Behavioral Ecology Vol. 13 No. 6: 808-815
© 2002 International Society for Behavioral Ecology
Male reproductive success and sexual selection in northern water snakes determined by microsatellite DNA analysis
a Department of Biology, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada b Department of Biology, McMaster University, Hamilton, Ontario L8S 4K1, Canada
Address correspondence to P.J. Weatherhead, who is now at the Program in Ecology and Evolutionary Biology, University of Illinois, Shelford Vivarium, 606 E. Healey, Champaign, IL 61820, USA. E-mail: pweather{at}uiuc.edu. H.L. Gibbs is now at the Department of Evolution, Ecology and Organismal Biology, Botany and Zoology Building, The Ohio State University, 1735 Neil Avenue, Columbus, OH 43210-1293, USA. G.P. Brown is now at the School of Biological Sciences, A08, University of Sydney, Sydney, NSW, Australia.
Received 2 January 2001; revised 29 October 2001; accepted 12 March 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Male northern water snakes (Nerodia sipedon) have high variance in reproductive success relative to females. We used DNA-based paternity analyses from a 3-year study of two marsh populations of water snakes to investigate the factors that contribute to variation in male success. Male traits investigated included body size, condition, tail length, home range size, activity during the mating season, and genetic profile (genetic similarity to females, heterozygosity, and genetic variability [d2]). We successfully assigned > 80% of offspring to sires from a sample of 811 offspring from 45 litters. Male reproductive success did not vary significantly with body size, tail length, condition, home range size, or the number of microsatellite loci at which males were heterozygous, nor with other features of their genetic profiles. However, we found evidence of positive assortative mating by size in the marsh in which receptive females were not spatially clumped. Also, males that were most active during the mating season were more successful, particularly where females were not clumped. We failed to find evidence of selection acting on male size through variance in reproductive success, indicating that sexual selection does not have an important influence on sexual size dimorphism in this species (males are smaller than females). We propose that males are smaller than females because the lack of advantage to large size allows males to adopt a low-energy, low-growth strategy that reduces their risk of predation outside the mating season.
Key words: assortative mating, DNA loci, heterozygosity, microsatellites, Nerodia sipedon, northern water snake, paternity analysis, sexual selection, sexual size dimorphism.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Snakes are underrepresented as subjects of sexual selection research, despite their rich variation in mating behavior and sexual size dimorphism, which suggests there is much to be learned from their study (Duvall et al., 1993
;
Olsson and Madsen, 1998;
Shine, 1993). Sexual size
dimorphism in snakes is of particular interest because available evidence is
poorly explained by current theory. In many species of snakes males are
smaller than females, despite intense mating competition among males (e.g.,
Madsen and Shine, 1993b;
Weatherhead et al., 1995).
Furthermore, in these same species, larger males appear to have a mating
advantage. These patterns are contrary to the conventional view that intense
mating competition among males is associated with male-biased sexual size
dimorphism (Darwin, 1871),
particularly given evidence that large males have an advantage. DNA-based
analyses of paternity have the potential to contribute fundamentally to our
understanding of sexual selection in snakes
(Duvall et al., 1993;
Gibbs and Weatherhead, 2001).
Snakes are often difficult to observe systematically in the wild, and even
when observations are possible, they may be unreliable predictors of parentage
(Prosser et al., 2002). We
used genetic methods to identify the males that sire offspring, and thus also
to identify those males that do not, in a population of northern water snakes
(Nerodia sipedon). We used this approach to identify the factors that
contribute to variation in male reproductive success and that therefore may be
subject to sexual selection, toward our general goal of understanding why male
water snakes are smaller than females
(Brown and Weatherhead,
1999c).
Male northern water snakes compete for access to females through scramble
competition. Two previous studies of the population we report on here
identified factors affecting male reproductive success by quantifying
observations of mating behavior in the field. Weatherhead et al.
(1995) found that large males
had a mating advantage in 1 of 2 years, whereas relative tail length appeared
unrelated to mating success. Brown and Weatherhead
(1999b) investigated the effect
of both behavioral and morphological attributes of males on mating success, as
well as how the spatial distribution of females affected males. Where females
were dispersed, males increased the size of their home ranges, and males with
larger home ranges encountered more females. The opposite patterns occurred
where females were spatially clumped. Where females were dispersed, larger
males and males in better condition appeared to have a mating advantage, but
neither factor appeared to affect male success when females were clumped.
Evidence that behavioral data poorly predict productive success in northern
water snakes (Prosser et al.,
2002) renders the results of these previous studies suspect. Thus,
we reexamined all of these results using genetic measures of paternity.
Understanding why male water snakes are smaller than females requires
knowing how body size affects male reproductive success. In addition to the
evidence for water snakes, there is behavioral evidence of a large-male mating
advantage in other snake species in which males compete for matings by
scramble competition (Madsen and Shine,
1993a; Shine et al.,
2000). It is possible that genetic evidence will confirm that
larger males have a reproductive advantage because of superior competitive
ability. If so, the question arises as to what selection pressure prevents
males from increasing in size to become more similar to females
(Weatherhead et al., 1995).
