Sperm morphometry (i.e., size and shape) and function are important determinants of male reproductive success and are thought to be under stabilizing selection. However, recent studies suggest that sperm morphometry can be a phenotypically plastic trait, which can be adjusted to varying conditions. We tested whether different behavioral strategies in aggression between aggressive red and nonaggressive black males of the color polymorphic Gouldian finch (Erythrura gouldiae) can influence sperm morphometry. We show pronounced within-individual phenotypic plasticity in sperm morphometry of male Gouldian finches in three different social environments. Both red and black males placed in intermediate to high competitive environments (high frequency of red males) increased the relative length of their sperm midpiece. By contrast, red males placed in low to intermediate competitive environments (higher frequency of black males) increased the length of the sperm flagellum. Significant changes in stress and sex steroid hormone levels (in response to the competitive environment) appear to influence sperm traits in red but not in black males, suggesting that changes in hormonal levels are not solely responsible for the observed changes in sperm morphometry. These findings imply that males can adjust sperm morphometry across social environments.

Sperm morphometry (i.e., size and shape) is thought to be closely associated with sperm function and therefore male fertilization success and fitness (Birkhead and Pizzari 2002; Pizzari and Birkhead 2002). The association between sperm function and morphometry, including the size of the head, midpiece, and flagellum, is empirically supported across different species (Fitzpatrick et al. 2009; Lüpold et al. 2009), and also within species (Gage et al. 2004; Birkhead et al. 2005; Humphries et al. 2008; Mossman et al. 2009). It seems plausible that intraspecific variation in sperm morphometry causes variation in sperm function, and hence variation in fertilization success and male fitness. Postcopulatory sexual selection and sperm competition in particular is thought to exert selection pressure on sperm function and morphometry stabilizing around an optimal sperm design (Parker 1998; Pizzari and Birkhead 2002; Calhim et al. 2007; Immler et al. 2008). In fact, increased selection due to increased risk of sperm competition has been shown to reduce intraspecific variation in sperm morphometry (Calhim et al. 2007). Despite this, variation at the intraspecific level persists and may be an important determinant of relative reproductive fitness among males.

Several studies have found sperm morphometry to be highly heritable (Woolley and Beatty 1967; Morrow and Gage 2001a; Simmons and Kotiaho 2002; Birkhead et al. 2005) and sperm morphometry is known to exhibit little variation within males (Morrow and Gage 2001b; Immler et al. 2008). However, two recent studies indicate that sperm morphometry can also be—at least to some extent—a phenotypically plastic trait (Crean and Marshall 2008; Morrow et al. 2008), which may be adjusted to population density (and hence intensity of sperm competition; Crean and Marshall 2008). Given that the intensity of sperm competition may vary between males within a species (Parker 1990), and that males alter other ejaculate traits, such as sperm number and sperm motility (Pilastro et al. 2002; Cornwallis and Birkhead 2007), according to varying social conditions, it seems plausible that at least some of the intraspecific variation in sperm morphometry may relate to male quality or strategic investment.

Dominant and subordinate males often use different mating strategies, which may have a pronounced effect on their physiology, including stress (corticosterone) and sex hormone levels (testosterone; Husak and Moore 2008). These hormones are also known to strongly influence ejaculate traits, often with contrasting effects. For example, testosterone is vital for sperm production (Maddocks and Setchell 1988; Jones and Lin 1993; Kast et al. 1998), whereas corticosterone may inhibit the production of sperm (Wingfield and Sapolsky 2003). However, it is unclear how these differences in hormone levels between dominant and subordinate males influence the strategic investment in ejaculate traits, such as sperm number (Gage et al. 1995; Neff et al. 2003; Pizzari et al. 2003) and sperm velocity (Burness et al. 2004; Cornwallis and Birkhead 2007). Furthermore, it is unknown whether changes in the social environment and resulting changes in hormone levels have any impact on sperm morphometry and/or function and also whether males are able to adjust sperm morphometry to different conditions and situations. In the bluegill sunfish (Lepomis macrochirus), for example, dominant (i.e., territorial) and subordinate (i.e., sneaker) males differ in sperm length (Burness et al. 2004), but whether this difference exists due to adjustment by the males to given conditions or whether this is correlated with other traits relating to male reproductive strategy is unknown.

