R. Kortet, Neurobiology, Physiology and Behavior, University of California, Davis, One Shields Avenue, Davis, CA 95616, USA. Tel.: (530)-752-1562; fax (530)-752-5582; e-mail: firstname.lastname@example.org
According to the ‘good genes’ hypothesis, sexual ornaments provide an indication of the ‘quality’ of the bearer. In roach, Rutilus rutilus, breeding tubercles (BTs) may signal resistance against the digenean parasite, Rhipidocotyle campanula. Life history theory predicts that there should be a trade-off between parasite resistance and other life history traits. In roach, this could imply a trade-off between parasite resistance in mature fish and some larval feature. We studied embryo survival and the early viability of larvae of male roach in relation to expression of BTs and parasite resistance in maternal half-sibling families. Highly ornamented males had higher resistance against R. campanula than less ornamented males, but the BTs were not related to either embryo survival or larval viability. However, sires having higher resistance to R. campanula had lower larval viability. These results suggest that BTs of male roach do not indicate ‘quality’ in terms of early survival or viability, but rather in terms of adult parasite resistance.
Secondary sexual ornaments of males (and female preferences for them) may have evolved as indicators of some aspects of male ‘quality’. Originally, this ‘quality’ might have been the adaptiveness of a male to his environment (e.g. Pomiankowski & Iwasa, 1998). In ‘Fisherian’ models of the evolutionary process, females are suggested to prefer highly ornamented mates because the preference for these traits becomes genetically coupled with ornamentation genes (Fisher, 1930). In ‘good genes’ models of sexual selection, females prefer highly ornamented males because they have superior genes (e.g. Hamilton & Zuk, 1982; Folstad & Karter, 1992; Westneat & Birkhead, 1998; Kokko et al., 2003). However, Kokko et al. (2002) recently showed that sexual selection may result from simultaneous effects of mate choice both on viability and attractiveness of offspring, the processes that had long been considered as two separate mechanisms. In addition, as sexual conflict theory predicts that males and females will have divergent interests in reproduction (review by Chapman et al., 2003; Kokko et al., 2003), it is likely that females prefer male traits that are genetically linked to features that are advantageous for females or features that provide advantages to both sexes, e.g. parasite and pathogen resistance. In support of this, female preferences for traits signalling male ability to resist parasites and pathogens have been described in several species (e.g. review by Hillgarth & Wingfield, 1997; Rantala et al., 2002; Rantala & Kortet, 2003).
A meta-analysis by Møller & Alatalo (1999) suggests that although ‘good genes’ effects in sexual selection occur, those effects are minor. Empirical evidence for female choice for ‘good genes’ is mixed, as most of the previous studies have been conducted using species, e.g. birds, that do have parental care for their offspring (but see Sheldon et al., 1997). Moreover, in earlier studies, the possible negative genetic correlations between male ornamentation and the male offspring's survival may have been interpreted incorrectly as falsifying the ‘good genes’ effects (see Kokko, 2001). In the scorpionfly, Panorpa vulgaris, the sons of highly ornamented males have been found possess a slightly better immunocompetence than sons of less ornamented males (Kurtz & Sauer, 1999). However, females may bias their investment to their eggs with respect to male attractiveness, giving advantage to the eggs fertilized by a high-quality male.
Many teleosts have nonrandom mating with external fertilization without parental care. Therefore, fishes provide ideal taxa for studies on sexual selection. Moreover, in certain fishes, maternal half-sib designs are possible, allowing direct studies of the ‘good genes’ hypothesis (e.g. Barber et al., 2001; Wedekind et al., 2001).
The few maternal half-sibling studies conducted so far indicate a strong genetic effect of parental sexual advertisement on offspring viability in whitefish, Coregonus sp. (Wedekind et al., 2001), and on parasite resistance in stickleback, Gasterosteus aculeatus (Barber et al., 2001). In studies conducted in tree frogs, Hyla versicolor, the offspring of long-calling males had overall better survival-related performance (Welch et al., 1998) and spent more time feeding (Doty & Welch, 2001) than did the offspring of short-calling males, giving some support to the ‘good genes’ hypothesis. Moreover, moor frog, Rana arvalis males with bright blue dorsal coloration sired offspring that had higher survival when exposed to predatory water beetles than males with dull coloration (Sheldon et al., 2003).
