Mechanisms of Sexual Selection in the , minutus

Nuutti Kangas

Department of Ecology and Systematics Division of Population Biology University of Helsinki Finland

Academic dissertation

To be presented, with permission of the Faculty of Science of the University of Helsinki, for public criticism in the lecture room of the Department of Ecology and Systematics, P. Rautatiekatu 13, on December 16, 2000, at 12 o’clock noon.

Helsinki 2000 © Nuutti Kangas (Summary and II, IV, V) © Oxford university press (I) © Academic press (III)

Technical editing by Johan Ulfvens

Author’s address: Department of Ecology and systematics Division of Population Biology P.O.Box 17 (Arkadiankatu 7) 00014 University of Helksinki Finland

ISBN 952-91-2870-3 (printed) ISBN 952-91-2871-1 (pdf)

Oy Edita Ab Helsinki 2000 Mechanisms of Sexual Selection in the Sand Goby, Pomatoschistus minutus

Nuutti Kangas

Department of Ecology and Systematics Division of Population Biology P.O.Box 17 (Arkadiankatu 7) 00014 University of helsinki Finland

The thesis is based on the following articles:

I Lindström, K. and Kangas, N. 1996. Egg presence, egg loss, and female mate preferences in the sand goby (Pomatoschistus minutus). – Behavioral Ecology 7, 213–217.

II Kangas, N. and Haldin, C. The effects of egg age on mating success in the sand goby, Pomatoschistus minutus. – Submitted manuscript.

III Kangas, N. and Lindström, K. Male interactions and female mate choice in the sand goby, Pomatoschistus minutus. – In press ( Behaviour).

IV Kangas, N. Sexual selection and male colour in the sand goby, Pomatoschistus minutus. – Submitted manuscript.

V Kangas, N. and Viitasalo, M. affects nest site colonisation, courtship and mate choice in the sand goby, Pomatoschistus minutus. – Submitted manuscript.

These are referred to by their numerals in the text. 4 Contributions

Contributions

The following table shows the major contributions of authors to original articles

I II III IV V

Initiative KL NK NK NK NK Experimental planning NK & KL NK NK & KL NK NK & MV Data gathering NK NK & CH NK NK NK Analysis NK & KL NK NK & KL NK NK Manuscript preaparation NK & KL NK NK & KL NK NK

Supervised by Dr. Kai Lindström, University of Helsinki Finland

Reviewed by Prof. Liselotte Sundström University of Helsinki Finland

Dr. Johanna Mappes University of Jyväskylä Finland

Examined by Prof. Arja Kaitala University of Oulu Finland Contents 5

Contents

0. Summary ...... 7 Introduction ...... 8 Sexual selection ...... 8 The focus of the thesis...... 9 Natural history ...... 10 General methods ...... 12 Main results ...... 13 Study I ...... 13 Study II ...... 13 Study III ...... 13 Study IV ...... 14 Study V ...... 15 Discussion ...... 16 1. Factors affecting mate choice and male-male competition...... 16 2. Habitat constraints on sexual selection...... 20 3. Conclusions ...... 21 References ...... 22

I. Egg presence, egg loss and female mate preferences in the sand goby (Pomatoschistus minutus)...... 29 Materials and methods ...... 30 Egg choice experiments...... 30 Effect of nest success ...... 30 Results ...... 30 Discussion ...... 31 References ...... 33

II. The effect of egg age on mating success in the sand goby, Pomatoschistus minutus...... 37 Material and methods ...... 38 Natural history ...... 38 Egg age and mating success in the wild ...... 39 Standardation of the egg mass area ...... 39 Male condition and egg age in the wild...... 39 The effect of egg age on female choice ...... 40 Results ...... 42 Egg age and mating success in the wild ...... 42 Standardation of the egg mass area ...... 42 Male condition and egg age in the wild...... 42 The effect of egg age on female choice ...... 42 Discussion ...... 43 Literature ...... 45 6 Contents

III. Male interactions and female mate choice in the sand goby, Pomatoschistus minutus...... 49 Material and methods ...... 50 Phase 1 ...... 51 Phase 2 ...... 52 Results ...... 53 Female preference ...... 53 Male-male interaction...... 53 Courtship ...... 54 Discussion ...... 55 References ...... 57

IV. Sexual selection and male colour in the sand goby, Pomatoschistus minutus . . . 61 Material and methods ...... 62 Natural history ...... 62 experiment...... 62 Experiment with non-manipulated males – males interacting ...... 62 Mate choice experiment – interactions prevented ...... 64 Experiment with manipulated males – males interacting ...... 64 Ornament size and mating success in the field ...... 65 Results ...... 65 Non-manipulated males interacting and mate choice ...... 65 Mate choice when interactions prevented...... 66 Manipulated males interacting and mate choice ...... 66 Ornament size and mating success in the wild ...... 67 Discussion ...... 67 Literature ...... 69

V. Predation affects nest site colonisation, courtship and mate choice in the sand goby, Pomatoschistus minutus ...... 73 Materials and methods ...... 74 Natural history ...... 74 Bird predation and nest colonisation in the field ...... 74 Field experiment on predation ...... 75 Nest colonisation under predation risk ...... 76 Predation avoidance with and without female ...... 77 Mate choice experiment ...... 77 Results ...... 77 Bird predation and nest colonisation in the field ...... 77 Field experiment on fish predation ...... 78 Nest colonisation under predation risk ...... 79 Predation avoidance with and without female ...... 79 Mate choice experiment ...... 80 Discussion ...... 80 Literature ...... 83 7

Mechanisms of Sexual Selection in the Sand Goby, Pomatoschistus minutus

Abstract This thesis describes five basic mechanisms of sexual selection in the sand goby, Pomatoschistus minutus, a small fish with paternal care. These are: (1) the female preference for eggs in the nest, (2)the effect of brood age on mating suc- cess, (3) the effect of male–male interactions on mate choice, (4) sexual selec- tion of male colour and (5) the effect of predation on sexual selection. All the studies test in the laboratory and in the field, factors that may give direct benefits to females in terms of offspring survival. The studies show that females avoid nests were potential of egg survival is poor. Also, females with strong males, which might be able to carry the costs of hard male–male competition. Furthermore, in all of the studies the constraints of mate choice are tested. Male interactions are shown to affect mate choice and dominant males may some- times override female choice. Besides this, predation is shown to reduce female activity towards potential mates. Also, in all the studies, factors affecting male– male competition are tested. It is shown that mate choice may affect the result of male–male competition. Parental stage of the male, nest size and predation af- fect the competition for mates among males. Furthermore, it is shown that males avoid high predation risk nest sites during colonisation, increasing com- petition for safe sites. Female choice is not necessarily always operating in the same direction as male–male competition. Either one may outweigh the other, but which one is stronger may be dependent on local ecological conditions. For example, it is shown that brood age variation may affect competition for spawnings and con- sequently female spawning patterns. Furthermore, this spawning pattern may constrain female choice in the wild, when most males are guarding old eggs. Yet, male interactions may outweigh mate choice but not in cases where males differ clearly in ornamentation. I also show that under certain conditions female choice may constrain male–male competition. For example females may choose males that are weaker in physical fights between males. These results suggest that different components of sexual selection operate together resulting in stabilising selection as well as directional selection depending on the envi- ronment. 8 Sexual selection in the sand goby