Here we experimentally evaluate within-individual plasticity (i.e., relative changes before vs. after the experiment for the same males) in sperm morphometry in varying social contexts, and the associated changes in aggression and stress hormones in the color polymorphic Gouldian finch (Erythrura gouldiae). Head color in this socially monogamous bird is genetically determined and related to distinctive dominance behaviors: red males are dominant over black males and overtly aggressive, whereas black males are submissive to red males (Pryke and Griffith 2006). These different competitive strategies between morphs are tightly linked to the divergent production of aggression (testosterone) and stress (corticosterone) hormones (Pryke et al. 2007). Red males are highly reactive to changes in the social environment, especially towards the relative densities of their own morph, displaying elevated aggression (testosterone) and stress (corticosterone) responses in socially competitive environments (Pryke et al. 2007). In contrast, the nonaggressive black males follow a more passive strategy, where aggression and stress hormone levels are largely unaffected by the relative competitive environment (Pryke et al. 2007). We took advantage of these behavioral differences between the morphs to create environments with varying social competition to test potential effects on sperm morphometry and function.

Material & Methods


Wild-type Gouldian finches used in all experiments were from a large captive-bred colony (>1000 breeding birds) at the Save The Gouldian Fund Research Facility in Martinsville, Australia (see Pryke and Griffith 2006 for details). All males and females used in the experiment were sexually naive birds (i.e., never reproduced) in their first year of adult plumage (i.e., 1-year olds). From independence from their parents (i.e., 60-days old) and through their molt into adult plumage, all birds were housed with their siblings, as this minimizes aggression between birds (S. R. Pryke, pers. obs.) and provides a standardized housing environment for all birds prior to the experiment.


To test the potential influence of social interactions on sperm morphometry, we placed males in a range of social environments, differing only in the relative frequency of aggressive red (R) and nonaggressive black (B) males (Pryke et al. 2007). Experiments were run during January and February (2008), which coincides with their natural breeding season. Six replicates of four different treatments with varying morph ratios (3R/0B, 2R/1B, 1R/2B, 0R/3B; see also Fig. 1) were created using three unfamiliar (and unrelated) males. This resulted in a total of 36 black males and 36 red males (see Table 1). The males were allowed to interact for 28 days to ensure that they had completed at least one spermatogenic cycle during the experimental treatment. Spermatogenesis (from the initial division of the stem spermatogonia to the mature sperm) in birds has been estimated to take approximately 11 days (Jones and Lin 1993). Each triplet of males was in visual, but not physical contact, with two females (one red and one black) to maintain males in breeding condition. In birds, males continuously release aged, superfluous sperm with the feces during the breeding season, presumably to maintain high quality sperm in their ejaculates (Riley 1937; Quay 1987; Immler and Birkhead 2005). Differences in sperm age between males could potentially occur if males engaged in homosexual copulations (and ejaculation) with other males. However, homosexual copulations have never been observed (in this and other experiments with male-only cages; S. R. Pryke, pers. obs.), suggesting that they have little or no effect on our results.

Figure 1.

Mean (±SE) relative changes for red (white bars) and black (gray bars) males in response to variation in competitive environments (red-black ratio). (A) There were no significant effects of social environment on sperm head length in either red or black males. As the relative frequency of aggressive red males increased, both red and black males (B) increased their relative midpiece length, whereas in low competitive environments (high black frequencies), red males (C) increased flagellum length, whereas flagellum length in black males did not differ significantly across environments. Statistics were performed separately for red and black males.

Table 1.  Sample sizes for both red (R) and black (B) males for which measurements could be obtained both before and after the experiment. Hormone samples were obtained for all birds both times.
3R/0B18 017 012 0
2R/1B12 6 8 5 5 2
1R/2B 612 6 6 5 2
0R/3B 018 016 014


Sperm and blood samples were taken from males immediately prior to and at the end of the experiment; all results reflect these within-individual changes in sperm morphometry and plasma hormone levels. Blood samples were taken prior to sperm samples and within 2 min of capture. Blood samples were centrifuged and plasma stored at −20°C. Testosterone and corticosterone were measured in duplicate from the plasma samples using Cayman Enzyme Immunoassay kits (see Pryke et al. 2007 for details). Sperm samples were collected by manually massaging the cloaca (Thiede et al. 1981): if sperm samples could not be obtained within 5 min of handling, birds were released to avoid stress. For a total of 58 (out of 72) males, both before and after measurements of morphometric sperm traits were obtained (see Table 1). Because of contamination with feces, not all samples could be assessed for sperm swimming velocity. Therefore, sample sizes for measures of swimming velocity (before and after experiment) were lower than the actual number of males involved in the experiment (see Table 1).