According to the theory of life history evolution, there should be trade-offs in the investment of resources between different life history traits (Roff, 1992; Stearns, 1992). Finding evidence for such trade-offs can be a challenge, however, because individuals of better overall ‘quality’ may show superior performance in several traits, even if a trade-off operates between them (De Jong & van Noordwijk, 1992). Recent studies indicate a trade-off between immune defence and other life history characters (reviewed by Lochmiller & Deerenberg, 2000; Norris & Evans, 2000). If parasite resistance is heritable, as has been shown in birds and mammals (review in Wakelin & Apanius, 1997), the offspring of high-resistance sires should be better at resisting parasites, but could be poorer in some other life history trait, for example, early survival (Kokko, 2001). This might be due to either antagonistic pleiotropy between parasite resistance genes and survival genes, or linkage disequilibrium between resistance and deleterious alleles (see Kokko et al., 2003). However, such life history trade-offs between adults and their offspring with respect to parasite resistance have only rarely been studied.
We studied the predictions of the ‘good genes’ hypotheses by testing the early survival of the eggs and the viability of the offspring in relation to the sexual ornamentation of the fathers using maternal half-sibling families. Since previously, roach males that differed in their BTs were found to differ in their sperm longevity (Kortet et al., 2004) we predicted differences in the early survival of embryos sired by males with different degrees of BT-ornamentation. More specifically, we examined (1) sexual ornamentation, i.e. BTs on the sides and operculum, (2) resistance against R. campanula, a digenean parasite (proportion of the dead parasites in natural infections), (3) in vitro embryo survival and (4) viability of larvae at days 7 and 14. As the infective cercariae of R. campanula are not found during early summer (Taskinen et al., 1991, 1994), we assumed that roach fry cannot be infected by the parasite during the first weeks of their life. Therefore, we also studied the relationship between the parasite resistance of sires and survival of the offspring during the first 2 weeks of their life to detect a possible trade-off between early larval viability and heritable parasite resistance.
Materials and methods
Fish collection and experiments were conducted on 1 June 2002, with permission from the Central Finland Regional Environment Center (permission LS-31/00). Roach were gill-netted on the spawning grounds of Lake Konnevesi, Finland (62 °15′N, 26 °26′E) at a water temperature of 16 °C. The spawning of roach occurred in Lake Konnevesi between 1 and 4 June. After collection, the fish were transported 1 km to University of Jyväskylä's Konnevesi Research Station in tubs filled with lake water.
The BT-ornamentation of males (roughness on the sides of the body and opercula) was scored (Taskinen & Kortet, 2002) and the males were classified into three categories based on their sexual ornamentation: (0) no clear BTs, smooth skin, (1) BTs clearly visible, slightly or moderately rough skin, and (2) very rough skin with many large BTs. Males that were difficult to classify were excluded. One person conducted the ranking of all fish. This ranking method of male roach has been shown previously to be repeatable (Taskinen & Kortet, 2002). The number of fish within each categories were nine, 18 and 17 (groups 0, 1 and 2, respectively). After the collection of milt and roe all the fish were killed, their total length was measured and their gills were dissected for parasite examination (see below).
Rhipidocotyle campanula uses predatory perch or pike perch (zander) as the final host, whereas bivalve clam (Anodonta piscinalis) and roach are the first and the second intermediate hosts, respectively (Taskinen et al., 1991; Gibson et al., 1992). The transmission of R. campanula to roach is limited to the period between mid-June and September (Taskinen et al., 1994; Taskinen, 1998). Thus, the proportion of dead R. campanula found at the time of spawning in early June should indicate the defensive responses of the host against the parasites that invaded into gills at least 8–11 months ago. Roach have been shown to mount a humoral (antibody-mediated) response against Rhipidocotyle metacercaria (Aaltonen et al., 1997). Rhipidocotyle campanula is assumed to be harmful to roach since the larvae migrate to the gills (Taskinen et al., 1991) where they interfere with blood circulation (Baturo, 1977). Heavy infections by bucephalid digeneans, like R. campanula, have been reported to cause considerable mortality in roach (e.g. Hoffmann et al., 1990). Moreover, R. campanula has been observed to be most prevalent metazoan parasite of the current study area (Taskinen & Kortet, 2002).
Rhipidocotyle campanula on all gill arches of each fish were examined microscopically using transmitted light and by pressing the gills between two large glass plates. Number of living and morbid R. campanula were counted and the proportions of dead parasites were used as a measure of parasite resistance of the host (Taskinen & Kortet, 2002). Live parasites were moving inside the metacercarial cyst, by contrast to dead specimens that were motionless and more or less brownish in colour due to variable degrees of melanization, and often clearly degenerated. This method for separating live and dead parasites has been shown previously to be repeatable (Taskinen & Kortet, 2002).
