Behavioral Ecology Vol. 14 No. 5: 602-606
© 2003 International Society for Behavioral Ecology
The cost of aggregation: juvenile salmon avoid sharing winter refuges with siblings
a Cardiff School of Biosciences, Main University Building, PO Box 915, Cardiff CF10 3TL, UK b F.R.S. Freshwater Laboratory, Faskally, Pitlochry, Perthshire PH16 5LB, UK c Fish Biology Group, Division of Environmental and Evolutionary Biology, Institute of Biomedical and Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow G12 8QQ, UK
Address correspondence to S.W. Griffiths. E-mail: griffithssw{at}cardiff.ac.uk.
Received 15 October 2001; revised 17 September 2002; accepted 2 October 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Kin selection advantages are usually accrued by individuals that associate with close relatives. But aggregation may also be costly, by increasing the risk of predation or resource competition, for example. As a result, individuals should increase their inclusive fitness by trading the costs and benefits of kin association and aggregation. Studies of kin selection to date have focused on situations where there is ample opportunity for kin-biased behavior and therefore for the formation of kin groups. Here we used juvenile Atlantic salmon to test an alternative strategy: that under conditions where the potential for kin-biased behavior is negligible, individuals should, when aggregating, avoid rather than associate with kin to avoid imposing the costs of aggregation upon close relatives. By testing salmon during winter, when juveniles shelter inactively in streambed refuges, we tested whether individuals associate with or avoid their siblings at a time when the opportunity for kin-directed behaviors is restricted. Our results provide the first evidence of kin avoidance in nonreproductive animals studied under semi-natural conditions.
Key words: aggregation, kin avoidance, kin selection, nocturnal sheltering behavior, Salmo salar, winter refugia.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The theory of kin selection is based on Hamilton's model for the evolution of social behavior (Hamilton, 1964
), which states that an individual can potentially increase its inclusive fitness by behaving in such a way as to enhance the reproductive success of its relatives. This is because individuals have many genes in common with their close relatives and can influence the chances of these genes being propagated to the next generation not only through their own offspring but also through those of their kin. A prerequisite for individuals accruing advantages through kin selection is interaction among relatives.
Usually animals that show any kin bias in spacing patterns aggregate with their relatives (Fletcher and Mitchener, 1987; Hepper, 1991). For example, the tadpoles of many amphibians aggregate in kin groups (Blaustein and Waldman, 1992), as do many invertebrates (Fellowes, 1998), and birds and mammals (Greenwood, 1980). There is evidence for kin selection even within groups of highly territorial animals competing for space. For example, sibling groups of juvenile salmonid fishes held in uniform laboratory environments during the summer are less aggressive than groups of non-kin and as a consequence grow faster and achieve higher densities (Brown and Brown, 1993; Olsén et al., 1996).
Kin selection is presumed to operate in each of these examples due to behaviors that increase the individual's inclusive fitness and that can occur because of aggregation among relatives. However, aggregation may also have costs, such as increased competition for local resources and exposure to predators. If, under circumstances where some animals within populations have no choice but to aggregate, and benefits of kin association outweigh costs of aggregation, then clusters of kin would be expected. However, if aggregation is costly and opportunities for kin-biased beneficial behaviors are negligible, then inclusive fitness may be enhanced by kin actively avoiding one another. We tested this possibility by observing the distributions of juvenile Atlantic salmon (Salmo salar) during winter.
During winter juvenile Atlantic salmon (parr) switch from being active throughout both day and night to being predominantly nocturnal (Chapman and Bjorn 1969; Fraser et al., 1993, 1995). The switch is temperature dependent so that below 10°C fish limit most activity such as foraging at night and increasingly seek shelter in small streambed crevices during the day (Cunjak, 1988; Riddell and Leggett, 1981; Smirnov et al., 1976).
Sheltering behavior is thought to be a strategy for reducing the risk of predation (Metcalfe et al., 1999; Valdimarsson and Metcalfe, 1998). Shelter use by salmon parr is strongly density dependent, with a smaller percentage of the fish using available shelters when at high densities (Armstrong and Griffiths, 2001). This may be because sheltering fish rely on crypsis, and the likelihood of a group of fish in a refuge being discovered by a predator may increase disproportionately with the size of the group. Aggressive competition for shelters occurs predominantly when fish are seeking refuge at dawn (Gregory and Griffith, 1996). However, once in the shelter, fish are inactive, and there appears to be little potential for behaviors that might benefit kin over non-kin.