Immediately after ejaculate collection, sperm were transferred into phosphate buffered saline solution. Within 5 min after collection, a subsample of the collected ejaculate was used for recording of sperm motility. A subsample was placed on a microscope slide previously painted with strips of nail polish to elevate the cover slip by about 200 μm and allow sperm to move naturally. To standardize the procedure, the same slide was used for all males. The slide was placed under a phase contrast microscope at 100 × magnification with a stationary heater kept at 37°C allowing to record swimming sperm. Video recording started 10 sec after placing the slide on the heater (i.e., allows time for the sample to reach 37°C and sperm reaching maximum swimming velocity—note that sperm kept at room temperature are not motile) and continued until no sperm were moving. The video recordings were analyzed for sperm swimming velocity on a Hobson sperm tracker (property of the University of Sheffield, U.K.) and data of straight-line velocity (VSL), curvilinear velocity (VCL), and average path velocity (VAP) were recorded. Only tracks measured in the first minute were used to standardize the time frame in which velocity was measured as swimming velocity decreases over time. The measurements obtained from the Hobson sperm tracker were filtered and tracks where (1) VSL < 12, (2) VCL = VAP and VCL = VSL, and (3) beat-cross frequency + ALH (amplitude of lateral head displacement) = 0 were excluded (see also Mossman et al. 2009). This filter excludes any data obtained from dead sperm or other particles showing some minimal movement, which can be picked up by the sperm tracker. The mean values of VCL, VAP, and VSL were calculated for each male and only means based on 10 or more tracks were included in the analyses (average number of tracks per male = 56 [10–211]; repeatability [following Lessells and Boag 1987] based on two means per male calculated from 10 tracks each for 20 male: VCL: ri= 0.80, F1,19= 8.88, P < 0.001; VAP: ri= 0.96, F1,19= 9.75, P < 0.001; VSL: ri= 0.81, F1,19= 7.50, P < 0.001).

The remaining sperm sample was fixed in 5% formalin solution (5 parts 40% formaldehyde [equivalent to 100% formalin] in 95 parts phosphate buffered saline solution) for assessment of sperm morphometry. Passerine (oscine) sperm have a very characteristic bauplan, which consists of a distinct spiral-shaped head, a mitochondrial helix (midpiece), which is twisted around the flagellum (and is not as distinct as the mammalian midpiece), and the flagellum (Jamieson 2007; see also Birkhead et al. 2006; Immler and Birkhead 2007). Using brightfield microscopy at 400 × magnification, digital images of five sperm per male were taken. This sample size has been shown to be appropriate for reliable measures of sperm trait dimensions of individual males in passerine birds (see Birkhead et al. 2005; Immler and Birkhead 2007) and repeatabilities based on two means per male calculated from measurements of two sperm (note that this is a conservative estimate as we measured five sperm per male) each for 20 males (Lessells and Boag 1987) for all three traits were intermediate (head: ri= 0.62, F1,19= 4.33, P= 0.001; midpiece: ri= 0.61, F1,19= 4.12, P= 0.001; flagellum: ri= 0.73, F1,19= 6.50, P < 0.001). Lengths of the head, midpiece helix, and flagellum were measured to the nearest 0.1 μm and the number of helical turns of the midpiece was counted using the software Leica IM50 (see also Immler and Birkhead 2007 for further details). From midpiece helix length, we calculated straight helix length (hereafter referred to as midpiece length) according to Birkhead et al. (2005).


We tested for differences between red and black males prior to the experiment using simple t-tests. Furthermore, we investigated the association between sperm morphometry and swimming velocity. Because sperm swimming velocity traits may be correlated with each other, we performed a principal component analyses (PCA) with a varimax rotation including all three measures (VCL, VAP, and VSL) and used the scores of the first principal component (PC1) in statistical analyses. PC1 for the three swimming velocity traits prior to the experiment explained 72% of the variation (loadings: VAP: 0.98, VCL: 0.98); second principal component (PC2) explained 28% based on VCL (loading 0.88). We tested the relationship between the PC1 and all morphometric sperm traits, including the ratio between flagellum and head length, which has been suggested to play a major role in swimming velocity (Humphries et al. 2008).