In vitro fertilization, embryo survival and larval viability
In the laboratory, the milt of 44 males (mean length ± SD; 164.7 ± 4 mm), and roe of 18 gravid females (186.6 ± 5 mm) were collected for the study. Prior to collecting gametes, the fish were anaesthetized using 0.01% MS-222 (Sigma Chemical Co., St Louis, MO, USA). All the fish were sexually mature and ready to spawn, which was ensured when collecting the gametes: they self-released milt or roe immediately when touched. Before sampling milt or roe, the gonopore area was dried with a tissue paper. Sperm were collected using a 1-mL syringe and then stored in tubes on ice for an hour at maximum. Urine-contaminated sperm was not used. This sampling method has been successful to ensure viability and motility of roach sperm (Kortet et al., 2004). Roe was carefully stripped into a plastic cup by applying slight pressure to the belly of a female. Ova were then collected carefully using an edge-rounded 5-mL syringe in preparation for in vitro fertilization. Fertilization was conducted not later than 1 h after sperm collection and 3 min after roe collection.
Size-matched pairs of males (18 pairs, mean difference within pair ± SD, 7 ± 1.5 mm) were used to fertilize eggs from a single female to give separate half-sibling groups. In cyprinids, like roach, fish size corresponds very strongly with fish age (e.g. Spivak et al., 1979). Therefore, by controlling the size of the male pairs, we controlled their age as well. The males within a pair differed in their ornamentation. The half-sibling clutches of eggs per female were produced by mixing, using a small feather, 30 μL of semen and 3 mL (mean ± SD, 162.3 ± 13 eggs) of roe in a 0.5-L plastic box. The number of sperm cells in the given volume of semen is extremely high. Therefore, all the eggs in the experiment received a superabundance of sperm cells and the number of sperm cells was not a limiting factor. The number of the males within each classification groups were 7, 15 and 14 (groups 0, 1 and 2, respectively). The fertilizing order of the sperm of the fathers that was used was randomized. After careful mixing of the roe and the sperm, the fresh 16 °C lake water was added immediately to activate the sperm cells. The room temperature was 18 °C, so the water temperature gradually rose to that point.
Eggs were examined 2 days after fertilization, and the number of eyed eggs in relation to the initial count of eggs was used as a measure of embryo survival. In roach, the eyes of eggs become visible after 2 days after fertilization at 18 °C. Eggs that were unfertilized, not developed or showed signs of fungal infection were removed using pasteur pipette.
Eggs were expected to hatch in approximately 9 days at 18 °C. Viable eggs and larvae were counted every day for 2 weeks and the number of survivors in relation to the initial number of living eggs at day 2 was used as a measure of larval viability. For simplicity, viability was assessed 7 and 14 days after fertilization. New aerated water in the incubation chambers was changed daily using a syringe, and dead larvae were removed. New water was from the pelagic area of the lake of origin of the experimental roach population. Water exchange was not aseptic, and no effort was made to treat possible infections. Hatched fry were fed zooplankton (mainly juvenile water fleas, Daphnia sp.) collected using plankton traps (0.4 μm) from the pelagic area of nearby Lake Konnevesi. The food was provided using a syringe and the fry were observed to ingest the food. After day 14, all the remaining fry were killed.
The relationships between male ornamentation and resistance against R. campanula, and male ornamentation and intensity of the R. campanula infection were tested using ancovas, using length as a covariate. To study possible maternal effects on embryo survival and larval viability we used Pearson correlation analysis as used by Barber et al. (2001).
The effect of male ornamentation on in vitro embryo survival and larval viability was analysed using a Wilcoxon signed rank sum test for dependent samples (size-matched male pairs) since the data were not normally distributed. Males having the higher classification rank of BTs were nominated as the highly ornamented individual within a pair. Finally, the paired data set was arranged according to resistance against R. campanula, (proportion of dead parasites) and analysed using a Wilcoxon signed rank sum test. Both the tested hypothesis (i.e. ‘good genes hypotheses’ and ‘hypotheses for life history trade-offs’) allow us to use directed statistics. Therefore, we use α = 0.05 (α = critical value set for statistical significance after Bonferroni correction) for the interpretation of the paired tests. Because of limited sample sizes, we analysed the statistical power of the paired tests as described for paired samples in Zar (1999). For the studied parameters the power varied from 0.05 (embryo survival) to 0.31 (larval viability).