Consider a juvenile salmon faced with the choice between two shelters, one of which is already occupied by a relative and the other by an unrelated fish. Because aggregation is costly, the focal individual's inclusive fitness is greatest if it shares with the unrelated fish, since by doing so it does not inflict the cost of sharing on its kin. This yields two predictions: fish should tend to avoid sharing shelters with each other, and this avoidance should be stronger between closely related individuals.
We tested these predictions using a population of wild juvenile salmon housed in a series of enclosures in a glass-sided stream, which allowed us to observe marked individual fish of known relatedness within their under-gravel shelters. Our results provide the first demonstration of kin avoidance in nonreproductive animals studied under natural conditions.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Experimental animals
The experiment required separate groups of fish that were raised apart and that were either related or unrelated. Kin groups were produced by harvesting the eggs and milt from wild adult Atlantic salmon that were caught at the River Braan, Perthshire, Scotland. The eggs of one female were fertilized with the milt of one male to produce a family of full-sibling offspring. This process was repeated to produce two families, and each family was split into two groups after the eggs were water hardened. Eggs in the four groups were incubated (at the FRS Freshwater Laboratory hatchery, Almondbank, Perthshire) separately with no chance of cross-contamination of water supplies.
We released the resulting fry between 17 and 21 April 1998, at the stage before they had started to feed exogenously, into tributaries of the River Braan that did not contain salmon due to an impassable waterfall. Each family was released into a separate stream (the Dullator and Glenfender streams) and the two groups comprising each family were released at different locations (>100 m apart) within their respective streams. We chose this distance between locations to ensure low interference between groups from dispersing fish based on models of existing data (Crisp, 1995; Martin-Smith K and Armstrong JD, unpublished data). It is therefore extremely unlikely that salmon from one release location would become familiar with juvenile salmon from any of the other release locations. At each location, we released 250 fry at a density of 5 fish/m2. On 22 September 1998, 20 fish from each of the four groups were recaptured by electric fishing and transported to the FRS Freshwater Laboratory, Pitlochry, in separate containers of oxygenated water. Each of the four groups was housed in a separate 250-l tank so that fish from each site were visually and chemically isolated from one another. This procedure allowed us to compare the behavior of unfamiliar, wild-reared fish that were either related (from the same stream, but different release locations) or unrelated (different streams).
We placed numerous rocks and pebbles within each holding tank to increase the heterogeneity of the environment. Live chironomid larvae (blood worm) were provided as food daily ad libitum by placing the larvae in a fine-mesh net suspended over the surface of the water. The larvae wriggled through the net slowly, so that over a period of many hours the fish were provided with a source of natural food.
Test aquaria
The experiment was conducted at the FRS Almondbank laboratory. A large indoor stream facilitated the observation of fish behavior in near-natural conditions of temperature and flow. The stream consisted of a glass-sided tank (80 m long x 2 m wide x 2 m deep) through which water from the River Almond passed. A length of the stream channel was divided by mesh screens into 10 test arenas, each measuring 1.8 m long x 0.6 m wide x 0.27 m water depth. The test arenas were arranged in sequence so that, for example, arena 2 was positioned downstream of arena 1, but upstream of arena 3, and so on. Marks were drawn on the glass of each section to denote 30-cm length divisions. We placed pieces of gravel into the stream channel to form the streambed substratum. The substratum was formed from a base layer (10 cm depth) of coarse gravel (5 cm diam) and an upper layer (5 cm depth) of fine gravel (1.5 cm diam). The mean velocity of water in the center of each test arena was (mean ± SE) 0.114 ± 0.01 m/s, n = 10 at the water surface, and 0.103 ± 0.01 m/s, n = 10 at the streambed.
Four shelters were embedded equidistant from one another into the gravel streambed of each test arena. The shelters were made from plastic bottles cut lengthwise in half in order to create cavities measuring 17 cm x 9 cm x 4 cm deep. Fish entered each shelter through a small opening (2.5 x 3 cm wide) that was level with the gravel surface. The bottles were punctured and opaque, as juvenile salmon are known to prefer dark shelters through which water can flow (Valdimarsson and Metcalfe, 1998). We positioned the open side of each bottle against the glass so that we could observe fish within each shelter. The upper layer of fine gravel used in forming the streambed discouraged fish from hiding in interstitial spaces other than those provided by the shelters. A series of glass windows above the length of the stream allowed observations to be made under natural light conditions, with additional illumination provided from 0700 to 1900 h by 11 lights (400 W each) positioned 1.8 m above the gravel surface.