Alternatively, it is possible the genetic evidence will show that small males
have a reproductive advantage perhaps because smaller males are more mobile
(Shine, 1978). We consider
various alternative explanations for sexual size dimorphism in northern water
snakes in light of what our results reveal about the relationship between male
body size and reproductive success.
Variation in body size could also affect reproductive success through
size-assortative mating. Larger female water snakes are more fecund
(Brown and Weatherhead, 1997;
Weatherhead et al., 1999) and
therefore should be preferred by males. If large males have a competitive
advantage, large males should tend to produce offspring with large females.
Shine et al. (2001) reported
this pattern in red-sided garter snakes. Thus, in addition to testing the
prediction that larger males will be more successful, we also tested the
prediction that there will be a positive correlation between the size of males
and females that produce offspring together.
Although our central interest is body size, we also considered other
morphological attributes of males that could affect reproductive success.
Relative to body size, male snakes have longer tails than females
(King, 1989). Males
"wrestle" with their tails when courting females, and the length
of a tail may affect success in this competition. Shine et al.
(1999) reported a mating
advantage to male red-sided garter snakes with intact tails relative to
individuals that had lost part of their tail. Among all males, Shine et al.
(1999) found stabilizing
selection acting to maintain sexual dimorphism in relative tail length. Thus,
we tested the prediction that males with tails that are either longer or of
intermediate length should be more successful. We also tested the prediction
that males in better condition (mass corrected for body size;
Weatherhead and Brown, 1996)
at the start of the mating season will be more successful. Locating and
competing for mates seems likely to be energetically costly, so males in
better condition should be able to devote more energy to mating.
We studied two populations of water snakes that differed substantially in
the extent to which females were spatially clumped
(Brown and Weatherhead, 1999b)
and predicted that male attributes should affect reproductive success
differently in the two populations. Where females are clumped, and thus
competition among males is higher, larger males and males with relatively
longer tails should have a greater reproductive advantage. Where females are
more dispersed, we predicted that the most successful males should be those
that were more active, in better condition, and with larger home ranges
because these attributes should enhance mate-searching ability.
A final question we addressed is whether genetic differences affect male
reproductive success directly. If females choose among potential mates, they
may prefer genetically dissimilar males to avoid the costs of inbreeding
(Charlesworth and Charlesworth,
1987; Partridge,
1983). Females might also prefer the most heterozygous males to
increase the heterozygosity, and thus the fitness, of their offspring
(Brown, 1997). Recently,
Pemberton et al. (1999) showed
that offspring fitness was positively related to measures of individual
genetic variation in two species of mammals. Even without active female
choice, multiple mating by females in combination with sperm competition could
result in some males having a reproductive advantage. For example, Olsson et
al. (1996) found that although
female sand lizards (Lacerta agilis) appeared to mate randomly,
unrelated males sired more offspring. We tested two predictions. First,
male(s) that sire a female's offspring will be more dissimilar genetically
from the female than expected by chance. Second, male reproductive success
will be positively correlated with individual male heterozygosity.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The methods we used to collect blood for DNA analysis and to assign paternity are explained in detail in a companion paper from the same study (Prosser et al., 2002
). We
collected DNA samples from 811 offspring from 45 litters that were
collectively sired by 48 males. Overall, 92.8% of offspring were assigned to
sires in one study population (Barbs Marsh) and 68.8% of offspring to sires in
the other study population (Beaver Marsh). Across both marshes 58% of litters
were sired by more than one male. All unassigned neonates were attributed to
unsampled males.
The study populations and the individual snakes we studied here were also
the same as those used by Brown and Weatherhead
(1999b), so field methods for
the two studies are identical. Therefore, we only provide a brief overview of
those methods here. We conducted this study from 1994 to 1996 at two beaver
ponds (Barbs Marsh and Beaver Marsh) near the Queen's University Biological
Station in eastern Ontario. We captured snakes by hand in the two marshes,
marked them individually, and collected blood for paternity analysis. We
weighed each individual and measured total length, tail length (cloaca to tail
tip), and determined snout-to-vent length (SVL) by subtracting these two
measures. To estimate condition we used Weatherhead and Brown's
(1996) equation to estimate
the lean mass of each male from its SVL. By subtracting lean mass from total
mass, we estimated total fat. Condition is total fat expressed as a percentage
of total body mass. To control for growth during the active season, we used
the equation in Brown and Weatherhead
(1999c) to adjust all SVL
measures to the median capture date for that year. No formula was available to
adjust tail length, but we assumed that any growth during the season would be
less than our measurement error.