All males were infected with R. campanula, with the number of parasites varying from 13 to 123. The proportions of dead parasites ± SE within ornamentation classes were 0.45 ± 0.05, n = 9 in class (0) smooth skin, 0.55 ± 0.03, n = 18 in class (1) moderately rough skin and 0.65 ± 0.03, n = 17 in class (2) very rough skin. The BTs of a male were significantly related to his proportion of dead R. campanula (F43,2 = 5.035, P < 0.05). The length as a covariate was also significant (F43,1 = 11.229, P < 0.01) so that larger males had a better resistance. However, the intensity of the R. campanula infection (i.e. parasite load) was not related to the sexual ornamentation of males (F43,2 = 2.137, ns), or to the length as a covariate (F43,1 = 0.028, ns).
The parasite load and the proportion of dead R. campanula were not correlated (Pearson's r = 0.038, P = 0.808, n = 44). However, when the proportion of dead R. campanula was weighted by the parasite load by using the following formula: resistance = proportion of dead parasites × [number of parasites/123 (=the maximum count of parasites)] to control for possible increased energetic costs associated with an increased number of parasites, the positive association with resistance and BTs was still found: (F43,2 = 4.329, P < 0.05), length as a covariate (F43,1 = 1.255, ns).
All maternal half-sibling clutches produced fertilized eggs. The mean percentage of fertilized eggs that survived to day 2 was 28 ± 5% (mean ± SE, n = 36 half-sibling clutches). The eggs hatched 7–12 days after fertilization, with 6.3% of the larvae surviving to day 7 and 1.7 % to day 14.
The results of Pearson correlation analysis indicated that there were strong positive correlations between the half-siblings from the same mother in embryo survival (r = 0.927, n = 18, P < 0.001) and also in larval viability to day 7 (r = 0.724, n = 18, P < 0.005), indicating a strong effect of a female on these variables. However, no such correlation was observed in viability to day 14 (r = 0.235, n = 18, ns). Paired tests revealed that there was no difference between males of different ornament categories in embryo survival and larval viability was not influenced by paternal ornamentation (Table 1, Fig. 1).
Table 1. Wilcoxon statistics of size-matched paired data.
Males sorted (within pair) by the breeding tubercle (BT)-ornamentation. ‘*’ indicates that higher-ornamented males had higher values.
Resistance against Rhipidocotyle campanula
Larval viability to day 7
Larval viability to day 14
The mean difference in the resistance of sires, measured as the proportion of dead parasites ± SE, was 0.16 ± 0.03 ranging from 0.001 to 0.47 within a pair of males. Paired tests revealed that the proportion of dead R. campanula of sires was negatively related to the viability of their larvae, but not to survival of their embryos (Table 2, Fig. 2). However, no differences were found in offspring survival (Z = 0.840, n = 18, ns; Z = 0.631, n = 18, ns, days 7 and 14, respectively) between high- and low-resistance sires, when the proportion of dead R. campanula was weighted by the parasite load.
Table 2. Wilcoxon statistics of size-matched paired data.
Males sorted (within pair) by the resistance against Rhipidocotyle campanula. ‘*’ indicates that higher-resistance males had higher values. ‘†’ indicates that higher-resistance males had lower value.
Breeding tubercle (BT)-ornamentation
Larval viability to day 7
Larval viability to day 14
‘Good genes’ hypotheses suggest that male sexual ornamentation could signal indirect benefits to females, and there is empirical evidence to support this (Møller & Alatalo, 1999; Kokko et al., 2003). In the current study, no differences were found with respect to the early survival of eggs and larvae produced by differently ornamented male roach. However, male sexual ornamentation was strongly related to resistance against the digenean parasite, R. campanula. This result about relationship between BT-ornamentation and parasite resistance is similar to our previous findings from the same population in spring 1999 (Taskinen & Kortet, 2002). The relationship we have found between sexual ornamentation and parasite resistance may provide support for ‘good genes’ hypotheses in relation to parasite-mediated sexual selection (Hamilton & Zuk, 1982).