Experimental procedure
At the beginning of the experiment (16 February 1999) each fish was weighed, measured, and tagged with an individually identifiable passive integrated transponder (PIT) tag (Prentice et al., 1990). Each PIT tag weighed 0.1 g, which is less than 3% of the total weight of the fish at the time of implantation. The lengths and weights of the fish did not vary between families (one-way ANOVA on fork length: F1,39 = 0.96, p =.328 and wet weight: F1,39 = 1.46, p =.235). The mean fork length and wet weight (mean ± SE, n = 44) was 72.27 ± 1.0 mm and 3.58 ± 0.19 g, respectively. We placed four length-matched juvenile salmon (one from each of the separate groups collected from the wild, so that all fish were unfamiliar with each other) into each of the test arenas. The fish were allowed to acclimate for 5 days before observations on shelter use began on day 6. Previous work suggests a settling time of about 4 or 5 days, during which groups of juvenile salmon in both summer and winter gradually establish prolonged patterns of space use (Armstrong et al., 1997; Armstrong and Griffiths, 2001), presumably as they develop dominance hierarchies.
We made scan sample observations on each of the 10 groups every 2 h from 0530 to 1130 h and from 1530 to 2330 h over 3 consecutive days (days 6–8). Observations were made from within a dark hide to avoid disturbing the fish. We used a hand-held PIT detector to identify which fish occupied each shelter during the day. The proximity of the shelters to the glass, and thus of the sheltering fish to the PIT detector, allowed individual identification of all fish. At night, when fish were located on the streambed, we recorded the x–y coordinates of each individual to the nearest 5 cm. During the night we identified fish individually using naturally occurring marks on their bodies and fins. A red-light torch allowed the fish to be observed without being startled. For those fish swimming in the water column, distance above the streambed was also noted.
Live chironomid larvae were provided in excess during the day and night, using a belt feeder that dropped small containers of chironomids into a fine-mesh net suspended over the surface of the water (4 containers per 24 h). This created a small but steady supply of drifting food. Mean (±SE) water temperature during the experiment was 3.6 ± 0.4°C, varying from a maximum of 6.3°C to a minimum of 1.5°C.
Statistical analysis
Proportional data were normalized by arcsine square-root transformation. We calculated the mean distance between pairs of related and unrelated juvenile salmon for each group at 2330 h each day (midpoint of observations made during darkness). We calculated the likelihood of parr occurring at densities of 1–4 in shelters by considering all possible unique combinations of fish and shelters, each fish distributed at random, and totaling the numbers of occasions that each density arose. The first fish to choose a shelter has a free choice. The probability of it picking an unoccupied shelter at random is 1.00 because all shelters are unoccupied. For the next fish, this probability is 0.75 (3 out of 4), for the next one it is 0.5, and for the final fish it is 0.25. Because the fish are assumed to choose independently, we multiplied these four probabilities together to get 9.38%. The binomial solution, for 4 fish and 4 shelters, is that 9.38, 56.25, 14.06, 18.75 and 1.56 % of observations should be of all 4, 3 (with one shared by 2 fish), 2 (with both shared by two fish), 2 (with one shared by 3 fish) and 1 (shared by all fish) shelter(s) being used, respectively. To test whether choice of shelter mates was influenced by kinship, family 1 (salmon obtained from the Glenfender Stream) was chosen a priori as the focal family. We noted the first incidence on day 6 (when use of shelters had stabilized; see below) of a fish from family 1 sharing a shelter, and recorded the relatedness of its shelter mate (kin or non-kin). Next, when the second fish from family 1 was observed to share a shelter, the relatedness of its shelter mate was also noted. This was repeated for all other possible members of this family. As the occurrence of shelter sharing was low throughout the experiment, this method of analysis allowed us to maximize the use of available data without biasing the results or losing independence of data. Where G tests were used, the G values were adjusted using a Williams's correction factor (Sokal and Rohlf, 1995).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Temporal patterns of shelter use
The number of fish sheltering in streambed refuges varied significantly with time of day (repeated-measures ANOVA with replicate arena as subjects, based on total number sheltering per test arena per scan sample observation, F8,89 = 72.75, p =.001). As expected, fish tended to use shelters during the day, leaving them to enter the water column during darkness. Thus, the mean number of fish sheltering on day 6 (± SE, n = 10 arenas) was 1.3 ± 0.2 at 2330 h (midpoint of observations made during darkness), and 3.9 ± 0.1 at 1530 h (midpoint of daylight observations).
