Mate sampling in a population of sand gobies

Anim. Behav., 1997, 53, 267–276

Mate sampling in a population of sand gobies ELISABET FORSGREN Department of Zoology, Uppsala University, Sweden (Received 4 December 1995; initial acceptance 19 February 1996; final acceptance 13 April 1996; MS. number: 5100)

Abstract. Female mate-sampling behaviour and mate choice were investigated in a natural population of sand gobies, Pomatoschistus minutus. The sand goby is a promiscuous fish with paternal care. Gravid females were released in the field and watched while they inspected potential mates. In total, 26 females were observed until they spawned. Half of the females spawned with the first male encountered. The other females inspected two or more males in sequence and never returned to a previously inspected male to mate. Females mated preferentially with males that courted intensively. The mate-sampling behaviour of the female sand gobies was most consistent with a threshold-criterion tactic, which is to be expected if search costs are important. Moreover, there was an effect of season, such that females sampled very few males early but apparently became more selective later in the breeding season. This is discussed and interpreted as an effect of varying costs and benefits of being choosy over the season. ?

Many studies of mate choice have tried to answer questions such as if, and why, mate choice exists, and what traits are being chosen. Fewer studies have been devoted to information gathering and decision making of individuals during mate choice, that is, how mates are selected. The process of mate sampling is still an underinvestigated area in sexual selection. Although the subject has received theoretical attention (Janetos 1980; Wittenberger 1983; Real 1990), only a few empirical studies have tried to demonstrate how animals search for and select mates in the wild. Most of our current knowledge derives from studies of birds (e.g. Dale et al. 1990; Petrie et al. 1991; Bensch & Hasselquist 1992; Choudhury & Black 1993), although there are a few studies of fish (Brown 1981; Gronell 1989; Reynolds & Côté 1995). Theoreticians have identified five main tactics that animals may follow when choosing mates (Janetos 1980; Wittenberger 1983; Real 1990): (1) random mating; (2) threshold-criterion tactic; (3) sequential-comparison tactic; (4) one-stepdecision tactic; and (5) pool-comparison tactic (best-of-N-males). In random mating, females have no mate preferences and mate with all males with equal probability. In the threshold-criterion Correspondence and present address: E. Forsgren, Kristineberg Marine Research Station, S-450 34 Fiskeba¨ckskil, Sweden. 0003–3472/97/020267+10 $25.00/0/ar960374

?

1997 The Association for the Study of Animal Behaviour

tactic, females inspect males in sequence and choose the first to meet some minimum specification, such that males are judged against an absolute standard. This leads to females always mating with the last male encountered. A variant of this tactic is when females have an option to lower the threshold when they are running out of time. Females using a sequential-comparison tactic inspect potential mates sequentially and compare the two most recently encountered males according to some rule. Either one of them is accepted, or the search is continued. For instance, a female might continue as long as male i is better than male i"1. In the one-step-decision tactic, the female decides at each encounter whether to accept or reject the potential mate. She should end her search and mate if the male at hand is better than she could expect to find by continuing her search. Time constraints can alter the choosiness of the female. At each encounter the female should take into account the probability of encountering a better male and the costs of delaying mating. Females may thus begin by being very choosy, but become less so as they run out of time. This tactic also leads to females mating with the last encountered male. In the pool-comparison tactic, females assess a number of potential mates and then choose the best one among these. Theoretical models make different predictions about how choosiness should change in the course 1997 The Association for the Study of Animal Behaviour

267

Animal Behaviour, 53, 2

268

of the breeding season. One model predicts that choosiness should decrease as the search progresses, given a finite time horizon (Real 1990). This is often thought to be the case in birds, which may be time-stressed by a short breeding season over which breeding prospects decline (e.g. Hochachka 1990). However, other models predict the opposite (Crowley et al. 1991); in animals that have only a single breeding season during which they reproduce repeatedly, choosiness is expected to increase late in the season. This might be the case if mate search were costly in terms of increased predation: it would be less costly to be choosy late in the season because individuals would risk losing a smaller proportion of their lifetime reproductive success. Empirical data on seasonal changes in choosiness are scarce, however (but see Hovi & Rätti 1994). The present study is the first to investigate mate-sampling tactics of a fish in the field. I observed female mate-sampling behaviour and mate choice in the sand goby, Pomatoschistus minutus (Pisces, Gobiidae). In this species, males build nests to which they attract females for spawning. I released gravid females in their natural habitat and watched them by snorkling until they spawned. My aim was to answer the following questions. (1) How long do females take to choose a mate, how many males are inspected, and how far do females travel while inspecting potential mates? (2) What characterizes males that are chosen? (3) Do females change their behaviour over the season? (4) Can the behaviour of female sand gobies be attributed to any of the matesampling tactics discussed above?