The resistance of a sire against R. campanula was negatively related to the survival of his larvae. This supports the life history theory concept in which certain life history characters are traded-off against others (Roff, 1992; Stearns, 1992). In the stickleback, G. aculeatus offspring of highly ornamented males grew slowly but had higher resistance against parasitic infections than offspring of less ornamented males (Barber et al., 2001). Barber et al. (2001) suggested that increased resistance may involve a cost in terms of reduced growth rate. Our finding about the negative relationship between a sire's parasite resistance and the early viability of his offspring may indicate the presence of such a cost also in roach. Alternatively, that result could be explained by a trade-off that would occur in the adult males between immune function and investment in seminal substances that might increase juvenile survival. The observed negative relationship between sire's resistance and early viability of his offspring may mean that those offspring with innate high levels of resistance may have a lower probability of survival during the first weeks of their lives. Survival in early life might be pleiotropically impaired for other reasons, such as the acquisition of resources for later investment in sexual ornamentation or parasite resistance. Supporting this idea, Brooks (2000) found a negative genetic correlation between a male trait (sexual attractiveness) and survival in guppies, Poecilia reticulata. However, other studies on guppies have found that offspring of large males had higher survival rates (Watt et al., 2001) and higher growth rates (Reynolds & Gross, 1992) than those sired by smaller males.
The present results suggest a strong maternal effect on the early survival of larvae of roach. Similar findings in half-sibling experiments have also been reported by Wedekind et al. (2001) and Barber et al. (2001). However, the variation in the quality of roe between females might not reflect the genetic differences between them, and another experiment should be conducted to assay the genetic maternal effects on offspring viability. Moreover, in the present experimental design the various females were all paired with different males, i.e. the design did not control for male effects and some of the observed maternal effect could be caused by variation among the fathering males.
In whitefish, the late mortality (but not the early mortality) of eggs, from a bacterial infection was correlated with male BTs (Wedekind et al., 2001). However, the developmental time of roach eggs (10–15 days) is much shorter than the developmental time of the eggs of whitefish (60 days at 8 °C, Wedekind et al., 2001), which prevent us from directly comparing our results. Moreover, Barber et al. (2001) found in their maternal half-sibling experiment that the sexual ornamentation of a male stickleback was related positively to his offsprings’ growth rate and parasite resistance, but not to survival.
In the wild, mortality values of early larval stages of freshwater fishes may be extremely high, occasionally over 99%. High mortality may be caused by unfavourable environmental conditions (e.g. temperature, food availability) during spawning and early life, predation, diseases or competition for essential resources (e.g. Braum, 1978; Karjalainen et al., 2000). In our laboratory study with roach, environmental variables were held constant and predation was absent, but mortality was nevertheless relatively high. Water in the incubation containers was changed each day and dead eggs and larvae were removed to minimize the risk of fungal and bacterial infections. Water was aerated, so oxygen deficiencies could not have caused the high mortality. However, possible infections were not treated and conditions were not sterile. Thus, we think that both the water quality and infection may have been responsible for fish mortality. The exact cause of mortality is unknown, but signs of both developmental problems and fungal infection of dead eggs and larvae were observed. According to our knowledge, there are no other studies on survival of the roach larvae in laboratory conditions. It may be possible that our artificial rearing environment provided better conditions for low-quality offspring than for high-quality offspring. Therefore, our result about the possible trade-off between larval mortality and adult parasite-resistance should be carefully interpreted within the constraints of the experimental design. Moreover, further studies are needed to confirm this phenomenon.
In group spawners such as the roach (e.g. Diamond, 1985; Wedekind, 1996), sperm competition may occur (cf. Parker, 1970), and the fertilizing ability of the ejaculate produced by an individual male may play an important role in reproductive success. Roach males do not offer shelters to females, in which to spawn, which may increase the possibility of simultaneous parasitic spawnings. BTs may also signal the male dominance hierarchy (Kortet, 2003), and highly ornamented males produce sperm with higher longevity (Kortet et al., 2004). Thus, we predicted differences in the in vitro embryo survival (including fertilization ability) between males with different degrees of ornamentation. However, we found no such differences in early survival of embryos. Likewise, in the corkwing wrasse, Symphodus melops, no differences in the in vitro fertilization ability were found between differently ornamented males (Uglem et al., 2001).
Our results indicate that the BTs of male roach may not signal quality, when the quality is measured as offspring survival in embryonic and larval stages. However, these results do not rule out the possibility that the offspring of highly ornamented males have better fitness over their total lifespan (for example, in terms of better parasite resistance genes). Moreover, sires having higher resistance to R. campanula had lower early viability of their larvae. Assuming that parasite resistance is heritable, this could mean that those roach offspring with innate high levels of resistance might have a lower probability of survival during the first weeks of their lives.
Authors are grateful to L. Vainio for the assistance in the field and laboratory. Also thank M. Jobling, J. Jokela, H. Kokko, T. Ketola, A. Hedrick and anonymous referees for helpful comments on the manuscript. Authors thank Emil Aaltonen Foundation (to RK), Academy of Finland (to RK and MJR) and Niilo Helander foundation (to JT) for financial support.