In late July we searched both marshes intensively to capture all adult
females (Weatherhead et al.,
1995) and brought them into the lab until parturition. We returned
females to the marshes after they gave birth or it became apparent they were
not gravid. To obtain DNA from neonates we collected 50 µl of blood from
the caudal vein of each live young and removed the terminal 1 cm of the tail
from young that were stillborn. After sampling we released neonates at their
mother's marsh. When assigning paternity for a given litter, we screened all
males sampled in the marsh from which that litter came, including males
sampled in years other than that in which the litter was produced. We used
eight highly variable microsatellite DNA loci for assigning paternity that had
a combined probability of exclusion across all loci of > 0.99 in both
marshes (Prosser et al., 1999,
2002).
Male spatial patterns
We collected data on the spatial patterns of males using observations of
marked individuals collected during the mating seasons. We searched both
marshes intensively for several hours at least once each day beginning in
mid-April before the snakes emerged from hibernation and continuing until
mid-June after all mating activity had ceased. Snakes captured during this
period were individually marked with spots of non-toxic, acrylic paint, in
addition to being given PIT (passive integrated transponder) tags. Paint marks
allowed us to identify snakes at a distance without disturbing them. Most
adults were measured and marked soon after emergence from hibernation, before
mating began. We also used radiotelemetry to monitor the activity of a sample
of snakes throughout the active season
(Brown and Weatherhead, 1999b).
Transmitters (Holohil Systems Ltd., Ottawa, Ontario) were surgically implanted
in the snakes' coelom. Transmitter mass was < 7.5% of body mass of each
snake. We tracked 19 males during the mating season, during which time we
located each individual at least once a day.
Scale maps of both study sites were overlain with a 20 m x 20 m grid.
Each time we saw a painted snake or located a snake with a transmitter, we
recorded its grid location. We estimated home ranges using the number of grid
squares in which an individual was seen, using only individuals seen a minimum
of five times. This method has the advantage of simplicity, and home range
estimates derived using this method were highly correlated with estimates from
more conventional methods (e.g., correlation with concave polygon method:
r = 0.81, p < .0001). Because home range size was
correlated with the number of times a male was observed
(Brown and Weatherhead, 1999b),
we included the number of observations as a factor in multiple regressions
investigating the effects of home range area on reproductive success. We also
used the number of times we observed males without transmitters as an index of
their activity. Males that were more actively searching for females should
have been easier for us to see and should also have been more likely to find
females. Brown and Weatherhead
(1999b) found that males
observed attempting to mate were seen more than twice as often overall during
the mating season as males never seen courting a female.
We restricted analysis of spatial patterns (and all other patterns) to
adult males (individuals potentially capable of siring offspring). We assumed
all males longer than 43.2 cm SVL were sexually mature, based on observations
of mating behavior (Weatherhead et al.,
1995) and on the smallest reproductively successful male
(Prosser et al., 2002). The
two methods were consistent to within < 1 cm SVL.
Data analysis
We conducted analyses for each year separately and pooled across years for
each marsh. We considered pooling across years justified for several reasons
(Prosser et al., 2002).
Particularly relevant here is that water snakes continue to grow after sexual
maturity, so most attributes of interest (e.g., body size) of individuals
sampled twice change between samples. Also, among males, where the proportion
of resampled individuals is highest, the variance in reproductive success was
as high within as among resampled individuals (F = 0.67, df = 36,42,
p = .89), suggesting that pooling data should not bias our results
(Leger and Didrichsons, 1994).
We inspected plots and distributions of residuals from all parametric analyses
to determine whether our data met assumptions of normality and homogeneity of
variance. We used transformations where necessary to meet those assumptions,
and when transformations were unsuccessful we used nonparametric analysis.
To assess the effects of male traits on reproductive success, we calculated
selection gradients using the standardized partial regression coefficients of
each character from a multiple regression of relative reproductive success
(Lande and Arnold, 1983).
Relative reproductive success (w'i) was estimated by
dividing an individual's success by the mean success of the sample. All
morphological traits were standardized (zi) to have a mean
of zero and standard deviation of 1 for pooling across years. We estimated
directional selection from the partial regression coefficients from a model
containing standardized linear terms for tail length, SVL, and condition:
 | (1) |
 | (2) |
The coefficient ß1 represents the intensity of directional selection on each trait after the effects of correlated traits have been removed. A positive coefficient indicates that directional selection is acting to increase that trait. The coefficients of the quadratic terms (ß2) represent stabilizing or disruptive selection. Stabilizing selection is inferred to be acting on traits if the sign (positive or negative) of the quadratic term is opposite to that of the corresponding linear term.
We performed three variations of these analyses. First, we excluded males with obviously damaged (and shortened) tails. Second, we included only males that sired offspring in case unsuccessful males included some individuals that had not attempted to mate, as opposed to having been unsuccessful. Third, for the analyses involving data pooled across years, we used a mean trait and reproductive success value for each male present in multiple years. Because none of these analyses produced results qualitatively different from the original analyses, we do not present results from these alternative analyses.