Introduction (Andersson 1994, Hoelzer 1989, Motro 1982, Bakker 1993, Brooks and Couldridge 1999, This thesis examines behavioural and evolu- Penn and Potts 1999, Wilkinson and al. tionary aspects of a small marine fish , 1998). Indirect benefits could be explained by the sand goby, Pomatoschistus minutus. The the Fisherian model of sexual selection in focus of the study was on mechanisms of sex- which the evolution of the ornament is based ual selection in a Baltic population. on a genetic correlation between trait and preference (Fisher 1958). Indirect benefits could also arise if females mated with geneti- Sexual selection cally superior males in which case their off- spring would inherit the high genetic quality Sexual selection is based on two components, of their fathers (Andersson 1994). Direct ben- competition for mates and mate choice (Dar- efits operate if mate choice increases off- win 1871, Bradbury and Andersson 1987). spring number or survival of the choosy sex These are usually termed intra- and inter-sex- (e.g. Rohwer, 1978, Hoelzer 1989). In ual selection, respectively. Sexual selection with paternal care direct benefits of mate operates within one sex only, i.e., individuals choice could involve qualities such as brood of one sex are competing with one another for defence and egg ventilation (e.g. Rohwer reproductive success (Kodric-Brown 1983). 1978, Colgan and 1986, Bisazza and Sexual selection may operate in the same di- al. 1989, Lindström 1992b, Lindström and rection as other processes of natural selection. Wennström 1994). Traits selected through sexual selection are In most species males are the competing often expected to become more exaggerated sex and females choosy. According to Trivers than would be expected under natural selec- (1972) the choosing sex is the one that has the tion only (Pomiankowski & al. 1991). At greater energy limitations in producing prog- some point, however, natural selection is ex- eny. Later, Clutton-Brock and Parker (1992) pected to limit the further exaggeration of proposed that the competing sex is the sex sexually selected traits (Fisher 1958, see also that has the higher potential reproductive rate, Burley and al. 1998, Holland and Rice 1998). in other words the sex towards which the op- Competition for mates may select for erational sex ratio is biased. qualities important in contests over mates and Mating is not free. In a resource defence resources necessary for reproduction while mating system males are usually stationary mate choice is based on an inequality of at- and defend some resource required by the fe- tractiveness among mates (Darwin 1871). males for mating, while females move around Mate choice may select for the same charac- searching for suitable males (Emlen and ters that are important in intra-sexual selec- Oring 1977). The costs of mate choice can be tion and thus intra- and inter-sexual sexual se- divided into energy costs, e.g. due to mate lection may operate in the same direction (e.g. sampling (Janetos 1980, Real 1991), and risk Kodric-Brown 1995). On the other hand, costs, e.g. through increased exposure to pre- mate choice may select for characters that are dation (reviewed by Magnhagen 1991). In not favoured by mate competition and thus male–male competition courtship competi- intra- and inter-sexual selection can operate tion or physical fights between males could in opposite directions (Andersson 1994). cause the energy costs. The risk of injury or Which of the two components of sexual selec- death due to rival males or predators during tion is more important often depends strongly fights could be a severe cost of male–male on ecological circumstances (e.g. Lindström competition (reviewed by Andersson 1994). 1988, 1992, Forsgren et al. 1996, Hastings Signals that prevent physical fighting could 1992, Simmons 1994). therefore be beneficial, as they would reduce Selection through mate choice may be these risks and costs (Maynard Smith 1982, based on direct and/or indirect benefits Qvarnström, 1997, Kohda & Watanabe 1990, Sexual selection in the sand goby 9

Frischknecht 1993, Barlow 1986, Grafen male–male competition is tested in more de- 1987). tail. How parental stage and the value of the In fish mating success is often constrained present brood of the male may constrain male by limited egg laying space (e.g. Hastings courtship competition was tested in paper II. 1992). As a consequence male potential re- How nest size may constrain mating competi- productive rate is determined by how quickly tion of attractive and strong males in the wild the eggs hatch and release space for new was tested in paper IV.Finally, how predation spawn. In sea horses and some pipefish the may affect the competition between the males number of eggs and hence reproductive rate is was tested in paper V.The focus of each of the limited by the size of the brooding structure of studies in more detail are as follows: the male (reviewed by Berglund 2000). Simi- Paper I deals with female preference for larly, the number of matings is limited by the males with eggs. This is a phenomenon com- size of the mouth in a number of mouth- mon in several species of fish with paternal brooding fish (reviewed by Godin 1997) or care (Ridley and Rechten 1981, Knapp and the size of the nest substrate in gobies (e.g. Sargent 1989, Marconato and Bisazza 1986, Lindström 1992b). Because of these limita- De Martini 1985, Unger and Sargent 1988, tions, males may sometimes be the choosy Petersen 1989, Hastings 1988, Kraak and sex and females the competing sex (reviewed Videler 1991). Common to the hypotheses by Berglund 2000). explaining female egg preference is that the survival potential of offspring increases with the presence of eggs in the nest. Offspring survival may increase directly through a dilu- The focus of this thesis tion of egg predation in nests containing many eggs (e.g. Rohwer 1978) or through in- This thesis consists of five separate parts, fo- creased parental investment in larger clutches cusing on the mechanisms of sexual selection (e.g. Sikkel 1994, Sargent 1988). The pres- in the sand goby. All of the studies test in the ence of eggs could also be an indicator of the laboratory and/or in the field, factors which quality of the parental male (e.g. Ridley may give direct benefits to females in terms of 1978). offspring survival. Specifically, the factors In species with paternal care the value of tested were: (I) presence of eggs and egg loss the present brood versus future broods may in the nest (potentially indicating safety of the cause a conflict between males and females, nest site in terms of offspring survival). (II) as it may be beneficial for the caring male to age of eggs in the nest (potentially indicating invest in the eggs that are closest to hatching the parental effort of the nest holding male). (e.g. Rohwer 1978, Clutton-Brock 1991, (III) male–male interactions and courtship Lindström 2000). Therefore females are ex- (potentially indicating male vigour). (IV) pected to avoid mating with males guarding male colour and male condition (indicating only old eggs (Sikkel 1994). Paper II deals male vigour and/or attractiveness in mate with female choice for males with old eggs choice selection). (V) occurrence of preda- and factors that may affect female spawning tion at nest sites (indicating safety of the nest patterns in relation to egg age in the wild. site). Furthermore, the constraints of mate One of the basic questions of sexual selec- choice were tested in all of the studies. tion is how male–male competition may af- Whether dominant males may override fe- fect female mate choice (Darwin 1871). male choice was tested in papers I, II, III and There are two ways in which inter- and IV, and how predation constrains female intrasexual selection can interact over evolu- spawning behaviour was tested in paper V. tionary time. In some cases, the two forces Factors affecting male–male competition will act in the same direction on the same were also tested in all of the studies. In papers traits and components of sexual selection, I, II, III, and IV, the effect of mate choice on mutually reinforcing the evolution of the trait. 10 Sexual selection in the sand goby