After the experiment, we tested the effects of the social environment on morphometric and velocity sperm traits and hormone levels. Because of the unbalanced experimental design (i.e., red and black males were not present in all environments), we performed the analyses for black and red males separately. To determine the effect of social environment on sperm and hormone traits, we performed linear mixed effect models including sperm/hormone traits as dependent variable, “morph ratio” (i.e., social environment) as a fixed factor and “cage” as random factor. We also performed a PCA on the changes in sperm velocity traits and tested for an effect of social environment on the resulting PC1 (which explained 65% of the variation with loadings for VAP: 0.96 and VSL: 0.98).

To test for associations between changes in morphometric sperm traits and hormone levels, we included sperm traits as the dependent variable, hormone levels as the independent variable, and cage as the random factor. We performed a PCA including VAP, VSL, and VCL after the experiment: The PC1 of the PCA performed after the experiment explained 65% of the variation based on VAP (loading 0.97) and VSL (loading 0.99) and PC2 explained 34% based on VCL (loading 0.99). We also tested for an effect of social environment on the change in swimming velocity (PC1 explained 65% of the variation with loadings for VAP: 0.96 and VSL: 0.98). To assess the relationship between changes in morphometric sperm traits and changes in swimming velocity traits, we included PC1 as the dependent variable, morphometric traits as independent variables, and cage as a random factor. We calculated the variance explained by random effects (hereafter v.r.e.) for each model, calculated as the random effect variance divided by the total variance after removing the variance explained by the fixed effects (see calculation for heritability value in quantitative genetics; Lynch and Walsh 1998). For multiple post hoc comparisons, we calculated effect size as suggested by Nakagawa (2006). All analyses were performed using the statistical software R version 2.8.1 (R Development Core Team 2008).



We used two-sample t-tests to compare sperm traits including length of head, midpiece, and flagellum and sperm swimming velocity traits between the morphs prior to the experiment. Between red and black males, sperm morphometry differed in midpiece length (mean ± black: 45.7 ± 2.3 μm; red: 42.3 ± 2.1 μm; t67= 6.47, P < 0.0001) and (almost significantly) in flagellum length (black: 61.6 ± 2.7 μm; red: 60.2 ± 3.2 μm; t67= 1.91, P= 0.06) but there were no significant differences for head length (black: 11.6 ± 0.6 μm; red: 11.6 ± 0.5 μm; t67= 0.29, P= 0.63) prior to the experiment. In contrast, there were no differences between morphs for sperm swimming velocity traits, including VSL (black: 37.65 ± 5.91 μm/s; red: 37.47 ± 9.10 μm/s; t50= 0.81, P= 0.42), VAP (black: 41.27 ± 6.44; red: 40.05 ± 11.06 μm/s; t50= 0.42, P= 0.68), and VCL (black: 91.04 ± 7.58 μm/s; red: 90.47 ± 7.54 μm/s; t50= 0.59, P= 0.56), or for testosterone (black: 0.33 ± 0.14, red: 0.37 ± 0.11 ng/mL, t70= 1.21, P= 0.23) or corticosterone levels (black: 3.89 ± 1.89 ng/mL, red: 3.93 ± 1.96 ng/mL, t70= 0.08, P= 0.93) prior to the experiment.


Social environment had a strong effect on morphometric sperm traits (Fig. 1). Although competitive environment had no effect on sperm head length in red males (F2,14= 0.16, P= 0.85, v.r.e. = 0.36), in black males, head length tended to decrease (although not significantly) with increasing social competition (high red frequencies; F2,14= 1.46, P= 0.27, v.r.e. < 0.01; Fig. 1A). Midpiece length increased in both red males (F2,14= 12.02, P < 0.001, v.r.e. < 0.01; Table 2) and black males (F2,14= 11.91, P= 0.001, v.r.e. < 0.01; Table 2; Fig. 1B). By contrast, with decreasing social competition (decreasing red frequencies), red males increased flagellum length (F2,14= 4.71, P= 0.03, v.r.e. < 0.01; Table 2; Fig. 1C), whereas black males exhibited no significant changes of flagellum length (F2,14= 0.17, P= 0.85, v.r.e. = 0.06).