To determine whether parentage was assortative with respect to body size, we regressed the SVL of successful males on the SVL of the female for each litter in which all fathers were identified. Regressions were weighted by the proportion of the litter sired by each father. Thus, males siring more offspring in a given litter were given a higher weighting than males siring fewer offspring in that litter. We excluded data from Barb's Marsh in 1995 because we only obtained one litter.
To determine whether parentage of offspring was nonrandom with regard to
the similarity of the parents' genetic profiles, we generated relatedness
scores (i.e., r values; Queller
and Goodnight, 1989) using Kinship 1.2
(Goodnight and Queller, 1999)
for every male and female pairwise combination in each population, pooled
across years. Therefore, individuals were included in each year they were
present. We generated distributions of r values, pooled into
intervals that differed by 0.1 units of r, for pairs that produced
offspring and for those that did not. We then assessed the significance of the
differences in these two distributions of r values using contingency
table analysis based on a randomization technique (STRUC subroutine of
GENEPOP; Raymond and Rousset,
1995).
To test the prediction that levels of individual genetic variation should
increase male reproductive success, we used only males that had been scored at
four or five of the same loci (Nsu 2, 4, 6, 9, and 10). We assumed that
variation in allele sizes at those loci was due to mutations causing the gain
or loss of one or a few repeats, and hence consistent with a stepwise mutation
model (cf. Valdes et al.,
1993). This assumption is supported by the roughly continuous
frequency distributions shown by alleles of different sizes in these
populations (Prosser et al., unpublished data). We estimated levels of
individual variation in two ways. First, we simply determined the percentage
of microsatellite loci at which each male was heterozygous. Second, we used
the method of Pemberton et al.
(1999) to estimate a measure
of individual internal genetic distance, d2, from the
difference in allele sizes at each locus. Formally, this measure of variation
(mean d2) is calculated as:
 | (3) |
).
Individual heterozygosity should reflect recent matings between relatives and
hence be a more sensitive measure of any inbreeding effects that are present.
In contrast, d2 should be more affected by events deeper
in a pedigree that affect allele length variation, and hence should be more
sensitive to outbreeding effects, in that it assumes that individuals with
high d2 scores are more likely descended from matings
between parents from genetically divergent populations. |
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Male morphology
Selection gradients revealed little evidence of selection acting on male morphology (Table 1). There was weak evidence of curvilinear selection on SVL in Barbs Marsh in 1996 and on condition at Beaver Marsh in 1996. However, given that the probabilities presented in Table 1 are not corrected for multiple comparisons, and that the overall models at each marsh were not significant, the simple inference from these analyses is that variation in male morphology did not affect variation in male reproductive success. It is possible that low statistical power may have limited our ability to detect selection. The mean absolute value of all our estimates of selection was 0.22 (Table 1). Power calculations indicate that the mean least-significant values of our regressions was 0.62, so selection on the traits we measured would have had to have been about three times greater, on average, before we could have detected it with a power of 50%. Thus, although we can be confident that no strong selection was acting on the traits we measured, we are less certain regarding more modest levels of selection. However, we replicated our analyses both spatially and temporally (which is rare among selection studies; Kingsolver et al., 2001
) and
drew the same conclusions from them all, which strengthens the inference that
selection on the male traits we considered was unimportant.
|
It is possible that male morphology does not affect mating success but does
affect sperm competition after mating. To assess this possibility we compared
traits of males that sired young in the same litter. The simplest approach was
to restrict the analysis to litters with only two sires, which were also the
most common among multiply sired litters. We had 11 litters for which both
sires were known and current morphological measurements were available. Within
litters the most successful male sired significantly more offspring than the
other male (74% vs. 26%, t = 8.9, df = 20, p < .0001). We
used logistic regression to determine whether male attributes (SVL, tail,
mass, condition) affected which male was more successful. None of the
attributes had a significant effect (all 
2 < 2.04, all
p > .15). A regression analysis of the difference in reproductive
success relative to the difference in morphology of sires within litters also
showed no significant effects. The result that was closest to being
significant was that the discrepancy in male success tended to increase as the
more successful male became larger relative to the less successful male
(F1,9 = 2.99, r = .50, p = .11).
Despite the lack of evidence that male morphology affected male reproductive success, we did find some evidence that mating was nonrandom with regard to male morphology. In Barb's Marsh, weighted regression indicated the SVL of females was significantly correlated with that of their mates (F1,23 = 5.99, p = .02, Figure 1). However, we found no evidence of this pattern in Beaver Marsh (F1,18 = 0.22, p = .65, Figure 1). In both these analyses we included each female—male pair that produced offspring, so a female with multiple mates was included once for each mate that we identified. Repeating the analysis using the mean size of each female's mates weakened the results. The positive assortative mating by size in Barbs Marsh became nonsignificant (F1,14 = 3.21, p = .10), while mating in Beaver Marsh remained random with respect to male and female size (F1,11 = 0.18, p = .65).