For example, females prefer dominant males ornament size. Furthermore, according to the to subordinate ones in pupfish (Kodric- indicator theories (e.g. Zahavi 1975), an orna- Brown 1995). In a finnish west coast popula- ment could also be an honest indicator of tion in Tvärminne (60o N, 23o E) where also male condition, which might be important in the present study was conducted (see Fig. 1), a species with male parental care, since males three-spined stickleback females prefer dom- in better condition may be able to provide inant males over subordinate ones (Candolin better quality care (e.g. Hoelzer 1989). To test 1999). whether the colour spot could be an indicator In other cases, the two forces might oper- of male condition, I also investigated the rela- ate on the same trait but in opposite direc- tionship between male nutritional condition tions. In these cases, components of sexual (energy reserves) and fin ornamentation in selection may interfere with, or even elimi- paper IV. nating, each other. Examples of this kind of Finally, paper V focuses on how predation dynamics include female three-spined may function as a constraint on sexual selec- sticklebacks, Gasterosteus aculeatus, from tion in the sand goby. In short living species, an eastern Canadian population, which do not such as the sand goby, individuals may have choose more aggressive or dominant males less chances to postpone reproduction than in (Ward and FitzGerald 1987) and female long-living species (e.g. Magnhagen 1990). three-spines from western Canada, whose Furthermore, under predation, individuals courtship of ’more attractive’ males is dis- may make a trade-off between reproduction rupted by less preferred suitors (McLennan & and risk-taking (reviewed by Magnhagen McPhail 1990). Because of these dynamics, 1991). In the study system, nest sites are male–male interactions should be controlled sparse and individuals may be forced to re- in mate choice experiments (e.g. see Bateson produce at predated sites. They may take the 1987; Andersson 1994; Houde 1997), which so called `best of a bad job option´, which is done in Paper III, (figure 2). In more detail, means that they have to choose between two paper III focuses on the role of male–male in- alternatives that both incur drawbacks (e.g. teractions in mate choice and whether various Lindström and Ranta 1992, see also Fretwell degrees of interactions between males will af- 1972). The effect of predation on reproduc- fect a female’s ability to exercise her prefer- tive behaviour was studied in the wild, where ences. Thus, paper III paves the way for other different constraints of reproduction may in- experiments in this thesis, where males may teract, e.g. because of strong competition for interact or interactions are prevented. nest sites (e.g. Lindström 1988), and in exper- The evolution of bright colours in species imental conditions, where different factors af- with parental care may be based on direct fecting male behaviour could be controlled benefits of mate choice (Hoelzer 1989), as for. More specifically, the effect of predation well as on genetic benefits (Fisher 1958). In on nest site colonisation, courtship and fe- paper IV, I investigated the mechanisms of male mating activity were investigated. Thus, sexual selection on male fin ornamentation. paper V improves our understanding of the In the sand goby, males have a bright blue suitability of nest sites and strong reproduc- spot on the first . This spot is much tive competition for nest sites in the study po- less developed and often missing completely pulation. in females. This was one of the reasons why I expected that the blue spot is under sexual se- lection. Erect fins also play an important part during courtship (e.g. Forsgren 1997a) and Natural history therefore the colour spot could function in mate choice. In paper IV, I tested these hy- The sand goby, Pomatoschistus minutus,isa potheses in separate experiments, by prevent- small bottom dwelling fish. It is common ing male–male interactions and manipulating along the coasts of Europe from the Mediter- Sexual selection in the sand goby 11

Figure 1. The studies of this thesis took place at Tvärminne Zooological Station on the South West Coast of Finland (60o N23Wo, map A). The fieldwork was conducted as shown in the large map (black squares, map B). ranean to the northern coast of Norway published data). (Miller 1986). It reproduces during one sea- Females can spawn at least three batches son only (Healey 1971). Males and females of eggs during the breeding season (Healey are of equal size, but differ in breeding colour. 1971) and the rate of egg development de- Males develop blue and black breeding color- pends on water temperature (Kvarnemo ation on the dorsal, pelvic, and tail fins. The 1994). During spawning the eggs are attached fin ornamentation is non-existent in females. as a single layer to the nest ceiling and thus Migration to shallow bottoms occurs in nest size may limit male reproductive success early spring (Hesthagen 1977) and nesting (Lindström 1992b). After spawning, only the starts when the water temperature reaches 5 to male takes care of the eggs. Care consists of 6 degrees C (Lindström personal communi- the removal of debris and dead eggs and cation). The male builds a nest under a suit- guarding and ventilating the eggs until they able substrate, e.g. a mussel shell or a flat hatch. During parental care, males may con- stone. Male courtship involves displaying sume part of their brood (Lindström 1998). with all fins erect, swimming rapidly with fins Nest take overs resulting in cannibalism of erect in short bouts close to the female and at- the entire brood are also common (e.g. Lind- tempts to lead the female to his nest (Forsgren ström and Hellström 1993, Hamerlynck and 1997a). During courtship males may change Cattrijsse 1994). After hatching, the larvae the intensity of the blue colour on the anal fin start their independent pelagic life (see and during fights between males the colour Méhner, 1992). may intensify and become black (Kangas un- At Tvärminne nest availability is limited 12 Sexual selection in the sand goby

B

A

Figure 2. Tank design in the interaction study (paper III). The experimental tanks were initially di- vided into three sections for all treatments except the interaction treatment. The female was placed into the long section and had visual contact with the males through the transparent divider A. The two males were placed into the two rear compartments, separated by the divider B. The divider B was either an opaque or transparent Plexiglas wall, except for in the interaction treatment, in which the divider B was not present. In the second phase of the experiment, all dividers were removed and all three fish had free access to all parts of the tank. The grey squares represent the ceramic tiles used as nest sites. and hence competition for nests is intense. period of parental care throughout the brood This should result in intra-sexual selection for cycle, without having to resort to filial canni- male characters that determine a male’spo- balism, should therefore be an important cri- tential to hold a nest site. Such a character is terion in female choice. In addition, females for example male size (Lindström 1988, should also pay attention to a male’s ability to Lindström 1992a, Lindström and Kangas defend the nest, as a take over would inevita- 1996). bly be very costly to the female’s reproduc- Egg losses may occur due to filial canni- tive success. Finally the safety of a male’s balism (Lindström 1998) but may also result nest is also important, as the eggs in a nest from nest takeovers as the new owner usually without a male die very quickly (Lindström consumes all of the previous owner’s eggs and Hellström 1993). (Lindström and Hellström 1993). Similarly, if a male dies, e.g. due to predation (Lindström and Ranta 1992), the eggs will be destroyed. General methods Predation by birds, e.g. the common tern, Sterna hirundo, plays an important role in the This study was done at the Tvärminne Zoo- reproductive ecology of the sand goby logical Station situated on the Southwest (Lindström and Ranta 1992), though little is coast of Finland (fig. 1). known about how fish predators affect their Observations and experiments were per- reproduction. Stomach content analysis of formed both in the field (fig. 1) and in aquaria perch, Perca fluviatilis, and pikeperch, (fig. 2). Fish for the aquarium experiments Stizostedion lucioperca, have revealed that were caught from nearby beaches using a reproducing sand gobies constitute a com- beach seine (fig. 1.). Artificial nest substrates mon food source (Koli et al 1985, Hansson were used in experiments. Male mating suc- and al. 1997, Kangas unpublished). cess was estimated by measuring the egg A male’s ability to sustain itself during the mass area in the nest. Sexual selection in the sand goby 13

1.0 ) 1,00 +se

(% 0.8

(%) 0,75 0.6 0,50 0.4 matings within 24 h Nest survival f 0,25 0.2 01234567891011 Frequency o Egg age (d) 0.0 Males with young eggs Males with old eggs Figure 4. Nest survival rate increases with egg Figure 3. In the wild males with late develop- age. mental stage eggs (hatched bar+S.E.) mate less frequently than males with new eggs to invest more in older broods (paper II, figure (white bar+S.E.). 4).

