To date, mate choice studies have mostly focused on establishing which mates are chosen or how the choices are performed. Here, we combined these two approaches by empirically testing how latency to mate is affected by various search costs, variation in mate quality and female quality in the sand goby (Pomatoschistus minutus). Our results show that females adjust their mating behaviour according to the costs and benefits of the choice situation. Specifically, they mated sooner when access to males was delayed and when the presence of other females presented a mate sampling cost. We also found a positive link between size variation among potential mating partners and spawning delay in some (but not all) experimental conditions. By contrast, we did not find the number of available males or the females' own body size (‘quality’) to affect mating latency. Finally, female mating behaviour varied significantly between years. These findings are notable for demonstrating that (i) mate sampling time is particularly sensitive to costs and, to a lesser degree, to variation among mate candidates, (ii) females' mating behaviour is sensitive to qualitative rather than to quantitative variation in their environment, and (iii) a snapshot view may describe mate sampling behaviour unreliably.
The high evolutionary potential of sexual selection by mate choice has become well established, with research efforts having mainly focused on determining which mates are chosen and why [1–3]. Recent research also suggests that a female's reproductive decisions are linked to her individual characteristics or the context of the choice situation [1,4–6]. This condition- and context-dependency of mating behaviour has mostly been considered in terms of its effects on female preferences and the characteristics of males that succeed in mating (assuming conventional sex roles). However, ultimately the problem for a female searching for a suitable mate is—within the framework of the choice situation—whether to accept a given male or to continue sampling mate candidates in order to find a mate of a higher quality. Typically, continued mate searching is likely to involve costs, whereas finding the preferred male type could provide fitness returns. Therefore, the time allocated to the sampling process should represent a balance between the benefits obtained by finding a high-quality mate and the costs associated with increased sampling effort [7,8]. Accordingly, here we focus on the factors that affect the time allocated to mate sampling, which is a key part of the mate choice process and also closely linked to the sampling strategy [8–14] the female should use.
Females are expected to react to higher search and sampling costs by lowering their mate acceptance threshold [1,8,15], a prediction that parallels the theory of optimal foraging time . Here, a mate acceptance threshold may refer to either the number of mates sampled or delay before mating when the number of mate candidates is fixed. For example, if female pied flycatchers (Ficedula hypoleuca) face increased costs of competition with other females, then they make more rapid mating decisions . Passing of time, in turn, has been suggested to represent a sampling cost to female three-spined sticklebacks (Gasterosteus aculeatus), causing them to choose males less discriminately . Likewise, in Túngara frogs, females are less likely to travel long distances in search of mates when predation risk is high [19,20].
Regarding the benefits of prolonged mate sampling, there is some prior evidence that time allocated to mate sampling should be sensitive to the distribution of mates and mate quality. Not all theoretical predictions, however, are concordant. On the one hand, sampling a larger number of mate candidates may take a longer time, and the benefits of continued sampling may also be higher when a fixed group of males is more heterogeneous . On the other hand, mate sampling may be faster when differences among males are more evident: females may have a higher probability of finding a satisfying male type among a larger number of males, and a female's uncertainty about mate quality may decrease with increased variation among males [14,15]. Finally, a female's own condition and quality are expected to influence her mating behaviour. In particular, lower-quality females are often assumed to be less choosy and to allocate less time to mate sampling [1,8,21,22], although the generality of this trend depends on the shape of the female preference function [15,23]. There has, however, been a general lack of empirical studies testing for more than one of these predictions at a time.
In this study, we aim at broadening the view of condition- and context-dependent female mating behaviour to include the period leading up to the actual choice itself. This has a direct bearing on the identification of how—and at what point—female phenotype or environmental factors affect female mating decisions. In particular, we studied how the time allocated to the mate sampling process is influenced by different types of search costs, female body size (a proxy of quality), and variation in male number and quality. We were particularly interested in the relative importance of the different factors as determinants of a female's latency to spawn. These goals were achieved by two controlled laboratory experiments using a small sexually dimorphic fish, the sand goby (Pomatoschistus minutus), as the model system. We were also interested in whether female preference for different male types was affected when search costs and male variability were manipulated. This is relevant especially because females of the study population have earlier been found to adjust their mating decisions depending on the presence (or absence) of interactions among males .
2. Material and methods
(a) Model system and study site
The sand goby is a small benthic fish with a single prolonged breeding season , which in our study population lasts from May to July. A reproductively active male establishes a nest under a suitable solid object, such as a flat rock or mussel shell . He then attracts passing females with courtship displays , and several females can spawn in the same nest at different times. The male cares for the eggs until hatching. Female sand gobies are known to sample multiple males before settling on a mate . Female mating decisions are based on multiple male traits (e.g. body size [29,30], nest construction [24,31] and parental skills [32–34]), although the relative importance of these traits seems to be context-dependent [24,35] and to vary both spatially [26,36] and over time .
