship between the intra-specific variances in sperm design and velocity, the other species of this study may in turn have too little variation in sperm velocity for enough statistical power, particularly given their relatively low sample sizes (N = 20–24). In fact, these three analyses exhibited low statistical power, and indeed the majority of intra-specific studies that found no link between sperm phenotype and swimming performance are based on fewer than 30 males (e.g. Burness et al. 2004; Minoretti and Baur 2006; Stoltz and Neff 2006; Locatello et al. 2007). The only study that found significant effects with such a small sample size is Malo et al. (2006) with 36 males, but their association may be driven mainly by three extreme values. Overall, it appears that even if sperm design affects sperm velocity, such a pattern may be detected only through a large sample size; when sample sizes are small, any pattern may be obscured too much by confounding factors.

Particularly for internal fertilisers, the environment in which sperm operate is difficult to simulate. Sperm performance is typically measured under standardised conditions (e.g. temperature and medium, such as DMEM), but within the female reproductive tract, sperm locomotion may be affected by wall effects and vaginal fluid or by components of the seminal fluid. Standardising the conditions for sperm analysis can control for some confounding effects, particularly for the examination of biomechanical links between form and function, but we do not know whether ejaculates from different male birds perform equally well in artificial media compared to their own seminal fluid. To test this possibility, we would need to separate seminal fluid from sperm and use this to re-suspend a subpopulation of the sperm contained in the same ejaculate. In passerine birds, this approach would be technically difficult because ejaculates are highly viscous and contain very little fluid, and they have to be greatly diluted to obtain a measurable concentration (personal observation).

Recent studies in the fowl (Gallus gallus) also show that sperm velocity can decrease over successive copulations (Cornwallis and Birkhead 2007) and that sperm performance may depend on social status (Pizzari et al. 2007). Sperm velocity has also been reported to vary with male body condition or quality (Malo et al. 2005b; Pitcher et al. 2007; Urbach et al. 2007), female attractiveness (Cornwallis and Birkhead 2007) or male mating tactic (Fitzpatrick et al. 2007; Locatello et al. 2007). In contrast, sperm design is highly heritable (Beatty 1970, 1972; Morrow and Gage 2001a; Birkhead et al. 2005) and may thus be limited in intra-male variation compared to sperm velocity. Although a genetic link between sperm design and motility exists in the zebra finch (Mossman 2008), sperm velocity is likely the result of a combination of factors that can depend on, or interact with, other ejaculate parameters (e.g. see Snook 2005 and references therein). Therefore, future research should be directed at identifying these factors, which would allow us to better control for confounding effects and establish whether sperm design translates into sperm velocity as predicted by theoretical models.

Conclusions

Our data suggest that post-copulatory sexual selection favours longer sperm components and reduces the intra-specific variation in sperm design among the Icteridae. Although theoretical models and a comparative study across the Icteridae indicate that selection on sperm design might be associated with selection for faster sperm, sperm design does not appear to dictate swimming performance within these species, or presently unknown factors may confound such a pattern in these and previously studied species. It is crucial that future research resolves the significance of sperm design for sperm function by understanding how both these sperm characteristics are influenced by environmental or conditional effects and controlling for them.