|
Male activity and home range
We used logistic regression to determine whether males observed more often
during the mating season, and thus assumed to be more active, were more likely
to achieve reproductive success. In Barbs Marsh, males that were seen more
often were more likely to be successful (2 = 6.36, N
= 67, p = .012). The same was true in Beaver Marsh, but the effect
was not significant (2 = 2.34, N = 114, p =
.13). In the two marshes combined over all years, successful males were seen
about twice as often as unsuccessful males (t = -3.18, df = 179,
p = .002, Figure
2).
|
In Barbs Marsh, males seen in more grid squares were more likely to be
reproductively successful (2 = 5.26, N = 57,
p = .022). In Beaver Marsh, the number of different grids a male
occupied did not affect the likelihood of being successful (2
= 0.35, N = 81, p = .55). A potential problem with this
analysis is that the estimated size of home ranges increased with the number
of times a male was observed (Brown and
Weatherhead, 1999b). We used multiple logistic regression to
examine the relation between male reproductive success and home range size,
controlling for the number of times a male was observed. In both marshes home
range size did not contribute significantly to the model in any year or
overall (Barbs Marsh overall: 2 = 0.88, N = 57,
p = .35; Beaver Marsh overall: 2 = 0.05, N =
81, p = .82).
Genetic similarity and variability
Mean relatedness scores (r values;
Queller and Goodnight, 1989)
for pairs that produced offspring together were similar to those for all pairs
of males and females from the same marsh that did not produce offspring
together in both Barbs (0.014 and 0.073) and Beaver Marsh (-0.039 and 0.005).
A randomization analysis (STRUC subroutine of GENEPOP;
Raymond and Rousset, 1995)
indicated that the distribution of r values for reproductive and
non-reproductive pairs of snakes were not significantly different in either
marsh (Barbs: p = .64; Beaver p = .16). Thus, water snakes
appear to mate at random with respect to genetic similarity.
Our first test of the prediction that individual variation should increase
male reproductive success simply involved determining whether male
reproductive success varied with the percentage of microsatellite loci at
which males were heterozygous. Male success did not vary significantly with
heterozygosity at Barbs Marsh in either year (1994: Spearman r = .19,
N = 36, p = .26; 1996: Spearman r = .17, N
= 19, p = .49) or overall (Spearman r = .10, N =
55, p = .47). The same was also true at Beaver Marsh (1994: Spearman
r = .15, N = 40, p = .36; 1995: Spearman r
= -.26, N = 36, p = .13; 1996: Spearman r = .17,
N = 21, p = .46; overall: Spearman r = -.02,
N = 97, p = .87). Our second method substituted natural
log-transformed mean d2
(Pemberton et al., 1999) as
our measure of individual variability. In Barbs Marsh male success did not
vary significantly with mean d2 in either year (1994:
Pearson r = -.15, N = 36, p = .37; 1996: Pearson
r = -.17, N = 19, p = .48) or overall (Pearson
r = -.16, N = 55, p = .23). Similarly, in Beaver
Marsh male success did not vary with mean d2 in any year
(1994: Pearson r = .24, N = 40, p = .15; 1995:
Pearson r = .09, N = 36, p = .61; 1996: Pearson
r = -.08, N = 21, p = .74) or overall (Pearson
r = .13, N = 97, p = .21).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Analysis of the relative variance in reproductive success for males and females in our study population of northern water snakes revealed that the potential for sexual selection was approximately five times greater for males (Prosser et al., 2002
).
Despite this potential, however, paternity analyses revealed no evidence that
any male traits were consistently or strongly targeted by sexual selection.
Male reproductive success did not vary significantly with body size, tail
length, or condition. Even within litters sired by two males, the relative
success of the two males was unrelated to their absolute or relative sizes,
tail lengths, or condition. Male success also did not vary significantly with
home range size, even though males responded to the highly clumped
distribution of reproductive females in Beaver Marsh by having much smaller
home ranges than those of males in Barbs Marsh
(Brown and Weatherhead, 1999b).
Finally, reproductive success did not vary with the number of microsatellite
loci at which males were heterozygous, with individual genetic variability
(d2), or with their genetic similarity to females. We did find some evidence that mating was not completely random. First, we found evidence of positive assortative mating by size in Barbs Marsh, although we found no evidence of assortative mating in Beaver Marsh. Second, males that were more active during the mating season were more successful, particularly in Barbs Marsh. We first discuss possible reasons that no male morphological features correlated with reproductive success and then discuss the implications of the few factors we did find that were correlated with male success. We conclude by considering the implications of all these results for the central goal of explaining sexual size dimorphism in water snakes.