This study was conducted at the Tvärminne Zoological Station, southern Finland (59°50.7′ N, 23°15.0′ E) during the sand goby breeding seasons (May–July) of 1995, 1996 and 1998 (see below for further details). Fish were collected from a nearby beach using a beach seine. In this area, nesting resources suitable for sand gobies are generally sparse, though with some aggregations of higher densities of suitable rocks and mussel shells [26,38]. Hence, in some areas, a female could potentially observe several breeding males simultaneously, whereas in other sites males would probably be encountered sequentially owing to their low density. In the laboratory, the fish were housed in unisex 250 l aquaria with through-flow of constantly renewed sea water. During a short stocking period, the fish were fed ad libitum with live shrimp (Neomysis), with occasional addition of commercial aquarium fish foods.
(b) Experiment 1: time cost
During the sand goby breeding season of 1996, we tested whether the time cost of mate sampling affected the time until gravid, ready-to-spawn females lay their eggs. The time cost was manipulated by varying the time during which mature females were in the presence of a male without having access to a nest in which to spawn, a manipulation that might also have affected the females' perceived encounter rate with males . To initiate the experiment, two randomly selected males were measured and placed in a tank (bottom area 80 × 40 cm), and were separated by an opaque Plexiglas divider. Both males were allowed to build a nest in a halved flowerpot (diameter 6 cm). After 24 h, two females, one at each end of the tank, were added in transparent Plexiglas cages with several holes in the walls to allow the exchange of water and chemical cues. Each replicate was randomly allocated to one of the three treatment levels representing a mating delay of 4 (n = 11), 8 (n = 9) or 24 h (n = 8). In addition to these, another six female pairs were initially included in the experiment, but later excluded from the data analyses, because the focal female had already released her eggs during the delay period, and was therefore unable to spawn after it.
One of the two females was randomly selected to be the focal female, while ensuring that in half of the replicates this female was familiarized with the smaller, and in the other half with the larger, of the two males. After the delay period, the focal female was released, whereas the opaque divider and the other female in the opposite end of the tank were removed. The focal female now had access to both males and could mate with one of them. By exposing both males to females in the delay phase of the experiment, we wanted to control for any possible effects of female presence on the male's sexual behaviour. The fish were then observed every half hour for the first 4 h and thereafter at least once every 2 h, until the female spawned in the nest of one of the two males. After spawning, the mated male was caught and measured to make sure that no nest switching had taken place.
We also tested whether the size difference between the two males or the focal female's own body size (total length) affected her latency to spawn. Body size was assumed to function as a proxy of female quality, because there is a tight correlation between fecundity and body size in this species . Similarly, we assessed whether female preferences for male size or familiarity were affected by the delay treatment. In particular, the design involved two males for the focal female to choose from, one of which she was familiar with and the other unfamiliar to her, giving us the opportunity to also test whether females establish a preference for a familiar male. Our assumption was that the choice of a more familiar male might involve lower costs than having to sample an unfamiliar one. Furthermore, even relatively short-term social familiarity with potential mates has potential to influence female mating preferences [39–41]. In theory, the presence of two competing males of different sizes in the same tank could prevent females from exercising free choice or sampling behaviour . However, our set-up prevented interactions between males until the fish were free to roam in the tank, a setting that has earlier been used to simulate low potential for male–male competition [24,43]. Furthermore, in sand gobies, females have been found to be in control of the choice situation even under conditions of high potential for competition among males .
We measured the time lapsed before females mated. When females are ready to spawn, the latency to spawn can be expected to relate directly to the time the females are willing to use for sampling suitable mates, and is sometimes also used as a measure of choosiness in general [44,45]. We therefore always tried to choose females that were fully gravid, as evidenced by the roundness of their bellies and dark eye coloration, and males that had demonstrated high breeding motivation by building a nest. In this respect, we do not have any reason to believe that any undetected variation in females' readiness to spawn could cause any systematic biases towards any particular treatment. Finally, a female's fresh weight-to-length3 ratio (Fulton's condition index) is a good predictor of whether a female will be ready to spawn (K.L. 1998, unpublished data), and females used in the three different delay time treatments did not differ in their condition index (ANOVA, F2,25 = 0.189, p = 0.83).