Main results Study III

Study I Interactions between males affected female choice (III). In the treatment where males Females frequently mated with egg-guarding lacked visual and physical contact, females males but this was mostly due to females hav- ing similar mate preferences (I). When eggs 15 were provided to random males, no egg pref- Phase I erence was found and females were instead Phase II found to avoid nests with earlier brood losses, (I) as indicated by the fact that males whose 10 eggs had been removed were less successful in attracting mates. Previous studies have rarely taken into account the possibility that 5 ’egg preference’may in fact stem from avoid- ing to mate with males with low previous frequency of aggression * hatching success. 0

treatment Study II Figure 5. Paper III: The frequency of aggres- A male’s attractiveness was not greatly re- sive interactions (attacks/h + S.E.) during duced by the age of his eggs if females ini- phases 1 and 2 of the experiment. Because tially preferred him. In the field, old eggs had males had no contact during phase 1 in the opaque treatment (star), no interactions could a clearly negative effect on male mating suc- be measured (see figure II for experimental de- cess, as they did not receive new eggs as fre- sign). Attack frequency was highest during quently as males guarding young eggs (II, fig- phase 2 in the opaque treatment. Open bars ure 3). The youngest eggs in a brood were in indicate phase one, before the removal of the greatest danger of being eaten (paper II). Egg dividers, while hatched bars represent phase age correlated with the survival rate of the two, when all three individuals were in physical brood, suggesting that males may be willing contact. 14 Sexual selection in the sand goby

Glass wall Physical interaction (%) Opaque wall 80 80 80 70 70 70 60 60 60 50 50 50 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0

Unmated male courting 0 102030405060708090100 0 102030405060708090100 0 102030405060708090100 Courtshiptime(%)ofthematedmale

Figure 6. When two males were separated by an opaque wall the courtship times did not correlate. When males could follow each other through a glass wall they showed courtship competition and courtship times correlated. When males were not separated by any wall, they could interrupt each others courtship by aggressive encounters and courtship times did not correlate (see also figure 2).

200 Phase I courtship times were positively correlated, ,s

h Phase II suggesting that males competed for matings 150 by courtship competition. Such a correlation engt l was not observed in the treatment where

out males had full physical contact with each oth- b 100

p ers, nor in the treatment where interactions hi were totally prevented. In the interaction 50 treatment males often interrupted each

courts other’s courtship through interference, and in 0 the noninteraction treatment, males were probably unaware of each other’s behaviour treatment during the first phase of the experiment (fig- Figure 7. Paper III: The average length of unin- ure 6). terrupted courtship bouts (seconds + S.E.). All Courtship during the first phase of the ex- courtship behaviours have been combined. periment differed between different treat- Open bars indicate phase one, before the re- ments. Courtship bouts were longest in the moval of the dividers, while hatched bars rep- treatment where the males were undisturbed resent phase two, when all three individuals and separated by an opaque wall, while in the were in physical contact (see figure 2 for exper- other two treatments the possibility for physi- imental design). cal and/or visual interaction made long court- ship bouts unlikely (Fig. 7). were consistent in their choice. However, when males could interact this was not appar- ent. The frequency of aggression was high if Study IV males had had no earlier contact with each other, and low when they had had earlier con- The dorsal fin ornament of males correlated tact with each other (III, figure 5). with male mating success in the lab experi- When males were freely competing to ments (IV, figure 8). Artificially decreasing spawn, the male that was better able to chase this ornamentat decreased male mating suc- his rival from his side was also more success- cess. In the wild the size of the ornament cor- ful in mating. When males were able to see related with egg mass area, but this correla- each other, but not interact physically, their tion was partly confounded by the size of the Sexual selection in the sand goby 15

2 2 4.5 mm ,sd=1 3mm,sd=1 2 2 2.5 mm ,sd=1 1mm, sd=0.3 Figure 8. Main results from the aquarium 8 16 experiments. In experiment with non-ma- nipulated males female spawned in 7 14 seven (7) replicates with the large-spot- 6 12 ted male (black column, mean and stan- dard deviation for spot sizes shown 5 10 above the figure) and in one (1) replicate with the small-spotted male (white col-

ngs 4 8 umn, mean and standard deviation i

at shown shown above the figure). In exper-

M 3 6 iment with manipulated males female spawned in four (4) replicates with the 2 4 manipulated male (cross hatched col- umn, mean and standard deviation 1 2 shown shown above the figure) and in 14 0 0 replicates with the control male (hatched column, mean and standard deviation shown shown above the figure).

2,0 Nesting males Males with large nest Non-nesting males 1,5

1,0

0,5

0,0

-0,5

Figure 9. Male ornament size -1,0 plotted against energy re- -1,5 serves (fat percentage). Relative colour spot size Open symbols: males holding -2,0 a nest. Solid circles: non- 0,00 0,01 0,02 0,03 0,04 0,05 0,06 nesting males. Fat (%) nest a male was guarding (IV). Ornament size ornamented males. At field sites with little did not correlate with male fat reserves, sug- bird predation males were more ornamented gesting that it may not be a condition depend- than at sites with intense bird predation (V). ent trait (paper IV,figure 9). Instead, nest pos- Predation affected nest site colonisation session seemed to be an indicator of male (V). In an experimental situation with no condition (IV, figure 9). competition for nest sites, males avoided nesting at predated sites (figs. 10a and 10b). Males at sites with a high predation risk were Study V smaller or differed in colour from males at safe sites (V). In the wild, attack frequency of Male ornamentation correlated positively diving birds resulted in decreased male brood with male swimming activity under predation guarding behaviour (figure 11). Male court- risk (V), showing that ornamented males be- ship and female activity towards males de- have differently under predation risk than less creased under predation risk (V, figure 12). 16 Sexual selection in the sand goby

Figure 10a. The experimental design of the colonisation experiment: In the predation treatment, an individual male could choose between a sheltered nest site (white square) or a visually predated site (black square). In the control, visual contact with the perch was prevented.

47.2 mm, sd=6.4 47.0 mm, sd=4.4 8 43.2 mm, sd=2.9 18 42.8 mm, sd=4.8 7 16 14 Figure 10b. Main results 6 from the colonisation experi- 12 ment (see figure 9a). Left: 5 The number of nests colo- 10 nised in the control treat- 4 ment (hatched column

nests colonised 8 =control site, white column f 3 =shelter site). Right: The 6 number of nests colonised in the predation treatment 2 4 (black column =predator site, white =column safe Number o 1 2 side). The mean sizes of the males at each site are given 0 0 above bars. Shelter Control Shelter Perch

Discussion vide the female with information about her expected reproductive success. They might 1. Factors affecting mate choice be a cue of a father which takes good care of and male–male competition his young (e.g. Sargent 1988). Thus, in spe- cies with parental care female might choose The stage of parental care good fathers instead of dominant males, and mate choice which has been shown in the Swedish West- Sexual ornaments could evolve because they Coast population of the sand goby (Forsgren signal a male’s condition and hence indicate 1997b). direct benefits such as offspring survival When a randomly chosen sand goby male (Motro 1982, Hoelzer 1992, Zahavi & Zahavi was given eggs, females did not show a pref- 1997). In many species of fish with paternal erence for eggs (I). However, when two males care egg presence and egg quality play impor- simultaneously had received eggs but the tant roles in attracting mates (Rohwer 1978, other one subsequently had his eggs removed, Sikkel 1994). Eggs in a male’s nest might pro- females preferred the male still guarding eggs Sexual selection in the sand goby 17

1,1 eggs (II), showing that egg age affects mating Deep 1,0 Shallow success. These results indicate that females

0,9 may avoid nests with old eggs, because such nests more often than nests with young eggs (%) 0,8 tend to suffer from partial egg losses (II). In