METHODS The sand goby is a marine fish common along the coasts of Europe (Miller 1986). Sand gobies live for 1–2 years and reproduce during one season only (Healey 1971). The breeding season lasts approximately from late April to late June in my study area, and individuals of both sexes can reproduce repeatedly. The mating system is promiscuous: both males and females spawn sequentially with different mates (Healey 1971). Males build nests under empty mussel shells by covering them with sand and digging underneath. They attract females by a courtship display, which involves lying with all fins erect, and/or swimming

rapidly, close to the female, followed by attempts to lead her into the nest. During spawning, the female attaches her eggs to the ceiling of the nest in a single layer. Afterwards she leaves the nest, and the male stays to guard and fan the eggs until hatching. Males can care for clutches from several females at the same time (Hesthagen 1977). Females are non-territorial while nest-holding males defend an area close to the nest. The sexes are equal in size, but males develop a breeding coloration with bluish-black fins. Females are known to prefer large and colourful males (Forsgren 1992), while males seem unselective in their mate choice (E. Forsgren & C. Kvarnemo, unpublished data). I conducted the study during May–June 1992 and 1993 at Klubban Biological Station, Fiskebäckskil (58)15*N, 11)28*E), on the west coast of Sweden. I caught sand gobies with a hand trawl near the field site (see below). Females were taken to the laboratory and kept in 120–130-litre storage aquaria with circulating sea water and a natural light regime set by a timer. They were fed chopped blue mussels, Mytilus edulis, once a day. Females were kept in the laboratory for a maximum of 1 week prior to release (a short time compared to our standard routines, which do not seem to affect the fishes negatively). The field site, Bökevik, is a shallow (approximately 0.4–1.0 m) bay of the Gullmarfjord. The bottom is predominantly covered with fine sand, but also with live blue mussel banks and dead mussel shells of various species. The vegetation is mainly eelgrass, Zostera marina, and Fucus spp. I measured water temperature at the surface with a thermometer. To facilitate orientation I put out a grid system, 30#30 m with a marker every 5 m, in the area. I picked highly gravid sand goby females from the storage aquaria and transported them (5 min by boat) to the field site. After measuring body length and estimating roundness (on a scale from 1–3) I released them, one at a time, and followed them by snorkling. Typically, females burrowed in the sand immediately after release. If a female stayed burrowed for more than 1 h, which happened only rarely, I stopped the observation and omitted her from the data set. When the focal female started swimming, I gently glided after her at a distance of 0.5–2 m, watching her movements and taking notes. I estimated body length, coloration (on a scale from 1 to 3, from pale to dark)

Forsgren: Mate sampling in gobies and presence/absence of parasites (Cryptocotula spp.) of encountered males (N=63). This parasite can be detected as white spots on the skin of the fish. I also estimated courtship intensity on a scale from 0 to 4 (0, no courtship; 1, display posture only; 2, slow courtship; 3, normal courtship; 4, very intense courtship). Classes 2–4 involve the male both lying in the display posture with erect fins and courtship swimming. In a few cases courtship was scored to 1/2 points on the scale. I estimated the male characters before knowing the outcome of the female’s visit. Observations ended when the female spawned. However, I stayed to observe the nest with the spawning couple for at least 20 min to make sure that the female did not swim out immediately (without spawning). Normally, spawning can be detected by watching a nest from outside. The head of the male can be seen in the entrance of the nest, panting, shivering and turning blueish. Occasionally he returns into the nest to fertilize the eggs, and often the moving tails of both fishes can be seen. Spawning takes about 1–3 h (personal observation). I marked all visible nests of encountered males, both rejected and spawning, with a coloured plastic ribbon tied to a lead weight. I returned, usually within 24 h, to catch these males with a hand net. I measured their total body length (N=22) to the nearest mm. I also determined the species of mussel used as a nest and measured nest size (width#length). No male was assessed by more than one of the females while occupying the same nest. But since I did not have a marked population, I cannot exclude the possibility that the same individual male may have occurred more than once in the data set. However, given the high density of breeding sand gobies in the area, I believe this risk to be extremely small. In total, 26 females were followed until spawning. Cases where I lost sight of the female, or had to stop the observation before she spawned (not all released females spawned within a reasonable time) are not included in the data set. The proportion of terminated observations was higher in the later half of the breeding season, as a consequence of females taking longer to spawn and inspecting more males then (see Results). I also observed mate-searching females which I encountered in the field. A couple of these were followed until spawning. Released females did not appear to behave differently from females encountered in the field. The data set is based on about 30 h of observations, although the