Table 2.  Comparison of changes in sperm morphometry between treatments (for main tests see text). In red males, both midpiece and flagellum length were significantly different between treatments, whereas in black males only midpiece length was significantly different. Statistics obtained from mixed effect models controlling for nonindependence of males by including “cage” as random effect (v.r.e., relative variance explained by random effects). We calculated effect size r and the confidence interval (CI) for each test.
 ComparisonTPd.f. CI
 Red males
  Head1R:2B–2R:1B0.370.71 9 0.610.12−0.59–0.79
 2R:1B–3R:0B0.150.88 9 0.310.05−0.64–0.74
 3R:0B–1R:2B0.580.5710 0.240.18−0.47–0.84
  Midpiece1R:2B–2R:1B2.200.06 9<0.0010.59−0.02–1.37
 2R:1B–3R:0B2.680.03 9<0.0010.67 0.12–1.50
 3R:0B–1R:2B4.520.00110<0.0010.82 0.50–1.81
  Flagellum1R:2B–2R:1B1.020.33 9<0.0010.32−0.36–1.02
 2R:1B–3R:0B1.920.09 9<0.0010.54−0.09–1.30
 3R:0B–1R:2B2.790.0210<0.0010.66 0.14–1.45
 Black males
 1R:2B–2R:1B0.500.63 9 0.730.16−0.53–0.85
 2R:1B–0R:3B1.570.15 9<0.0010.46−0.20–1.19
  Midpiece0R:3B–1R:2B3.710.00410<0.0010.76 0.34–1.65
 1R:2B–2R:1B0.640.53 9 0.730.21−0.48–0.91
 2R:1B–0R:3B4.160.003 9<0.0010.81 0.43–1.82
 1R:2B–2R:1B0.490.64 9 0.730.16−0.53–0.85
 2R:1B–0R:3B0.30077 9 0.120.10−0.59–0.79

Competitive environments had a significant effect on hormone levels. In particular, red males expressed significantly higher testosterone (F2,15= 65.89, P < 0.001, v.r.e. < 0.001, Fig. 2A) and corticosterone levels (F2,15= 52.41, P < 0.001, v.r.e. = 0.28, Fig. 2B) in highly competitive environments. In contrast, black males showed a marginal increase in testosterone (F2,15= 4.27, P= 0.03, v.r.e. < 0.001, Fig. 2A) but the social environment had no effect on corticosterone levels (F2,15= 2.15, P= 0.15, v.r.e. = 0.66, Fig. 2B).

Figure 2.

Changes in testosterone (A) and corticosterone (B) levels across social environments in red (white bars) and black males (gray bars). Statistical analyses were performed separately in the two morphs.

When investigating the relationship between changes in sperm morphometry and hormone levels, we found that in red (but not black) males the within-individual changes in sperm morphometry (midpiece and flagellum length) were significantly related to within-individual changes in testosterone and corticosterone levels (Fig. 3). Red males with elevated testosterone levels had significantly increased midpiece lengths (r= 0.88, P= 0.01, v.r.e. < 0.001; Fig. 3A) and decreased flagellum lengths (r=−0.88, P= 0.009, v.r.e. < 0.001; Fig. 3B), whereas there were no significant associations between testosterone and sperm traits for black males (midpiece: r= 0.04, P= 0.31, v.r.e. = 0.03; flagellum: r=−0.02, P= 0.34, v.r.e. < 0.001). Similarly, red males with elevated corticosterone levels had significantly longer midpiece lengths (t15= 4.68, P= 0.0003, v.r.e. < 0.001; Fig. 3C) and shorter flagellum (t15=−2.15, P= 0.05, v.r.e. < 0.001; Fig. 3D), whereas there were no associations between corticosterone levels and sperm traits for black males (midpiece: t9=−0.01, P= 0.99, v.r.e. = 0.02; flagellum: t14=−0.31, P= 0.77, v.r.e. < 0.001).

Figure 3.