In any study that involves assignment of paternity, unassigned young could
affect the results. Overall, we successfully assigned just > 80% of
offspring. At Barbs Marsh our assignment success was 93%, with only three
females producing young sired by unknown males. A 3-year telemetry study
indicated that both males and females rarely moved out of Barbs Marsh other
than to hibernate (Brown, unpublished data). Thus, the unassigned offspring in
Barbs seem likely to have been sired by males we failed to capture. It also
seems likely that snakes we considered resident in Barbs were not leaving to
mate with snakes elsewhere. Thus, our assessment of patterns of reproductive
success in Barbs is probably an accurate reflection of the true patterns. At
Beaver Marsh just fewer than one-third of all offspring were unassigned. The
capture patterns of females producing those young suggested there may have
been some mating activity by Beaver snakes else-where
(Prosser et al., 2002),
although telemetry again indicated strong site fidelity by snakes in Beaver
Marsh. For the mating activity outside the marsh to alter our results would
require that mating patterns outside Beaver Marsh were strongly and
consistently different from patterns within Beaver Marsh. We can envision no
plausible scenario for this happening, particularly given the overall
similarity of patterns we documented in Barbs and Beaver Marshes.
Sampling error could also have occurred if we failed to capture some gravid
females and thus did not attribute any reproductive success to the males that
sired their offspring. We have no way to assess this possibility directly.
However, it seems unlikely to have been a problem because of the behavior of
gravid females. Thermoregulatory demands cause females to bask more when
gravid (Brown and Weatherhead,
2000), making them much easier to find than other snakes. They
also are probably easier to capture because they are gravid. Because we
searched marshes intensively for gravid females, and because we caught many
females that were not gravid, we think it unlikely that we missed any gravid
females. Thus, it is reasonable to conclude that we have documented an
accurate picture of mating patterns in the water snake populations we
studied.
We predicted that more active males (i.e., those we saw more often) should
be more successful based on Brown and Weatherhead's
(1999b) observation that male
water snakes seen courting females were observed more than males not seen
courting. It is possible that courting males are more visible than noncourting
males or, alternatively, that courting males were more active than noncourting
males. The increase in reproductive success with the number of times males
were seen supports the latter interpretation. Observations of courtship did
not reliably predict with which male(s) a female mated
(Prosser et al., 2002), so
males we saw courting more must also have courted more when we did not observe
them. Madsen et al. (1993)
found that the distance traveled by male adders during the mating season was a
significant predictor of the number of females mated. In other taxa with
scramble mating competition among males (e.g., voles:
Gaulin and FitzGerald, 1989;
ground squirrels: Schwagmeyer,
1988), searching ability of males also appears to be important to
male success (Andersson,
1994).
Given that males that were more active during the mating season were more
successful, it seems surprising that male condition at the start of the mating
season was not correlated with reproductive success. However, Brown and
Weatherhead (1999b) found that
male water snakes do not lose weight over the breeding season. Furthermore,
males have been observed feeding during the mating season (Brown and
Weatherhead, personal observations), so, at least in this population, mating
does not rely on the use of stored fat. Because male mortality increases after
sexual maturity, and mortality is highest during the mating season
(Brown and Weatherhead, 1999a),
the cost of greater activity by males appears to be in the form of increased
risk of predation rather than increased risk of starvation.
We predicted that males that were genetically more variable or genetically
different from particular females would be more successful, either through
direct female preference or through sperm competition. Madsen et al.
(1999) recently demonstrated
experimentally that male European adders introduced into an inbred population
from elsewhere were disproportionately successful. We found no evidence that
any aspect of male genotypes we considered affected mating success. Multiple
mating by females, as indicated by the frequency of multiply sired litters,
should have provided adequate opportunity for sperm competition. Furthermore,
although there may be limited gene flow between water snake populations, they
do not appear to be highly inbred (Prosser
et al., 1999). It may be that genetic factors such as
heterozygosity or specific differences among mating partners only affect
reproductive success when inbreeding is more pronounced.
Although male success did not vary with body size, male body size did
affect paternity patterns, and it did so differently in the two marshes. In
Barbs Marsh, the sizes of males and females that produced young were
positively correlated, but the pattern was random in Beaver Marsh. Brown and
Weatherhead (1999b) found that
larger males courted more females in Barbs Marsh, but in Beaver Marsh no size
advantage existed. Thus, in Barbs Marsh (where females are more dispersed),
large males appear to have an advantage in finding and mating with larger
females. Furthermore, because larger females produce more offspring
(Weatherhead et al., 1999),
large males should realize a reproductive advantage from mating with those
females. A factor that might eliminate that advantage is that the ratio of
females to males appears to decline with size
(Brown and Weatherhead, 1999c;
Weatherhead et al., 1995), so
there could be proportionately more males competing for larger females.
Consistent with that interpretation is the observation that multiple paternity
increases with female size in water snakes
(Prosser et al., 2002). In
Beaver Marsh (where females are clumped), male size does not affect success at
finding females (Brown and Weatherhead,
1999b), and paternity is random with respect to male size. Thus,
for apparently different reasons, sexual selection does not favor males of a
particular size in either marsh.