269

total observation time (including terminated observations and females encountered in the field) was about 70 h, encompassing around 60 females. All test probabilities are two-tailed.

RESULTS The focal females, as well as all other sand gobies encountered, did not seem to be disturbed by my presence; in fact, most of them ignored me totally. They showed apparently normal behaviour, identical to that observed in aquaria for habituated fish. The swimming direction seemed random. Average female body length was 55.6 (range 42–68, N=26) mm. Over the season, there were no significant changes in either body length (Spearman rank correlation: rS =0.24, N=26, P=0.24) or female roundness (rS = "0.36, N=26, P=0.09). The trend towards a seasonal decrease in roundness reflects the situation in the field. Days were counted from the start of the study to the end. The first observation day in both years, 10 May, was hence number 1. Females behaved similarly in both years as there was no effect of year on number of males inspected (two-way ANOVA: F1,22 =0.93, P=0.35), however, there was an effect of season (early/late; F1,22 =7.74, P=0.01, see below for further details about the seasonal effect). There was no interaction between effects of year and season on number of males inspected (F1,22 =0.83, P=0.37). Data for both years were therefore pooled. Water temperature increased over the season from 8 to 20)C in 1992, and from 11 to 17)C in 1993. Water temperature was not significantly correlated with number of males sampled (rS =0.29, N=26, P=0.15). Mate-sampling Tactic Although sand gobies are cryptic bottomdwelling fish, mate-searching females usually swam 20–50 cm above the bottom in a quite conspicuous fashion, with short visits to the bottom. They often crossed areas of inhospitable substrate, for example live mussel banks or dense eelgrass. They did not forage during mate search. A distinct black pattern appeared around the eyes of females as they became ready to spawn. Typically, males would swim up and approach the focal female to court. She usually followed the courting male down to the bottom until she either

Animal Behaviour, 53, 2

270

Table I. Pair-wise comparison of accepted and rejected males from those females that inspected more than one male (N=12) Male trait Body length (mm) Accepted Rejected Coloration Accepted Rejected Courtship intensity Accepted Rejected

Mean



Range

53.8 50.1

6.6 3.3

42–64 45–55

paired t=1.95 P=0.077

2.06 1.83

0.16 0.24

1.75–2.25 1.25–2.00

z=2.05 P<0.05

3.21 2.11

0.40 0.77

3.0–4.0 0–3.0

z=2.94 P<0.005

Each female contributed one observation of an accepted and a rejected male. In cases where the female rejected several males the average of the different traits of these males was used. Coloration and courtship differences were tested with the Wilcoxon signedranks test. Coloration was scored on a scale from 1 (pale) to 3 (dark) and courtship was scored from 0 to 4.

left or followed the male into his nest. The median assessment time was 45 s (range 2–1800, =238, N=63). In the majority of cases females rejected males without entering their nests. In only one case did a female inspect the nest of a male that was later rejected. Females sampled on average 2.5 males (range 1–13, =2.6, N=26) prior to spawning. However, 54% of them mated with the first male encountered. Female roundness was not correlated with the number of males sampled (rS = "0.20, N=26, P=0.33). Mean search time was 26 min (range 3–71, =19.6, N=26), during which the females swam a distance of 34 m (range 1–100, =28.4, N=26). Females always mated with the last male encountered and did not return to a previously inspected male. The only exception was a female spawning in a situation with two courting males whose nests were less than 0.5 m apart. This female assessed both males for a while before making her choice. Male–male interactions, however, did not seem important to females. In only two cases out of 26, did the focal female observe male interactions involving the male with which she later spawned. On one occasion the focal female was clearly rejected and even attacked by several males before finally spawning. This female was omitted from the data set since she apparently could not choose freely between the males. In two cases, the female laid part of the clutch with one male, and, after having been disturbed, moved on to another male where she continued egg laying. In one case this was due to an aggres-

sive crab, Carcinus maenas, which attacked the male while spawning was going on. In the other case, I accidentally dropped a lead weight very close to the nest while the pair was spawning. In these cases I included only the first spawnings.