For observations made during daylight, a repeated-measures ANOVA (with replicate arena as subjects) tested the influence of day (days elapsed since fish were introduced into the laboratory stream) and time (3 observations per day at 0930, 1130, and 1530 h) on the proportion of observations of sheltering salmon that were of fish sharing a shelter. Neither time (nested within day; F6,89 = 0.07, p =.99) nor day (F2,89 = 1.46, p =.24) had an influence on occurrence of shelter sharing. The proportion of times that fish were observed sharing shelters with one or more conspecifics, calculated as a mean (±SE, n = 10 arenas) of data collected on days 6,7, and 8, was 15% ± 6.31 at 1530 h. Further analyses (below) were therefore conducted on data collected on day 6 at 1530 h (after 5 days acclimation), unless otherwise indicated.
Shelter sharing
Sufficient shelters were provided to allow each fish to shelter in isolation (1 fish/shelter). To determine whether fish shared shelters more often than expected by chance alone, the observed frequencies of the following possible shelter-use configurations were noted at 1530 h on day 6: 1+1+1+1 (i.e., all four shelters used, with one fish in each); 2+1+1 (three shelters used); 2+2 (two shelters used); 3+1 (two shelters used), and 4 (one shelter used). These observed frequencies differed significantly from those predicted by a binomial distribution (G test of goodness of fit, pooling observations of 2+2, 3+1 and 4 configurations with 2+1+1 data because the expected frequencies of the former were small: G1 = 11.45, p <.005). Fish were found alone more often than expected (40% vs. 9.38%), upholding our first prediction that fish preferred not to share shelters.
Shelter preference and fidelity
The observed frequencies with which each position of shelter in an arena was occupied did not vary from those expected by chance (G test of goodness of fit: G3 = 1.11, p >.25). However, there was evidence that individual fish had a preferred shelter. We tested this by first defining the preferred shelter of one randomly chosen focal fish per replicate as the shelter that it occupied on day 6 at 1530 h. We then calculated the proportion of scan sample observations made at 1530 h on subsequent days where the focal fish was sheltering within the preferred shelter. Juvenile salmon returned to their preferred shelter on 76.7% of occasions, which is more than would be predicted by chance, as indicated by the significant departure from the null hypothesis of no preference (25% use of each shelter; one-sample t test based on transformed data, t10 = 5.12, p =.004).
Do salmon prefer or avoid kin?
As each test arena within the indoor stream contained two fish from each of two families, the expected probability that a second fish entering a shelter would be related to the first (resident) fish was 1 in 3, whereas the probability of it being unrelated was 2 in 3. The observed frequency of sharing with kin was significantly different from these expected frequencies (G test of goodness of fit: G1 = 10.154, p <.005). This significant departure from random associations was obtained regardless of which family of salmon was taken as the focal family. Juvenile salmon clearly avoided kin when they shared shelters and preferred to associate with unrelated conspecifics (Figure 1).
|
For salmon moving between different shelters on consecutive days (between days 6 and 7 and days 7 and 8), 7 fish moved into shelters already occupied by kin, and 18 fish moved into shelters already occupied by non-kin, a ratio not significantly different from that expected on the basis of chance (G test of goodness of fit: G1 = 0.290, p >.50), where the expected frequencies of moving into shelters occupied by kin and non-kin were 0.33 x 25 = 8.25 and 0.67 x 25 = 16.75, respectively. These data suggest that fish did not avoid entering a shelter occupied by a close relative. Resident fish always (7 of 7 occasions) left a refuge if a sibling entered it, whereas they only left on 13 of 18 occasions if the intruder was unrelated. A G test of independence found that these ratios were not significantly different from one another (G1 = 3.293, p >.05), although this nonsignificant trend suggested higher dispersal among kin compared to non-kin.