(c) Experiment 2: competition, mate number and mate variability
In this experiment, conducted during the sand goby breeding seasons of 1995 and 1998, we manipulated the presence of same-sex competitors, and the number and size range of available males. For this purpose, we had two treatment categories. First, the focal female was allowed a choice between two, three or six nest-guarding males. Second, these three male availability levels were replicated in the absence or presence of competing females. Hence, in total, the experiment involved 3 × 2 treatments with the following sample sizes: n2,absent = 15, n3,absent = 15, n6,absent = 16, n2,present = 14, n3,present = 15 and n6,present = 14. To maintain an approximately constant male density in each treatment, the two-male treatments were run in the tanks with a bottom area of 40 × 50 cm, the three-male treatments in the tanks of 40 × 80 cm and the six-male treatments in the 80 × 80 cm tanks. Each male was given a halved clay flowerpot (diameter 6 cm) as a nest-building resource. The nests were positioned at equal distances from each other. Males were allowed 24 h for nest building before any females were added. The males were randomly selected from the storage tanks, and their total length and weight were measured before they were placed into the aquarium.
Female competition was manipulated by adding, simultaneously with the focal female, gravid ‘competitor’ females in the tank, enclosed in 10 × 10 × 10 cm transparent Plexiglas cages. Two of the walls of these cages were made of a dense mesh, allowing water exchange between the cage and its surroundings. In the two-male treatment with competition, two females were enclosed in one such cage; in the three-male (competition) treatment two females were in one cage and one female in another; and in the six-male treatment (with competition) two females were enclosed in each of three cages. In the corresponding no-competition treatments, the same number of empty cages was used. The males were generally very interested in these females, and the females responded to courtship by displaying their bellies and increasing the dark coloration around their eyes. Both the focal and competitor females were similarly gravid and ready to spawn. After the focal female had been released, the tank was observed at least once every 30 min for the first 4 h, and thereafter at least once every 2 h until spawning took place, except between 23.00 and 08.00 h. During each check, the behaviour and location of the fish were recorded. If spawning occurred when the tank was not observed, then spawning time was extrapolated as the point of time in the middle of the two observation periods. The male with whom the female spawned with was caught, and his length and weight were measured.
(a) Experiment 1: time cost
To test the effects of presentation delay (treatment group), male size variation (i.e. size difference between the two males) and female size (proxy of its quality), we applied a Cox proportional hazards survival analysis on the latency to spawn. After stepwise deletion of all non-significant interaction terms, the resulting model with main factors indicates significant effects of treatment and male size range, but not female size (table 1): females spawned sooner after longer presentation delays (figure 1) and when the size difference between the two males was small.
Females did not show any preference for male size: they mated with the larger of the two males in 15 out of 28 cases (binomial distribution, p = 0.85), independent of the treatment (G-test of independence, G2 = 1.24, p = 0.54). Similarly, females mated equally often with the ‘familiar’ and more distant male (i.e. they mated with the male with whom they had been presented in 14 out of 28 cases; binomial distribution, p = 1.0), independent of the treatment (G-test of independence, G2 = 1.85, p = 0.40).
(b) Experiment 2: competition, mate number and mate variability
Here, we applied a Cox proportional hazards survival analysis on spawning latency, with the number of males, the absence/presence of competition and year (1995 versus 1998) as categorical variables, and male size variation (the length difference between the largest and the smallest male) as a numerical variable. Again, we simplified our initial model containing first-order interactions by dropping all non-significant interaction terms, with our final model including only the main factors. We found that the number of available males did not affect the focal female's latency to spawn (table 2). However, both competition and year had a significant effect (table 2): females spawned faster when same-sex competitors were present than when there were no competitors (figure 2), and faster in 1998 than 1995. In this experiment, male size differences within the replicates did not have a significant effect (table 2).
To assess whether a female's choice of male was influenced by the treatments, we counted how often the focal female chose a male with higher than average total length in each replicate. This was the case in 16 replicates out of 28 (with one replicate in which an average-sized male was chosen), 18 out of 27 (three average-sized males chosen) and 18 out of 30, in the two-male, three-male and six-male treatments, respectively. This indicates that there is no difference among the treatment types, in terms of mating distribution relative to male size (G-test of independence, G2 = 0.592, p = 0.74). Similarly, the focal female spawned with a larger than average male in 27 replicates out of 45 (average-sized male chosen in two replicates) in the absence of female competitors, and in 27 out of 40 replicates (two average-sized males chosen) when female competitors were present, indicating no effect of competition (G-test of independence, G1 = 0.516, p = 0.47). These patterns also suggest that over the whole experiment, females mated with larger than average males (54 out of 85 replicates, binomial distribution, p = 0.017), whereas this tendency was not significant within any of the above subsamples (binomial distribution, p > 0.05 in all cases).