0,7 nests with old eggs the youngest eggs are in the greatest danger of being consumed before 0,6 hatching (Sikkel 1994). males guarding f 0,5 The theory that males with older eggs would have been of poorer condition than 0,4 males with young eggs (Ridley 1978, Number o 0,3 Goldschimdt & al. 1993) was not supported. Males with old eggs and males with young 0,2 eggs had similar body fat contents. Nesting 0 5 10 15 20 males may sustain their condition by canni- Attacks/30 min balising their own eggs during the brood cy- Figure 11. The proportion of males guarding a cle. It is also possible that males did not differ nest decreased as attacks by birds increased. in condition because food was plentiful and Open circles: males at deeper sites. Solid cir- that all males therefore were able to maintain cles: males at predated sites. their condition. Breeding sand gobies are ef- fective predators of macrofauna (Aarnio and (I, II). This suggests that females avoid nests Bonsdorf 1993), however, during the repro- from which eggs have been previously lost. ductive season, breeding and feeding habitats This is interpreted as female preference for of the species may vary in prey availability safe nest sites that are not vulnerable to egg (Zander 1990). For example, the availability predation. A similar case has been shown for of Mysis spp, common prey for sand gobies the bicolor damselfish, Stegastes partitus (Hostens and Mees 1999, Lindström K. per- (Knapp 1993). Thus, egg loss rather than egg sonal communication), may vary both tempo- presence may be an indicator of egg survival rally and spatially (Rudstam and al. 1992, in the sand goby. Hansson and al. 1990, own observations). In the wild, males with young eggs re- Shortage of food in turn may at some loca- ceived more spawnings than males with old tions increase filial cannibalism among car-

30 30 27 27 24 24 21 21 18 18 15 15 Figure 12. Left: Male 12 12 distance to predator and 9 9 control (no visual con-

Distance to female (cm) tact). Right: Male dis- Distance to predator (cm) 6 6 tance to female when 3 3 predator was present and male distance to 0 0 control with no predator Perch Control Perch Control present. 18 Sexual selection in the sand goby ing males for finishing their brood cycle For this reason, females are expected to (Lindström and Hellström 1993), as occurs in choose males that are able to overcome the conditions were interactions between males constraints of male–male competition (e.g. are intense (II). Kodric-Brown 1995). Furthermore, in the A lowered mating success of males with sand goby, male condition may be an impor- old eggs could also be due to lower invest- tant criterion of female mate choice ment in mate attraction by these males (Lindström 1998). Alternatively, in this kind (Rohwer 1978, Sargent 1988, Petersen & of a system, strong males might be able to Marchetti 1989, Lindström and Wennström override female choice. In a similar system of 1994). This hypothesis was not supported by Fiddler Crabs, which spawn in burrows dug the laboratory experiments in which males by males, males lack of females which are did not decrease courtship activity as eggs ready to spawn instantly. In this species mate grew older. However, in these experiments choice may be constrained by mate guarding males had only few eggs to care for and thus of burrow owners or burrow take overs of in- the overall value of the brood might have truding males (Koga and al. 1999). been too small to affect the mating effort of In the sand goby, at densely inhabited nest the males with old eggs. In support of this, in sites male interactions are likely to occur dur- the wild, it was shown that the number of old ing courtship. A female’s choice is therefore eggs in the nest affected mating success, al- likely to be affected by male interactions. though the nest size did not restrict the num- Such interactions could both hamper and fa- ber of spawnings the nest owner could re- cilitate female choice. Subdominant males ceive. Furthermore, in the wild nest survival may not be fully able to display their qualities increased as brood grew older, supporting the thus making it more difficult for female to study by Lindström and Wennström (1994), gather information about male quality. How- which shows that males increase parental ef- ever, interactions could also result in an in- fort as eggs grow older. Thus, increase of pa- creased mating effort by the males, which in rental effort may have decreased the mating turn could have a positive effect on female success of the males with old eggs in the wild. choice as it might allow her to assess the males thoroughly. Besides, an increased level Strong males, availability of nest sites of male–male competition may result in a and mate choice higher average quality of nest holders at In a system where resource availability deter- dense sites. mines male reproductive success one would Moreover, in my experiments, the males expect that a male’s competitive ability corre- decreased courtship behaviour while physi- lated with his condition (Emlen and Oring cally and visually interacting with one an- 1977, Lindström 1992a). In concordance other (III). This resulted in less consistent with this prediction, males without nests were choice by the females. Hence, male–male in- in poorer condition than males with nests. teraction may indeed hamper a female’s abil- Male condition correlated positively with the ity to make a choice. At densely inhabited number of eggs he was guarding (Paper IV). sites interference competition may play an Furthermore, males without nests in the wild important role in determining reproductive will usually build a nest in the laboratory. This success (II, III). At such sites dominant males shows that they are not nutritionally limited in may overcome their neighbours (III). For in- reproduction. Therefore, male–male compe- stance, females may not be able to consider all tition for nest sites probably restricts the nest- important factors, such as egg presence in ing of poor condition males, and not their mate choice, if a male is able to dominate all physiological state. Similarly competition for other males in his vicinity (I, II). On the other large nests might select the males in the best hand, interference from competing males condition to guard the largest nests (Paper does not necessarily reduce mating success of IV). attractive males (papers I and IV). Sexual selection in the sand goby 19