Our central goal was to determine how male reproductive success varied with
male body size and thus to determine whether sexual selection is responsible
for maintaining male water snakes at a substantially smaller size than females
(Brown and Weatherhead, 1999c).
We found no evidence that sexual selection favored smaller males, or indeed
that body size affected male success in any way. Given that sexual selection
does not explain sexual size dimorphism in northern water snakes, what does
explain why males are so much smaller than females? Based on our previous work
with this population, we propose the following hypothesis. First, female size
is a trade-off between survival and fecundity. Fecundity and survival both
increase with female size, although survival declines once females become
reproductive (Brown and Weatherhead,
1997,
1999a). Even though females
survive better as they grow, mortality is still high, so delaying maturation
too long to maximize fecundity makes the probability of dying before
reproducing too high. Thus, females mature at the size that maximizes lifetime
fecundity (Brown and Weatherhead,
1999a).
Male survival also increases with size up to the point of sexual maturity
(Brown and Weatherhead, 1999a).
However, because male reproductive success does not increase with size,
selection favors maturation at a smaller size in males than in females.
Furthermore, the lack of reproductive advantage to larger size means that
males do not need to forage extensively to promote growth, which allows them
to adopt a low-energy strategy outside the breeding season
(Brown and Weatherhead, 2000).
One aspect of water snakes' natural history that does not appear to fit this
scenario is that males grow more slowly than females beginning at birth
(Brown and Weatherhead, 1999c).
The optimal strategy for males would seem to be to follow the same growth
trajectory as females up to the size that males mature sexually and then to
adopt the low-energy/low-growth strategy thereafter. The fact that they do not
follow this pattern may indicate that the size at sexual maturation imposes
some constraint on the growth trajectory that can be followed. Comparative
analyses of patterns of growth in snakes will be necessary to assess this
possibility.
We have ruled out an important role for body size in affecting the outcome
of sexual competition among the male northern water snakes we studied.
However, the fact that body size did influence patterns of mating in one of
the marshes suggests the possibility that in different ecological
circumstances body size could be important. Brown and Weatherhead
(1999c) compared the water
snake population that we studied with two other populations and showed that
female growth patterns were quite similar among populations. However, male
growth patterns differed substantially among populations, particularly with
respect to asymptotic sizes. Thus, comparative studies of sexual selection on
male water snakes in different ecological circumstances both within and among
populations may yet reveal situations under which male size is subject to
sexual selection.
|
ACKNOWLEDGEMENTS |
|---|
We thank P. Comm and K. Kissner for help in the field, G. Alderson, F. Mohammed, and L. De Sousa for help in the lab, G. Blouin-Demers, M. Daly, R. King, and J. Quinn for comments on earlier versions of the manuscript, Queen's University for use of the Biological Station, and the Natural Sciences and Engineering Research Council of Canada and Parks Canada for providing financial support through grants to H.L.G. and P.J.W.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Andersson M, 1994. Sexual selection. Princeton, New Jersey: Princeton University Press.
Brown GP, Weatherhead PJ, 1997. Effects of reproduction on survival and growth of female northern water snakes, Nerodia sipedon. Can J Zool 75: 424-432.
Brown GP, Weatherhead PJ, 1999a. Demography and sexual size dimorphism in northern water snakes, Nerodia sipedon. Can J Zool 77: 1358-1366.
Brown GP, Weatherhead PJ, 1999b. Female distribution affects mate searching and sexual selection in male northern water snakes (Nerodia sipedon). Behav Ecol Sociobiol 47: 9-16.
Brown GP, Weatherhead PJ, 1999c. Growth and sexual size dimorphism in northern water snakes (Nerodia sipedon). Copeia 1999: 723-732.
Brown GP, Weatherhead PJ, 2000. Thermal ecology and sexual size dimorphism in northern water snakes, Nerodia sipedon. Ecol Monogr 70: 311-330.
Brown JL, 1997. A theory of mate choice based on
heterozygosity. Behav Ecol 8:
60-65.
Charlesworth D, Charlesworth B, 1987. Inbreeding depression and its evolutionary consequences. Annu Rev Ecol Syst 18: 237-268.[ISI]
Darwin C, 1871. The descent of man, and selection in relation to sex, vols. 1 and 2. New York: Appleton.
Duvall D, Schuett GW, Arnold SJ, 1993. Ecology and evolution of snake mating systems. In: Snakes: ecology and behavior (Seigel RA, Collins JT, eds). New York: McGraw-Hill; 165-200.
Gaulin SJC, FitzGerald RW, 1989. Sexual selection for spatial-learning ability. Anim Behav 37: 322-331.
Gibbs HL, Weatherhead PJ, 2001. Insights into
population ecology and sexual selection in snakes through the application of
DNA-based genetic markers. J Hered 92:
173-179.