Male Characteristics My estimates of male body length were very close to their actual length, as revealed by males subsequently caught and measured (linear regression: slope=0.97, r=0.96, N=18, P=0.0001), and the estimation error was small (range 0–4 mm, mean=1.44 mm). Parasite frequency was very low: only five males were visibly parasitized, one of the 26 accepted males, and four of the 37 rejected ones (÷21 =0.28, P=0.59). Most males encountered seemed interested in the gravid focal females. Only two males, one early and one late in the season, did not court at all. A pair-wise comparison between accepted and rejected males for those females that inspected more than one male revealed significant differences in male courtship intensity and coloration, and a tendency for accepted males to be larger than those that were rejected (Table I). The results are not affected if the two non-courting males are excluded from the analysis. However, there is a problem with these traits being intercorrelated. There were positive correlations between both body length and coloration (rS =0.50, N=63,

Forsgren: Mate sampling in gobies Table II. Logistic regression analysis with male mating success (accepted/rejected) as dependent variable and male body length, coloration, courtship intensity and female as independent variables

N=2

"2 Log LR

"8.64 "7.23 "17.62 "21.71

2.84 0.012 20.79 28.98

15

6

29

2

4

0.9 0.8

Significance df of Log LR 1 1 25 1

0.09 0.91 0.70 <0.0001

Probability of mating

Length Colour Female Courtship

Log likelihood

1

1.0

Model if variable removed Variable removed

4

271

The current model does not differ from a perfect model ("2 log likelihood ÷234 =14.4, P=0.99) and the variable coefficient(s) differ from zero (improvement ÷25 =22.13, P=0.0005). The table shows statistics after the first step, when all variables (and the constant) have been entered into the model.

0.6 0.5 0.4 0.3 0.2 0.1 0

1

2 3 Courtship intensity

4

5

Figure 1. Logistic regression of male mating success (mated/rejected) as a function of courtship intensity (logit: Z= "32.68+11.16X; =37.7). All variables except courtship intensity and the constant were removed from the model. The current model correctly classifies 85.7% of the observations.

Courtship intensity

P<0.001) as well as between coloration and courtship intensity (rS =0.30, N=63, P<0.05). Therefore, I carried out a logistic regression analysis. Backward step-wise selection was used to remove variables from the model (Norusis 1990). To control for a possible effect of nonindependent observations since half of the females assessed more than one male, female identity was entered into the model as a categorical variable, and was found to have no effect. This analysis, including all males, showed that the male character that seemed most important in female mate choice was courtship intensity. Male mating success was strongly associated with courtship intensity (Table II, Fig. 1), whereas male body length and coloration did not significantly add to the model, although body length tended to explain male success. Males with a courtship intensity below 3 were never accepted. Male courtship intensity declined significantly over the season (rS = "0.28, N=63, P<0.05; Fig. 2). Most of the early males courted intensely while there was a larger variation later in the season, leading to a higher average courtship intensity in the early half (2.81) than in the late half (2.45) of the season (Mann–Whitney U-test: Z= "2.11, Nearly =16, Nlate =47, P<0.05). Accepted males, however, did not differ significantly in courtship intensity over the season (early half: mean=3.15, range 3–4, N=13; late half: mean=3.23, range 3–4, N=13; Z= "0.21, P=0.8). Comparing the courtship intensities of rejected males, females

0.7

4 3 2 1

0

5

10 15 20 25 30 35 40 45 50 Day

Figure 2. Male courtship intensity in relation to day of the breeding season. The size of the circles reflects sample sizes. Day 1=10 May.