Changes in sperm morphometry and hormone levels for red (white circles and broken line) and black males (gray circles and solid line). Red males with elevated testosterone levels had significantly increased (A) midpiece length and decreased (B) flagellum length, whereas there were no significant associations between testosterone and sperm traits for black males. Similarly, red males with elevated corticosterone levels had significantly (C) longer midpiece lengths, and (D) shorter flagellum, whereas there were no associations between corticosterone levels and sperm traits for black males. Regression lines show significant relationships in red males. Statistics were performed separately for red and black males.

Flagellum length and flagellum:head ratio showed a significant positive association with sperm swimming velocity (PC1) prior to the experiment (Table 3A, Fig. 4A,C) but not after (Table 3B, Fig. 4B,D). Changes in swimming velocity traits were not significantly associated with variation in competitive environments (Table 4).

Table 3.  Relationship between sperm motility (PC1) and sperm morphometry controlling for morph at the beginning of the experiment (A) and at the end of the experiment (B). (A) Statistics were obtained from multiple regression analyses using values for individual males. (B) Statistics were obtained from a mixed linear model with “cage” as random factor to account for nonindependence between males (v.r.e., relative variance explained by random effects).
Midpiece 2.090.15 0.06
Morph 1.080.30 
Flagellum 8.810.005 0.16
Morph 0.440.51 
Flagellum:head 4.710.035 0.09
Morph 0.130.71 
PredictorMorphFPdf v.r.e.
HeadRed 4.430.061,9 0.45
MidpieceRed 0.660.441,9 0.32
 Black 1.530.241,12<0.001
FlagellumRed 0.700.421,9 0.36
Black 1.080.321,12<0.001
Flagellum: headRed 0.900.371,9 0.38
 Black 0.640.441,12<0.001
Figure 4.

Association between sperm swimming velocity (PC1 in both cases based on VAP and VSL) and sperm morphometry before (A,C) and after (B,D) the experiment. Regression lines indicate the significant associations between PC1 and both flagellum length (A) and flagellum:head ratio (C) across males of both morphs prior to the experiment. In (B,D) white circles and broken lines represent red males and gray circles and solid lines represent black males. For the data obtained after the experiment (B,D) analyses were performed separately for red and black males.

Table 4.  Changes of sperm swimming velocity in response to competitive environment. Statistical values were obtained from a linear mixed effect model with “cage” as random factor (v.r.e., relative variance explained by random effects).
 MorphFPdf v.r.e


Our results demonstrate pronounced and rapid within-individual changes in different social environments for two important morphometric sperm traits—the length of midpiece and flagellum. Here we discuss our findings in the light of male reproductive strategies and possible mechanisms.


The relationship between morphometric sperm traits and sperm function has been explored theoretically (Humphries et al. 2008) and empirically, both within (Malo et al. 2006; Mossman et al. 2009) and between species (Fitzpatrick et al. 2009; Lüpold et al. 2009; Kleven et al. 2009). However, how changes in sperm morphometry affect sperm function and fitness is still poorly understood. To date, the only evidence comes from a study of the hermaphroditic broadcast spawning ascidian (Styela plicata) where an increase in sperm head length was associated with an increased fertilization success (Crean and Marshall 2008). Changes in social structures may cause a change in a male's mating opportunities and the level of sperm competition he is exposed to and adaptations in sperm morphometry and function may be the consequence. In the jungle fowl, for example, subordinate males produce more motile sperm, however, when given the opportunity to be dominant, sperm motility decreases (Cornwallis and Birkhead 2007). Whether this change in motility is due to a change in sperm morphometry needs to be tested. Similarly, in some marine gobies, young, small males use alternative mating strategies and their sperm differ in function from sperm of older, larger territorial males (e.g., Locatello et al. 2007). As these males grow older and larger, they switch reproductive tactics to territorial male and their sperm characteristics change accordingly. However, this has not been tested at the intramale level. Finally, a recent study in the house mouse, Mus musculus domesticus has demonstrated phenotypic plasticity in daily sperm production and total number of sperm produced in response to varying social environment (number of rival males present and hence perceived intensity of sperm competition), but they did not look at sperm morphometry (Ramm and Stockley 2009). It is possible that in the Gouldian finch a change in social structure affects a male's mating opportunities, but whether this is linked to postcopulatory sexual selection and sperm competition needs to be further investigated.