When male interactions were prevented, male nutritional condition (IV, e.g. Johnstone they courted for a longer time, and females and Grafen 1993). Thus it is unlikely that fe- showed consistent preference. This could be males could use male coloration in order to the situation at sparsely populated sites. At assess a male’s energetic ability to complete a such sites males experience less interference brood cycle. and traits such as courtship ability might be Colourful males may be better fathers dur- used and selected for in mate choice. On the ing male–male interactions during which contrary, dominant males that are successful males generally increase filial cannibalism in male–male competition but not otherwise (II). I also found that the spot size was an indi- attractive may not benefit from nesting in ar- cator of a male’s ability to obtain mates in a eas with sparse nest sites. competitive situation, where female choice Furthermore, I found that average egg for traits such as courtship may disappear mass area was positively correlated with nest (III). Thus, colourful males might well be density (Paper V). A decrease in nest density able to overcome the constraints of male– increases sampling distance between nests. male competition and their outlook might in- This may decrease the risk of predation and dicate this ability (sensu Zahavi 1975, the energetic cost of mate choice (e.g. Kodric.Brown 1995). In concordance with Forsgren 1997a, Janetos 1980, Real 1991). this absolute colour area of the fin correlated Also in areas with high nest density predation with male size. Furthermore, male colour risk might be diluted (reviewed by might indicate also other abilities to carry Magnhagen 1991). Hence, females may be costs besides social costs in male–male con- attracted to sites with a high nest density, al- tests. Males possessing a large spot in relation though at such sites male–male interactions to their body size may indicate their capacity might sometimes overweigh their choice. to carry costs involved with it. For example, a Nonetheless, it is currently not clear what fac- large spot size may require a fin size over the tors cause the observed relationship between optimum for swimming (sensu Zahavi 1975, mating success and nest density and further Zahavi and Zahavi 1997). It is therefore pos- studies are required. sible that colour also indicates a male’s ability This study provides evidence that male– to defend his nest against takeovers, which male competition and female choice operate would be an important consideration for the in the same direction, i.e. select for the same female. traits in the same direction. Females choose Colour could also correlate with a male’s characters important in male–male competi- life history strategy. In the laboratory I found tion, probably because of survival benefits to that under predation risk more colourful her offspring. However, I also found candi- males were swimming more than less colour- dates for opposing directions of sexual selec- ful males(V), suggesting that male colour tion. Male courtship and interactions might might be related to their predation avoidance both function as a cue in mate choice. None- strategy. This suggests that colourful males theless, strongly interacting males may con- may behave differently under predation risk strain the choice of females (Kangas own ob- than dull males. Ornamented males might af- servation, paper III). ford to take more risks and/or they might be more attractive as prey than dull males. More Sexual selection of male colour active swimming of ornamented males might Male sand gobies develop a bright blue spot have been a consequence of more frequent es- lined by a white area on their first dorsal fin capes from the predator and bold swimming during the reproductive season. The size of beside the predator’s aquarium, which both the colour spot correlated with male repro- were frequently observed in the experiments ductive success both in aquarium and field (personal observations). Furthermore, Magn- conditions (IV). However, I found no support hagen and Forsgren (1991) showed that indi- for the size of the colour spot being a signal of vidual sand gobies might approach to preda- 20 Sexual selection in the sand goby tory , Gadus morhua, seemingly inspect- able to compete for nest sites may benefit ing the predator. from their ornamentation in mating competi- In the wild males breeding at deep, safer, tion. sites tended to have larger ornamentation (V). This supports the hypothesis that ornamented Nest availability and predation and dull males avoid predation in different Individuals may often have to depart from ways. In pipefish, Syngnathus typhle, females their optimal decisions due to ecological con- that compete for males at predated sites are straints. Predation risk forces to com- less ornamented (Bernet & al. 1998). Colour- promise their behaviour. This may result in ful males may benefit more from nesting at the adoption of less optimal alternatives. In safe sites than dull males, because at safe sites reproduction this is often caused by the need females perform an active choice (Forsgren to simultaneously optimise both current and 1992). In populations of guppies, females future reproductive success. Consequently tend to decrease their sexual activity under individuals may have to choose less optimal predation risk (Godin & Briggs 1996). The breeding sites when faced with predation abilities to carry social costs in male–male threats. I found that male sand gobies avoided competition (paper IV) and avoid predation nesting close to a predatory perch in condi- are important direct benefits in terms of off- tions with no nest site competition (V). In- spring survival. These benefits might out- stead they colonised sheltered sites. Predation weigh any indirect benefits such as genetic may thus restrict the number of suitable nest quality. sites in the wild. Predation avoidance could play an important role in determining the dis- tribution of males especially in the beginning 2. Habitat constraints of the reproductive season when competition on sexual selection for nest sites may still be relatively weak (own observations). Nest availability and nest size In the wild I found that males also colo- When only males with small nests where in- nised high predation sites. However, I also vestigated in the wild, the size of the colour found a tendency for the males colonising spot did not correlate with reproductive suc- these sites to be smaller than males at safe cess. This suggests that nest size may restrict sites. These field results are in accordance reproductive success of ornamented males. with the laboratory results as this indicates Earlier studies have shown that nest size is that males competed more for the safe sites one of the basic determinants of male repro- (see also Lindström and Ranta 1992). ductive success in the sand goby (e.g. In shallow habitats predation by pisci- Lindström 1992b). The eggs are attached to vorous birds was higher than at deeper habi- the nest substrate in a single layer. Thus the tats where attacks by birds were rarely ob- nest size restricts the space available for eggs served. Nonetheless, nest colonisation rates and mating success becomes a correlate of the between these habitats were similar. This may size of the nest (e.g. paper IV). Consequently suggest that male sand gobies do not, or can- female spawning patterns are constrained by not, choose their nesting habitat in order to variation in nest size and egg numbers. The avoid predation by birds. However, these most fecund females are most restricted by habitats also differed in other aspects, e.g. in variation in available egg space between nests nest density and water temperature. Thus also (Lindström 1992b). Also previous studies these factors may have affected the colonis- support the idea that sexual selection in the ation rates between habitats. sand goby is constrained by nest availability In the laboratory, females decreased their and nest size in the study population (see also activity in the presence of a predator, even if Lindström 1988,1992a, 1992b, Lindström they didn’t clearly show any avoidance of the and Ranta 1992, paper V). Hence, only males predator. Consequently females might be less Sexual selection in the sand goby 21 likely to sample males in areas of high preda- that in the wild females spawn more often in tion risk. In addition, males also decreased nests containing young eggs than nests con- their courtship activity under predation, as taining old eggs. It is suggested that this pat- has also been shown for the wolf spider, tern is based on the higher risk for younger Hygrolycosa rubrofasciata (Kotiaho & al. eggs to be cannibalised by the males. Paper III 1998). Nests at high predation sites in the shows the importance of male interactions in field did, however, contain as many eggs as mate choice and courtship competition. It those at safe sites. Male mating success was also suggests that males in densely inhabited thus not necessarily affected by predation habitats might pursue a different reproductive risk. The decrease in courtship activity and strategy than males in sites with distant neigh- frequent absence from the nest by the male bours. Paper IV shows that male dorsal fin during periods of elevated predation risk, colour is a sexually selected trait. In the wild may make it more difficult for females to reli- this selection is constrained by nest size varia- ably assess male quality during the periods of tion. Furthermore, the results do not support high predation risk. Increased predation the idea that the fin ornament indicates male could thus result in an increased sampling nutritional condition. However, paper V sug- cost and perhaps lowered selectivity of fe- gests that male colour could be an indicator of males (Forsgren 1992, 1997a). male predation avoidance strategy. More fo- Adecrease in female choosiness would in- cused, paper V shows experimentally that evitably benefit unattractive males. The fact males avoid nest sites exposed to predators. that poorer quality males were found to nest at Strong reproductive competition among high predation sites could, however not nec- males may thus have caused the differences essarily be a consequence of this. The poor that were found in male quality between high quality males may have also been forced to predation and safe sites in the wild. It is also choose the `best of a bad job option´, and nest shown that males decrease courtship activity at predated sites, as the alternative option may and females decrease mate choice activity un- have been to be without a nest. der predation risk. Thus predation may be a cost of reproduction in the sand goby. Nutritional and energy demands These studies have shown that females for breeding prefer to mate under circumstances that indi- Besides predation, food could also affect the cate high offspring survival (I, II, IV, V). An distribution of breeding sand gobies. I found interesting finding is that female choice is not that in the field non-breeding gobies were necessarily always operating in the same di- usually in poorer body condition than nesting rection as male–male competition. Either one gobies (IV). I also found that male nutritional may outweigh the other, but which one is condition correlated with the number of eggs stronger may be dependent on local ecologi- he was guarding. If the ability to colonize and cal conditions. For example, females may hold a large nest requires good condition (e.g. prefer to spawn with males with young eggs, Lindström 1992b) then the quality of gobies but due to other differences among males they breeding at different sites should correlate may be forced to mate with males with old with food abundance (see also Kvarnemo eggs (II). Male interactions may outweigh 1997). mate choice (I, II, III), but not in cases where males differ clearly in ornamentation (IV). I also show that under certain conditions fe- 3. Conclusions male choice may constrain male–male com- petition (III, IV). For example females may This thesis describes five basic mechanisms choose males that are weaker in physical of sexual selection. Paper I brings up the im- fights between males. These results suggests portance of direct benefits of mate choice in that different components of sexual selection terms of offspring survival. Paper II shows operate together resulting in stabilising selec- 22 Sexual selection in the sand goby tion as well as directional selection depending company in the field. Jaakko Ilvonen was taking his on the environment. time to catch the sand gobies when I needed help From the point of view of reproductive (and other fishes as well). Janne Prokkola contrib- ecology it was an interesting finding that uted to the layout of this thesis and was good com- brood age variation may affect competition pany during all of those skating excursions in the na- for spawning and consequently female ture. I wish to thank also all my friends outside the spawning patterns (II). Furthermore, this University for being there. The members of the fam- spawning pattern may constrain female ily were always there to support and help of which I choice in the wild, when most males are am privileged and thankful. This study was funded guarding old eggs. Also, the finding that by Grants from Finnish Culture Fund, Emil Aaltonen males avoid high predation risk nest sites dur- Foundation, Walter and Andrée de Nottbeck Foun- ing colonisation does add a piece to the puzzle dation, the University of Helsinki and Finnish Acad- of understanding the strong competition for emy. nest sites in the study population (V).