may have been more selective later in the season, although the statistical test has little power since the sample size is small early in the season because few males were rejected (early half: mean=1.33, range 0–2, N=3; late half: mean=2.19, range 0–3, N=34; Z= "1.60, P=0.11). Nest size did not differ between mated and unmated males (t26 =0.65, P>0.5), nor did nest type (the white, hard Mya and Cardium were pooled into the same category and compared with

Animal Behaviour, 53, 2

the softer-shelled, darker Mytilus, Fisher’s exact, N=31, P=0.28). On several occasions sneaky spawning attempts were observed. Other males hid or discretely stayed close to the courting pair watching them just prior to spawning. When the pair entered the nest, the other male, in three cases, managed to follow quickly into the nest and stay inside for about 1 min (range 0.5–1.5 min) until chased away. Hence, potential sneaky fertilizations occurred in about 12% of the spawnings. Sneaking has also been reported in the laboratory in the closely related common goby, P. microps (Magnhagen 1992).

14 No. of males inspected

272

12

(a)

10 8 6 4 2 0

5

10 15 20 25 30 35 40 45 50

80 70

There was a significant increase in the number of males sampled (rS =0.59, N=26, P<0.01), search time (rS =0.45, N=26, P<0.05) and search distance (rS =0.52, N=26, P<0.01) over the season (Fig. 3). There was no clear difference in male density over the season as the distance between males that females visited early (mean inter-male distance&=22.3 &26.7 m) and late in the season (16.3&15.2 m) did not differ significantly (t24 =0.7, P=0.49).

60

Search time (min)

Seasonal Effect

(b)

50 40 30 20 10 0

5

10 15 20 25 30 35 40 45 50

120

DISCUSSION

(c)

Sampling Tactic My results show that female sand gobies actively sample and choose between males. Half of the females, however, mated with the first male encountered. The behaviour of these females, predominantly observed early in the season, is consistent with both random mating and a threshold-criterion tactic. The other females inspected two or more males in sequence before making their decision. They never returned to a previously inspected male, so their behaviour can best be described as following a thresholdcriterion tactic or the one-step-decision tactic. Throughout the season, only males that courted at levels scored as normal-intense, or above, were chosen by the females, which is not in accordance with random mating. This result, together with the observation that so many females mated with the first male encountered, seems most congruent with a threshold-criterion tactic (Janetos 1980),

Search distance (m)

100 80 60 40 20 0

5

10 15 20 25 30 35 40 45 50 Day

Figure 3. The number of males inspected (a), search time (b) and search distance (c) of mate-sampling females in relation to day of the breeding season. Day 1=10 May.

that is, males are sampled in sequence until a male above some threshold quality is encountered (which in some cases may be the first male) and mating takes place. This is to be expected if search costs are important, since a sequential search

Forsgren: Mate sampling in gobies tactic generates higher expected fitness gains than a pool-comparison tactic, if search costs are included in the model (Real 1990). Models excluding search costs, on the other hand, predict that animals should use a best-of-N tactic, that is, a pool comparison (Janetos 1980). Hence, in sand gobies search costs seem to be important. This implies that the females must gain some compensatory benefit outweighing the cost of showing discriminatory behaviour (Pomiankowski 1987). Potential search costs for female sand gobies include increased energy and time expenditure, predation and loss of mating opportunities while sampling and assessing additional males. Predation, for example, may be a severe cost to mate-searching females as they then swim, sometimes long distances, high above the bottom and away from shelter. In addition, their black eye markings and extended belly probably add to their conspicuousness. In lekking birds, where search costs are thought to be small (Gibson & Bachman 1992) a best-of-N tactic has been found in several species (cock-of-the-rock, Rupicola rupicola, Trail & Adams 1989; peacock, Pavo cristatus, Petrie et al. 1991; great snipe, Gallinago media, Fiske & Kålås 1994; black grouse, Tetrao tetrix, Rintamäki et al. 1995). I cannot exclude the possibility that female sand gobies, under natural circumstances, also assess and memorize males before they are ready to spawn. However, females who are not gravid seem uninterested in males (personal observations in aquaria and field). Furthermore, because of the temporary nature of the population, for example, nests are built and abandoned all the time and a male’s mating status may change quickly, such information should not be very useful. There may have been a potential problem with the gobies being disturbed by me, perceiving me as a potential predator. However, the typical antipredator response of sand gobies is to burrow in the sand, and neither females nor males burrowed as I followed/encountered them. Neither did they swim away from me. Also, most males immediately approached the focal female to court upon seeing her, whereas males threatened by a predator are known to reduce their activity drastically during the courtship phase (Forsgren & Magnhagen 1993). Since the gobies behaved calmly, it seems unlikely that they perceived me as a threat. In a previous experimental study, female sand gobies became unselective in the presence of