The divergent effects of social competition on the size of the sperm flagellum and the midpiece reported here suggest that there may be a direct structural or physiological trade-off between these traits. Such a trade-off between individual sperm traits could also partly account for our failure to find a difference in sperm swimming velocity between competitive environments. In addition, it may explain why the initially observed association between sperm morphometry and swimming velocity disappeared over the course of the experiment. Alternatively, a longer midpiece and shorter flagellum may be adaptive in a stressful or competitive social environment, and perhaps relate to functions other than swimming velocity, such as swimming stamina or longevity in the female's sperm storage tubules. In mammalian sperm, the relative sizes of the flagellum and the midpiece have been suggested to influence sperm longevity (Cardullo and Baltz 1991). Across passerine species, sperm longevity decreases with increasing absolute size of midpiece and flagellum (S. Immler and T. R. Birkhead, unpubl. data), but whether a similar pattern can be found at the intraspecific level is not known.


Stress and sex hormones are thought to play a crucial role in the expression of male sexual traits, including secondary sexually selected traits, as well as sperm and ejaculate (Husak and Moore 2008). A recent study of male red deer (Cervus elaphus) found that males with higher plasma testosterone levels had larger testes and improved sperm quality (i.e., higher ratio of normal sperm and faster swimming sperm) but also stronger antlers and higher hematocrit levels (Malo et al. 2009). In contrast, in a field study of the dark-eyed junco (Junco hyemalis), artificial manipulation of testosterone levels using implants showed no effect on ejaculate quality in terms of sperm numbers (Kast et al. 1998). Our findings suggest that, although testosterone (and corticosterone) may provide the physiological basis underlying agonistic behavior (see also Pryke et al. 2007), differential hormonal responses of the morphs to competitive environments are unlikely to drive the divergent variation in sperm traits alone. For example, in black males both testosterone and corticosterone levels were unaffected by competitive environments (see also Pryke and Griffith 2007), yet black males significantly increased the length of their midpiece and decreased head length, although this was not statistically significant in highly competitive environments. Thus, it seems that the mechanisms involved in producing rapid changes in sperm morphometry are remarkably complex. A recent comparative study across passerine birds further supports this view: a path analysis suggested that the phylogenetic association between testis size (and hence larger ejaculates) and levels of extra-pair paternity is stronger that that between testis size and testosterone levels (Garamszegi et al. 2005).


The step-wise changes in sperm morphometry in association with an increase in social stress and aggression levels suggest that strategic adjustment of sperm morphometry is possible. The adaptive significance of these adjustments cannot be fully understood at this stage, because this will depend on the mating strategies adopted by red and black males. Dominant red males have privileged access to limited food and roosting places (Pryke and Griffith 2006) and it seems likely that they would also have greater access to females. In this situation, we might expect red males in a black dominated environment to produce less competitive sperm (shorter midpiece and longer flagellum), as they have privileged access to females and hence higher copulation frequencies. In contrast, both red and black males in a red dominated environment would be expected to produce more competitive sperm with a longer midpiece. However, these predictions remain to be tested in relation to the alternative strategies adopted by red and black males.


The rapid, pronounced within-male changes in sperm morphometry across different socially competitive environments have not previously been empirically demonstrated, or theoretically anticipated. Although vertebrate sperm appear superficially to be relatively conservative in morphometry at the intraspecific level (Morrow and Gage 2001b; Calhim et al. 2007; Immler et al. 2008), our findings suggest that environmental changes can cause marked changes in several morphometric sperm traits. One hypothesis for the pronounced step-wise changes in sperm morphometry reported here is that males are able to strategically adjust sperm morphometry according to selective pressures caused by varying social environments, which in turn may affect the intensity of postcopulatory sexual selection. However, this hypothesis needs to be carefully tested. Our study highlights the need to account for the largely neglected, environmentally derived within-individual plasticity of sperm morphometry.

Associate Editor: P. Stockley


We thank L. Astheimer for help with the hormone analyses, B. Montgomerie and C. Cornwallis for comments on the article and J. Biggins, G. Arnqvist, and H. Schielzeth for statistical advice. This work was supported by an Australian Research Council (ARC) Discovery Grant DP0770889 (SRP), an ARC International Linkage Grant LX0668753 (SCG and TRB), a Leverhulme Grant F/00118AJ (TRB), and Save The Gouldian Fund.