Acknowledgements References I thank Kai Lindström for his guidance throughout this work. Tvärminne Zoological station offered ex- Aarnio, K. and Bonsdorff, E. 1993. Seasonal varia- cellent facilities and nice surroundings to conduct the tion in abundance and diet of the sand goby work. I am more than grateful to professor Lotta Pomatoschistus minutus (Pallas) in a Northern Sundström for her advice, help and teaching. I ac- Baltic Archipelago. Ophelia, 37:,19–30. Andersson, M. 1994. Sexual Selection. Princeton, knowledge professors Kristian Donner, Ilkka New Jersey: Princeton University Press. Hanski, Jari Niemelä, Hannu Pietiäinen, Esa Ranta, Bakker, T. C. M. 1993. Positive genetic correlation and Markku Viitasalo for the invaluable support, between female Preference and preferred male or- counsel and teaching during the different procedures nament in sticklebacks. Nature 363, 255–257. of my Ph.D.-work. I´m greatly indebted to Hannu Barlow, G.W. 1986. Mate choice in the monogamous Rita for the consultation and analytical expertise polychromatic midas cichlid, Cichlasoma citrinellum. Journal of Fish Biology, 29, 123–133. dealing with probabilities, coincidence and related Bateson, P. 1987. Mate Choice. Cambridge: Cam- topics. I owe my gratitude to Johanna Mappes for her bridge University Press. constructive criticism and comments during the end Berglund, A. 2000. Sex role reversal in pipefish: fe- of the study. Sirkka Liisa Nyeki was always there to male ornaments as amplyfying handicaps. help when it was question of finding books and other Annales Zoologigi Fennici, 37,1–13. literature. Kari Vepsäläinen offered me an extra li- Bernet, P., Rosenqvist, G., and Berglund A. 1998.Female–female competition affects female brary and was always kind and helpful. Lotta Haldin ornamentation in the sex-role reversed pipefish was good work company with sand gobies and she Syngnathus typhle. Behaviour, 135, 535–550. was always positive and helpful. For various kinds of Bisazza, A., Marconato A. & Marin G. 1989. Corre- contributions during this study I owe my recognition lates of male reproductive success in padagobius and thanks to Ilkka Teräs, Marja Koski, Mika Kilpi, martensi (Gobidae ). Journal of Fish Biology, 34, Markus Öst, Lauri Koli, Patrik Byholm, Matti 889–899. Bradbury, J. W. and Andersson, M. 1987. Sexual Koivula, Janne Sundell, Heikki Hirvonen, Paa Selection: Testing the Alternatives. Chistester, Helstedt, Jyrki Holopainen, Heidi Kinnunen, Jukka John Wiley & Sons. Lehtonen, Markku Lehtonen, Atte Moilanen, Anssi Brooks, R. and Couldridge, V. 1999. Multiple sexual Laurila, Teija Aho, Anou Londesborough, Minna ornaments coevolve with multiple mating prefer- Pyykkö, Jukka Palo and Janne Kettunen. Very spe- ences. American Naturalist, 154,37–45. cial thanks to Lallu, Magnus, Raija, Bebbe, Svante, Burley, N. Tyler and Symanski, R. 1998. “Ataste for the beautiful”: Latent aesthetic mate preferences Jari, Totti, Jouko, Kale, Eva, and all others in for white crests in two species of Australian Tvärminne. I also want to thank Lotta Kvarnemo, grassfinches. American Naturalist, 152, 792–802. Outi Pihlajamäki, Jutta Kujasalo and Helena Candolin, U. 1999. Male–male competition facili- Suomela, and Stefano Malawasi for their help and tates female choice in sticklebacks. Proceeding of Sexual selection in the sand goby 23

the Royal Society of London, Series B, 266, 785– gian coastal waters, and a comparison with the 789. food of its potential competitor P. lozanoi. Journal Clutton-Brock, T.H. 1991. The Evolution of Parental of Fish Biology, 44, 753–771. Care. Princeton, New Jersey, Princeton university Hansson, S, Arrhenius, F. and Nellbring, S. 1997. press. Diet and growth of pikeperch (Stizostedion Clutton-Brock, T.H.& Parker, G.A. 1992. Potential lucioperca L.) in a area. Re- reproductive rates and the operation of sexual se- search 31, 163–167 lection. Quarterly rewiew of biology. 67, 437– Hansson, S., Larsson, U. and Johansson, S. 1990. 456. Selective predation by and mysids, and Colgan, P. W. & Salmon. A. B. 1986. Breeding expe- zooplankton community structure in a Baltic Sea rience and parental behaviour in convict cichlids coastal area. Journal of plankton research. 12, (cichlasoma nigrofasciatum). Behavioural pro- 1099–1116. cesses, 13, 101–18. Hastings, P. A. 1988. Female choice and male repro- Darwin, C. 1871. The Descent of Man and the Selec- ductive succes in the angel blenny, Corralliozetus tion in Relation to Sex. London, John Murray. angelica (Blennoidea: Chaenopsidae). Animal DeMartini, E. E. 1985: Social behaviour and color- Behaviour., 36,115–124. ation changes in painted greenling, Oxylebius Hastings, P. A. 1992. Nest-site size as a short-term pictus (Pisces Hexagrammidae). Copeia, 4, 966 – constraint on the reproductive success of paternal 975. fishes. Environmental Biology of Fishes, 34, 213– Emlen, S.T. And Oring, L. W. 1977. Ecology, sexual 218. selectionand the evolution of mating systems. Sci- Healey, M. C. 1971. Gonad development and fecun- ence,197, 215–223. dity in the sand goby, Pomatoschistus minutus Fisher, R A. 1958. The Genetical Theory of Natural Pallas. Transactions of American Fisheries Soci- Selection. Second edition. New York, Dover pub- ety,100,520–526. lications. Hesthagen, I. H. 1977. Migrations, breeding and Forsgren, E. 1992 Predation risk affects mate choice growth in Pomatoschistus minutus, Pallas, in in gobiid fish. American Naturalist, 140, 1041– Oslofjorden , Norway. Sarsia, 63,17– 26. 1049. Hoelzer, G. A. 1989. The good parent process of sex- Forsgren, E., Kvarnemo C. and Lindström K. ual selection. Animal Behaviour, 38, 1067–1107. 1996. Mode of sexual selection determined by re- Holland, B. and Rice, W. R. 1998. Perspective: source abundance in two sand goby populations. Chase-away sexual selection: Antagonistic se- Evolution 50, 646–654. duction versus resistance. Evolution, 52,1–7. Forsgren, E. 1997a. Mate sampling in a population of Hostens, K. and Mees, J. 1999. The mysid-feeding sand gobies. Animal Behaviour, 53, 267–276. guild of demersal fishes in the brackish zone of the Forsgren, E. 1997b. Female sand gobies prefer good Westerschelde . Journal of Fish Biology, fathers over dominant males. Proceeding of the 55, 704–719. Royal Society of London, Series B, 264, 1283– Houde, A. E. 1997. Sex, Color and Mate Choice in 1286. Guppies. Princeton, New Jersey: Princeton Uni- Fretwell. S.D. 1972. Populations in a Seasonal Envi- versity Press. ronment. Princeton, Princeton University Press. Janetos, A. C. 1980. Strategies of female mate Frischkencht, M. 1993. The breeding colouration of choice: a theoretical analysis. Behavioural Eco- male three-spined sticklebacks (Gasterosteus logy Sociobiology, 17, 271–278. aculeatus) as an indicator of energy investment in Johnstone, R. and Grafen, A. 1993. Dishonesty and vigour. Evolutionary Ecology, 7, 439–450. handicap principle. Animal Behaviour, 46, 759– Godin, J. G. J. (ed). 1997. Behavioural Ecology of 764. Teleost Fishes. New York, Oxford university Knapp, R. A. 1993. The influence of egg survivor- press. ship on the subsequent nest fidelity of female Godin, J. G. J. and Briggs, S. E. 1996. Female mate bicolor damselfish, Stegastes patritus. Animal Be- choice under predation risk in the guppy. Animal haviour, 46, 111–121. Behaviour, 51.117–130. Knapp, R. A. & Sargent R. C. 1989: Egg mimicry as Goldschmidt, T., Bakker, T. C. M, Feuth-de a mating strategy in the fantail darter, Etheostoma Bruijin, E. 1993. Selective copying in female flabellare: females prefer males with eggs. Behav- sticklebacks. Animal Behaviour, 45, 541–547. ioural ecology and sociobiology. 1989, 25, 321 – Grafen, A. 1987. The logic of divisively asymmetric 326. contests: respect for ownership and the desperado Kodric-Brown, A. 1983. Determinants of male re- effect. Animal Behaviour, 35, 462–467. productive success in pupfish (cyprinodon Hamerlynck, O and Cattrijsse, A. 1994. The food of peconensis). Animal behaviour, 31, 128–137. Pomatoschitus minutus (Pisces, ) in Bel- Kodric-Brown, A. 1995. Does past reproductive his- 24 Sexual selection in the sand goby