273

a fish predator (Forsgren 1992). In the present study, many of the females at the beginning of the season mated with the first encountered male. However, if this was random mating caused by my presence, all females should have behaved similarly, which was not the case. Rather, I suggest that the females were following a thresholdcriterion tactic, since no males with poor courtship were ever accepted. Possible explanations for the observed seasonal effect in the mate-sampling behaviour are discussed below. Male Characteristics Most females rejected males without inspecting their nest. This suggests that characteristics of the nest are of minor importance and that females base their mating decisions mainly on charcteristics of the male himself. Male courtship intensity was the most important cue to females choosing a mate. However, this does not exclude the possibility that females may also base their choice on other, unmeasured cues. It is possible that the intensity of a male’s courtship reflects his ability and willingness to care for the eggs, as have been shown in the bicolour damselfish, Stegastes partitus (Knapp & Kovach 1991). Thus, by selecting males with intense courtship females might gain direct fitness benefits. Male sand gobies vary considerably in parental quality as assessed by the proportion of eggs hatching to fry, and females seem to prefer males of superior parental quality (Forsgren 1995). Seasonal Effect Female sand gobies spent more time sampling males and swam greater distances, inspecting a larger number of potential mates prior to mating, as the breeding season progressed. This is in accordance with Crowley et al.’s (1991) model. There may be several explanations for this increase in choosiness over the season. First, the operational sex ratio might be more skewed towards males later in the breeding season because of the shorter time needed for the eggs to hatch at warmer temperatures, thereby shortening the period when males are occupied by parental duties (Kvarnemo 1994). A male-skewed operational sex ratio can increase the opportunity for females to be selective (Emlen & Oring 1977; Berglund 1993). However, I did not find any

274

Animal Behaviour, 53, 2

evidence that males were more abundant later in the season. Male density remained approximately the same, and females travelled equally long distances between males at the beginning and end of the season. Also, most males courted the focal females and seemed interested in spawning throughout the season. How many of these males actually had nests with room for additional eggs is unknown, however, as I do not have complete data on whether a nest contained eggs at the time when females made their mating decisions. However, since females only exceptionally inspected a nest prior to rejecting a male, attributes of the nests cannot explain why males are rejected. Second, search costs may change over the season and influence the degree of choosiness, as predicted by Real (1990) and Crowley et al. (1991). For example, as in Crowley et al.’s (1991) dynamic model, the cost of being preyed upon should be higher for gobies early in the season since their residual reproductive value is high then. Later in the season when it is lower, being preyed upon is less costly. That predation risk may affect mate choice is supported by earlier experimental results showing that female sand gobies become less choosy in the presence of a fish predator (Forsgren 1992; see also Forsgren & Magnhagen 1993). Whether the energetic costs of searching decrease over the season with increasing temperature is not known, although this seems unlikely as temperature did not affect the number of males sampled. Also, temperature per se does not seem to affect activity level in this species (Kvarnemo 1996). Several studies have shown that increasing search costs lead to a decrease in choosiness; for example, in the pied flycatcher, Ficedula hypoleuca, distance between males (Alatalo et al. 1988), energy expenditure (Slagsvold et al. 1988), competition for mates (Dale et al. 1992), and perhaps time stress (Slagsvold et al. 1988; but see Hovi & Rätti 1994) constrain mate choice. In kestrels, Falco tinnunculus, females become unselective in poor vole years, when kestrel density is low and few unmated males are available (Palokangas et al. 1992). Experimentally increased search costs (in terms of time and energy) resulted in reduced choosiness in female sticklebacks, Gasterosteus aculeatus (Milinski & Bakker 1992). In another fish, the upland bullie, Gobiomorphus breviceps, heavily parasitized