Goodnight KF, Queller DC, 1999. Computer software for performing likelihood tests of pedigree relationship using genetic markers. Mol Ecol 8: 1231-1234.[Medline]
King RB, 1989. Sexual dimorphism in snake tail length: sexual selection, natural selection, or morphological constraint? Biol J Linn Soc 38: 133-154.
Kingsolver JG, Hoekstra HE, Hoekstra JM, Berrigan D, Vignieri SN, Hill CE, Hoang A, Gibert P, Beerli P, 2001. The strength of phenotypic selection in natural populations. Am Nat 157: 245-261.[ISI]
Lande R, Arnold SJ, 1983. The measurement of selection on correlated characters. Evolution 37: 1210-1226.[ISI]
Leger DW, Didrichsons IA, 1994. An assessment of data pooling and some alternatives. Anim Behav 48: 823-832.
Madsen T, Shine R, 1993a. Male mating success and body size in European grass snakes. Copeia 1993: 561-564.
Madsen T, Shine R, 1993b. Temporal variability in sexual selection acting on reproductive tactics and body size in male snakes. Am Nat 141: 167-171.
Madsen T, Shine R, Loman J, Hakansson T, 1993. Determinants of mating success in male adders, Vipera berus. Anim Behav 45: 491-499.
Madsen T, Shine R, Olsson M, Wittzell H, 1999. Restoration of an inbred adder population. Nature 402: 34-35.[ISI]
Olsson M, Madsen T, 1998. Sexual selection and sperm competition in reptiles. In: Sperm competition and sexual selection (Birkhead TR, Møller AP, eds). San Diego, California: Academic Press; 503-577.
Olsson M, Shine R, Madsen T, Gullberg A, Tegelström H, 1996. Sperm selection by females. Nature 383: 585.
Partridge L, 1983. Non-random mating and offspring fitness. In: Mate choice (Bateson P, ed). Cambridge: Cambridge University Press; 227-255.
Pemberton JM, Coltman DW, Coulson TN, Slate J, 1999. Using microsatellites to measure the fitness consequences of inbreeding and outbreeding. In: Microsatellites: evolution and applications (Goldstein DB, Schlotterer C, eds). Oxford: Oxford University Press; 151-164.
Prosser MR, 1999. Sexual selection in northern water snakes, Nerodia sipedon sipedon (PhD thesis). Hamilton, Ontario: McMaster University.
Prosser MR, Gibbs HL, Weatherhead PJ, 1999. Microgeographic population structure in the northern water snake, Nerodia sipedon, detected using microsatellite DNA loci. Mol Ecol 8: 329-333.[Medline]
Prosser MR, Weatherhead PJ, Gibbs HL, Brown GP, 2002.
Genetic analysis of the mating system and opportunity for sexual selection in
northern water snakes (Nerodia sipedon). Behav Ecol
13: 800-807.
Queller DC, Goodnight KF, 1989. Estimating relatedness using genetic markers. Evolution 43: 258-275.[ISI]
Raymond M, Rousset F, 1995. GENEPOP (version 1.2): A
population genetics software for exact tests and ecumenicism. J
Hered 86:
248-249.
Schwagmeyer PL, 1988. Scramble competition polygyny in an asocial mammal: male mobility and mating success. Am Nat 131: 885-892.
Shine R, 1978. Sexual size dimorphism and male combat in snakes. Oecologia 33: 269-277.
Shine R, 1993. Sexual dimorphism in snakes. In: Snakes: ecology and behavior (Seigel RA, Collins JT, eds). New York: McGraw-Hill; 49-86.
Shine R, O'Connor D, LeMaster MP, Mason RT, 2001. Pick on someone your own size: ontogenetic shifts in mate choice by male garter snakes result in size-assortative mating. Anim Behav 61: 1133-1141.
Shine R, Olsson MM, Moore IT, LeMaster MP, Greene M, Mason RT, 2000. Body size enhances mating success in male garter snakes. Anim Behav 59: F4-F11.[Medline]
Shine R, Olsson MM, Moore IT, LeMaster MP, Mason RT, 1999. Why do male snakes have longer tails than females? Proc R Soc Lond B 266: 2147-2151.
Valdes AM, Slatkin M, Friemer NB, 1993. Allele frequencies at microsatellite loci: the stepwise mutation model. Genetics 133: 737-749.[Abstract]
Weatherhead PJ, Barry FE, Brown GP, Forbes MRL, 1995. Sex ratios, mating behavior and sexual size dimorphism of the northern water snake, Nerodia sipedon. Behav Ecol Sociobiol 36: 301-311.
Weatherhead PJ, Brown GP, 1996. Measurement versus estimation of condition in snakes. Can J Zool 74: 1617-1621.
Weatherhead PJ, Brown GP, Prosser ML, Kissner KJ, 1999. Factors affecting neonate size variation in northern water snakes, Nerodia sipedon. J Herpetol 33: 577-589.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  B. D. Neff Mean d2 and Divergence Time: Transformations and Standardizations J. Hered., March 1, 2004; 95(2): 165 - 171. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||