tory predict competitive interactions and male 7, 213–217. mating success in pupfish? Animal Behaviour, 50, Lindström, K. and Ranta E. 1992. Predation by 1433–1440. birds affects population structure in breeding sand Koga, T., Murai, M. and Yong, H. S. 1999. Male– goby, Pomatoschistus minutus, males. Oikos 64, male competition and intersexual interactions in 527–532. underground mating of the fiddler crab Uca Lindström, K. and Wennström, C. 1994. Expected paradussumieri. Behaviour, 136, 651–667. future reproductive success and paternal behav- Kohda, Y. & Watanabe, M. 1990. The aggression- iour in the sand goby, Pomatoschistus minutus releasing effect of the eye-like spot of the (Pisces, Gobiidae). Journal of fish biology, 44, oyanirami Coreoperca kawamebari a freshwater 469–477. serranid fish. Ethology, 84, 162–166. Magnhagen, C. 1990. Reproduction under predation Koli, L., Rask, M., and Aro, E. 1985. Growth, age risk in the sand goby, Pomatoschistus minutus, distribution and year class strength of perch, and the , niger: the effect of age Perca fluviatilis L., at Tvärminne, Northern Baltic and longevity. Behavioral Ecology Sociobiology, Sea. Aqua Fennica, 15, 161–167. 26, 331–335. Kotiaho, J., Alatalo, R. V. Mappes, J. Parri, S. Magnhagen, C. 1991. Predation risk as a cost of re- Rivero, A. 1998. Male mating success and risk of production. Trends in Ecology and Evolution, 6, predation in a wolf spider: Abalance between sex- 183–186. ual and natural selection? Journal of Animal Eco- Magnhagen, C. and Forsgren E. 1991. Behavioral logy. 67, 287–291. responses to different types of predators by sand Kraak, S. B. M. & Videler, J. J. 1991: Mate choice in goby Pomatoschistus minutus: An experimental Aidablennius sphynx (teleostei, Blennidae); Fe- study. Marine Ecology progress series, 70,11–16. males prefer nests containing more eggs. Behav- Marconato, A & Bisazza, A. 1986: Males whose iour.,119, 243–266. nests contain eggs are preferred by female Cottus Kvarnemo, C. 1994. Temperature differently affects gobio L. (Pisces Cottidae). Animal Behaviour. 34, male and female reproductive rates in the sand 1580–82. goby: consequences for operational sex ratio. Pro- McLennan, D. A. & McPhail, J. D. 1990. Experi- ceedings of the Royal Society of London Series B. mental investigations of the evolutionary signifi- 256, 151–156. cance of sexually dimorphic nuptial colouration in Kvarnemo, C. 1997. Food affects the potential repro- Gasterosteus aculeatus (L.): the relationship be- ductive rates of sand goby females but not of tween male colour and female behaviour. Cana- males. Behavioral Ecology, 8, 605–611. dian Journal of Zoology, 68, 482–492. Lindström, K. 1988. Male–male competition for nest Maynard Smith, J. 1982. Evolution and the Theory sites in the sand goby, Pomatoschistus minutus. of Games. Cambridge, Cambridge University Oikos, 53,67–73. Press. Lindström, K. 1992a. The effect of resource holding Méhner, T. 1992. Diet spectra of Pomatoschistus potential, nest size, and information about re- microps (Kroyer) and Pomatoschistus minutus source quality on the outcome of intruder–owner (Pallas) (Teleostei, Gobiidae) during first weeks conflicts in the sand goby. Behavioral Ecology after hatching. Zoologischer anzeiger, 229,13– and Sociobiology, 30,53–58. 20. Lindström, K. 1992b. Female spawning patterns and Miller, P. J. 1986. Gobiidae. In: Fishes of north-East- male mating success in the sand goby, Pomato- ern Atlantic and Mediterranean. Paris, UNESCO. schistus minutus. Marine Biology, 30,53–58 Moore, A. J. and Moore, P. J. 1999. Balancing sex- Lindström, K. 1998. Effects of costs and benefits of ual selection through opposing mate choice and brood care on filial cannibalism in the sand goby. male competition. Proceedings of the Royal Soci- Behavioral Ecology and Sociobiology. 42, 101– ety-Biological Sciences Series B. 266,711–716. 106. Motro, U. 1982. The courtship handicap: phenotypic Lindström, K. 2000. The evolution of filial cannibal- effect. Journal of theoretical biology. 97, 319– ism and female mate choice strategies as resolu- 324. tions to sexual conflict in fishes. Evolution . 54, Penn, D. J. and Potts, W. K. 1999. The evolution of 617–627. mating preferences and major histocompatibility Lindström, K. and Hellström, M. 1993. Male size complex genes. American Naturalist. 153, 145– and parental care in the sand goby, Pomato- 164. schistus minutus. Ethololgy Ecolology and Evolu- Petersen, C. W. 1989., Females prefer mating males tion, 5,97–106. in the carmine triplefin, Axoclinus carminalis,a Lindström K, Kangas, N. (1996) Egg presence, egg parental brood guarder. Environmental Biology of loss, and female mate preferences in the sand goby Fishes. 26, 213–221. (Pomatoschistus minutus) Behavioural Ecology Petersen , C.W. & Marchetti, K. 1989. Filial canni-