females were less selective than less parasitized females (Poulin 1994), apparently because they had depleted energy stores and were in poor physical condition. Although choosiness may apparently disappear with high search costs, females may pay some costs to obtain certain mates, as in redlip blennies, Ophioblennius atlanticus, were females pay higher search costs (measured in terms of attacks by heterospecifics) in order to reach high quality males (Reynolds & Côté 1995). Third, if the mean and/or variance of male quality change over the season, the benefits from being selective will also do so, leading to a change in choosiness (Parker 1983; Real 1990). My results support this idea, as male courtship intensity was more variable later in the season, yielding a lower mean intensity. It thus seems as if males on average are in better condition early in the season. Hence, early in the season, it may not pay females to be selective, because almost all males court vigorously, presumably indicating that they are in good condition and able to care for eggs. Late in the season, on the other hand, it should be more important to be choosy. Another aspect of this seasonal change in courtship (and possibly condition) is that more males would be above the females’ threshold of acceptance early in the season. This would result in females accepting a male more quickly early than late, even without a reduction in their choosiness. Hence, whether female sand gobies actually change their choosiness over time is unclear. Presumably they have a fixed threshold regarding courtship intensity, below which males are always rejected. Indeed, all males showing a weaker courtship than level 3 were rejected. Early in the season, all males courting at this intensity were accepted by the females. However, this was not the case in the later half of the season, when some males were rejected even if courting at this level of intensity. Thus, females might really become more choosy later in the breeding season. Hence, seasonal changes in both search costs and male quality could potentially explain the observed increase in female choosiness over the season. Mate search might be more costly in terms of predation early in the season, and the decrease in male courtship intensity over the season might indicate a greater variance in male quality. In other words, the cost–benefit ratio of being choosy is likely to decrease over the season in this population.

Forsgren: Mate sampling in gobies ACKNOWLEDGMENTS I am very grateful to Johan Dannewitz who with much patience spent long hours in cold waters helping me with observations. I also thank Annika Bokström, Ola Hallberg and Jenny Jewert for help in the field, Ed Connor for statistical advice, and Jacob Höglund, Ben Sheldon, Staffan Ulfstrand and two anonymous referees for valuable comments on the manuscript. The Royal Swedish Academy of Science provided financial support.

REFERENCES Alatalo, R. V., Carlson, A. & Lundberg, A. 1988. The search cost in mate choice of the pied flycatcher. Anim. Behav., 36, 289–291. Bensch, S. & Hasselquist, D. 1992. Evidence for active female choice in a polygynous warbler. Anim. Behav., 44, 301–311. Berglund, A. 1993. The operational sex ratio influences choosiness in a pipefish. Behav. Ecol., 5, 254–257. Brown, L. 1981. Patterns of female choice in mottled sculpins (Cottidae, Teleostei). Anim. Behav., 29, 375– 382. Choudhury, S. & Black, J. M. 1993. Mate-selection behaviour and sampling strategies in geese. Anim. Behav., 46, 747–757. Crowley, P. H., Travers, S. E., Linton, S. E., Cohn, S. L., Sih, A. & Sargent, R. C. 1991. Mate density, predation risk, and the seasonal sequence of mate choices: a dynamic game. Am. Nat., 137, 567–596. Dale, S., Amundsen, T., Lifjeld, J. T. & Slagsvold, T. 1990. Mate sampling behaviour of female pied flycatchers: evidence for active mate choice. Behav. Ecol. Sociobiol., 27, 87–91. Dale, S., Rinden, H. & Slagsvold, T. 1992. Competition for a mate restricts mate search of female pied flycatchers. Behav. Ecol. Sociobiol., 30, 165–176. Emlen, S. T. & Oring, L. W. 1977. Ecology, sexual selection, and the evolution of mating systems. Science, 197, 215–223. Fiske, P. & Kålås, J. A. 1994. Mate sampling and copulation behaviour of great snipe females. Anim. Behav., 49, 209–219. Forsgren, E. 1992. Predation risk affects mate choice in a gobiid fish. Am. Nat., 140, 1041–1049. Forsgren, E. 1995. Sexual selection in the sand goby, Pomatoschistus minutus. Ph.D. thesis, Uppsala University. Forsgren, E. & Magnhagen, C. 1993. Conflicting demands in sand gobies: predators influence reproductive behaviour. Behaviour, 126, 125–135. Gibson, R. M. & Bachman, G. C. 1992. The costs of female choice in a lekking bird. Behav. Ecol., 3, 300–309. Gronell, A. M. 1989. Visiting behaviour by females of the sexually dichromatic damselfish, Chrysiptera