The results of this study show that female sand gobies adjust their mating behaviour according to costs and benefits of the choice situation. Specifically, they mated sooner with delayed access to males (time cost, experiment 1) and when other females were present in the same tank (competition cost, experiment 2). Size variation among the available males was also related to mating latency in one of the two experiments. In particular, the bigger the size difference among males, the longer it took females to spawn, which fits the ‘conventional’ prediction that choosiness increases with variation in mate quality . Furthermore, if body size (or its correlates) conflicts with other relevant mate choice cues [46,47], such as nest-building or indicators of male parental ability [32,34,48], mate sampling by females could become more challenging when males vary greatly in size. By contrast, we did not find the number of males (between two and six) to affect the time delay before spawning. Similarly, female size was not found to affect mating latency. Hence, female mate sampling seemed to be especially sensitive to the associated costs, and to a lesser extent to variation among males, but not to the number of available mates or female body size (a proxy of her quality).
The implications of our results for understanding patterns of sexual selection are fundamental. For instance, mating decisions may become more random in environments with high sampling costs, with females investing less time in mate sampling. Indeed, under time constraints, there could be selection on females' ability to quickly and adequately assess potential mates, introducing sexual selection also on females. Such conditions of particularly high sampling costs may arise, for example, when females have to wait for access to spawning sites (experiment 1) or when competition for spawning space is intense (experiment 2), as observed in the field under nest site limitation . Our results therefore predict that when mate sampling costs diverge on a local or temporal scale (in sand gobies, see [26,36,37]), sexual selection regimes deviate correspondingly, which is relevant, for example, in the context of divergence between phenotypes or populations . The results also provide insights into the long-standing debate on the female decision rules (strategies) in mate choice, with the two most prominent models having been the ‘fixed sample search strategy’ (sometimes called the ‘best-of-n’) and the ‘fixed threshold strategy’. The latter should, in theory, outperform the former [7,13,14], but empirical studies have struggled to differentiate between the two because their predicted search behaviour patterns may appear superficially similar [10,11,14]. Here, we show that mate sampling time was sensitive to costs and variation among males (under some conditions), but not to the number of available males. These results are most consistent with the ‘threshold strategy’, with females potentially updating their thresholds according to the costs of the choice situation.
To date, empirical studies on mate choice have strongly focused on the questions of which male traits females prefer and why. To also address this topic, we assessed, in our second experiment, whether females choose different males under different experimental conditions. Despite the above-specified various effects of costs on spawning latency, we found no significant effects regarding the relative size of males that the females chose. This result may imply that how females sample mates is more sensitive to external conditions (especially costs) than which males they prefer. It is important to note, however, that we assessed only mating distribution with regard to male size, whereas previous studies have shown that female preference towards other male traits might be even more important . We nevertheless assert that male body size was a relevant male trait to assess here, because a previous study found that females paid less attention to male size in the absence of male–male competition than when male interactions were allowed . In general, female mating behaviour seems to be sensitive to qualitative rather than to quantitative differences in their environment (e.g. absence versus presence of male–male competition). Indeed, previous investigations also suggest that female sand gobies mate quicker when presented with two versus only one male , whereas here we did not find a difference between the treatments of two, three and six males to choose from.
The study was run over several years (1995, 1996 and 1998). This should be seen to strengthen the results, because the targets of female choice can vary in importance among years [6,37], and results obtained over multiple years can therefore be seen to be more general. Indeed, we also show here that mating latency of females varies between years, in line with the results of Lehtonen et al. , which suggested that female responsiveness can vary from year to year. Taken together, the previous and current results suggest that a snapshot view of female mating behaviour may not give a reliable account of the sexual selection regime operating in a population.
We also tested the prediction that the focal female would be more likely to mate with a male she is already familiar with, in order to lower the time cost of mating sampling. Indeed, familiarity has been found to affect mating behaviour in some other systems [39,40]. However, sand goby females did not spawn any more often, nor sooner, with males they had been pre-exposed to. This could be interpreted to suggest that mating time latency of females was not solely due to the time needed to evaluate males, or that mere evaluation of another male was not a major constraint to females, at least under laboratory conditions.
In conclusion, by focusing on the time used for mate sampling, while simultaneously considering several factors that are expected to influence mate choice behaviour, we found support for the prediction that females have the capacity of adjusting their mating behaviour in relation to costs of mate sampling. In particular, females were sensitive to delays in access to mates and presence of same-sex competitors. By contrast, the costs of evaluating another mate candidate, variation among male candidates and female size (‘quality’) were found to be less important. The results also support the view that female mating behaviour may significantly vary from year to year. Future studies may therefore want to assess temporal differences in female mate sampling in a more detailed manner.
T.K.L. acknowledges the financial support by the Department of Biology at the University of Turku.
We thank Tvärminne Zoological Station for working facilities, and Karine Gagnon, Bob Wong and two reviewers for helpful comments.
- Received April 17, 2013.
- Accepted June 7, 2013.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.