275

cyanea (Teleostei: Pomacentridae): a probable method of assessing male quality. Ethology, 81, 89–122. Healey, M. C. 1971. Gonad development and fecundity of the sand goby, Gobius minutus Pallas. Trans. Am. Fish. Soc., 3, 520–526. Hesthagen, I. H. 1977. Migrations, breeding and growth in Pomatoschistus minutus (Pallas) (Pisces, Gobiidae) in Oslofjorden, Norway. Sarsia, 63, 17–26. Hochachka, W. 1990. Seasonal decline in reproductive performance of song sparrows. Ecology, 71, 1279– 1288. Hovi, M. & Rätti, O. 1994. Mate sampling and assessment procedures in female flycatchers (Ficedula hypoleuca). Ethology, 96, 127–137. Janetos, A. C. 1980. Strategies of female choice: a theoretical analysis. Behav. Ecol. Sociobiol., 7, 107– 112. Knapp, R. A. & Kovach, J. T. 1991. Courtship as an honest indicator of male parental quality in the bicolor damselfish, Stegastes partitus. Behav. Ecol., 2, 295–300. Kvarnemo, C. 1994. Temperature differentially affects male and female reproductive rates in the sand goby: consequences for operational sex ratio. Proc. R. Soc. Lond. Ser. B, 256, 151–156. Kvarnemo, C. 1996. Sexual selection and the influence of environmental factors on operational sex ratios in the sand goby. Ph.D. thesis, Uppsala University. Magnhagen, C. 1992. Alternative reproductive behaviour in the common goby, Pomatoschistus microps: an ontogenetic gradient? Anim. Behav., 44, 182–184. Milinski, M. & Bakker, T. 1992. Costs influence sequential mate choice in sticklebacks, Gasterosteus aculeatus. Proc. R. Soc. Lond. Ser. B., 250, 229–233. Miller, P. J. 1986. Gobiidae. In: Fishes of the Northeastern Atlantic and the Mediterranean, Vol. 3 (Ed. by P. J. P. Whitehead), pp. 1019–1085. Paris: UNESCO. Norusis, M. J. 1990. SPSS Advanced Statistics User’s Guide. Chicago: SPSS. Palokangas, P., Alatalo, R. V. & Korpimäki, E. 1992. Female choice in kestrels under different availability of mating options. Anim. Behav., 43, 659–665. Parker, G. A. 1983. Mate quality and mating decisions. In: Mate Choice (Ed. by P. Bateson), pp. 141–166. Cambridge: Cambridge University Press. Petrie, M., Halliday, T. & Sanders, C. 1991. Peahens prefer peacocks with elaborate trains. Anim. Behav., 41, 323–331. Pomiankowski, A. 1987. The cost of choice in sexual selection. J. theor. Biol., 128, 195–218. Poulin, R. 1994. Mate choice decisions by parasitized female upland bullies, Gobiomorphus breviceps. Proc. R. Soc. Lond. Ser. B., 256, 183–187. Real, L. 1990. Search theory and mate choice. I. Models of single-sex discrimination. Am. Nat., 136, 376–404. Reynolds, J. D. & Côté, I. M. 1995. Direct selection on mate choice: female blennies pay more for good mates. Behav. Ecol., 6, 175–181. Rintamäki, P. T., Alatalo, R. V. Höglund, J. & Lundberg, A. 1995. Mate sampling behaviour of

276

Animal Behaviour, 53, 2

black grouse females (Tetrao tetrix). Behav. Ecol. Sociobiol., 37, 209–215. Slagsvold, T., Lifjeld, J. T., Stenmark, G. & Breiehagen, T. 1988. On the cost of searching for a mate in female pied flycatchers, Ficedula hypoleuca. Anim. Behav., 36, 433–442.

Trail, P. W. & Adams, E. S. 1989. Active mate choice at cock-of-the-rock leks: tactics of sampling and comparison. Behav. Ecol. Sociobiol., 25, 283–292. Wittenberger, J. F. 1983. Tactics of mate choice. In: Mate Choice (Ed. by P. Bateson), pp. 433–447. Cambridge: Cambridge University Press.