There was no significant difference in the mean distance between pairs of kin and non-kin (calculated as means of four possible pair combinations) during the night (two-tailed t test, day 6 at 2330 h: t9 = 0.66, p =.53; Figure 2).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
We have shown that juvenile Atlantic salmon actively avoid their kin when sheltering in streambed refuges. Salmon tended to shelter singly, but when sharing of refuges did occur (on 29% of occasions), siblings were never observed to shelter together. Both results support our predictions. The significant tendency to avoid sharing with other fish indicates that there is some cost to aggregating, presumably because groups of hiding fish are more likely to be detected by predators using olfactory or visual cues than are solitary fish. Fish seeking refuge should therefore not inflict this cost of sharing on their relatives. When animals have been observed to avoid their relatives, the behavior has been explained by inbreeding avoidance (Bateson, 1979
, 1980) or reduced competition between parents and offspring (Holmes and Sherman, 1982; Hurst and Barnard, 1991). However, these explanations do not apply to the situation we examined because the fish were immature and were all the same age. We believe this is the first demonstration of active kin avoidance that can be ascribed to costs of aggregation exceeding benefits of association in a natural ecological context.
While there are documented examples of invertebrates in simple culture conditions dispersing more actively away from kin than non-kin (e.g., the mango shield scale, Milviscutulus mangiferae: Kasuya, 2000; the flour beetle, Tribolium castaneum: Jasienski et al., 1988), the interactions between these behaviors and the natural ecology of the populations is not clear. Moreover, the situation of dispersal is fundamentally different from the case of a static population where avoidance is on a much more local scale.
Although there was a tendency for fish to show some fidelity to shelters from one day to the next, movements between shelters occurred throughout the period of observation. This continued movement probably reflects exploration of the environment. Exploration would be expected because in rivers the quality of refuges during winter is likely to vary from day to day as areas become blocked with, or free from, silt and ice or as spates cause movement of the substratum. The opportunity to sample shelters probably depends on whether they are vacant and whether fish occupying them can easily be displaced. Competition between juvenile rainbow trout for shelter only occurred for a short period near dawn when fish were returning to shelters after foraging at night (Gregory and Griffith, 1996). Similarly, salmon parr tended not to move between shelters after dawn, presumably because moving after dawn would incur a high risk of predation in natural rivers. However, fish were sometimes recorded in shelters at night and would then have had opportunities to sample refuge options, albeit not under the conditions when shelter would have been of most importance.
Our data show that salmon did not show any kin discrimination when entering a refuge already occupied by another fish. Rather, the avoidance of kin may have arisen from the nonsignificant trend toward being more likely to leave a refuge if an entering fish was a sibling. This may be because fish only enter an occupied refuge if they perceive the occupier to be subordinate to them and/or it is only possible for them to determine their degree of relatedness after having entered a refuge due to the concentration of olfactory cues (see below).
During the night, when salmon emerged from streambed shelters to feed in the water column, kin discrimination was not evident. Salmon did not avoid or associate with siblings any more than unrelated fish. One possible explanation for this finding is that the ability of salmon to discriminate kin may depend on the availability and concentration of waterborne odor cues. Salmonids use waterborne chemical cues to distinguish kin (Stabell, 1982). However, although chemical communication allows the transfer of detailed information over long distances, chemical cues are also likely to be dispersed quickly in fast-flowing water. Therefore, it is likely that individual fish have difficulty in ascertaining the origin of any particular odor. Concentration of odor cues is known to affect kin discrimination in fish (Hiscock and Brown, 2000; Steck et al., 1999). An increase in the concentration of kin odor increases the accuracy with which salmon and sticklebacks are able to distinguish water scented with kin versus unscented water (Hiscock and Brown, 2000; Steck et al., 1999). Furthermore, a recent experiment showed that kin discrimination was observed in Atlantic salmon only when water was recirculated repeatedly past a pair of test fish, but kin discrimination was not observed when water flowed past the fish only once before flowing to waste (Griffiths and Armstrong, 2000). It is possible that the concentration of odor cues is not strong enough to allow fish in the natural conditions of fast-flowing water above the streambed to recognize their kin, whereas the increased concentration of odor in slow-flowing eddies within refuges provides ample opportunity for fish to identify shelter mates.
This study demonstrates that kin-biased behaviors may include local-scale avoidance as well as aggregation of relatives. Given that the strength of kin association may vary spatially depending on water flow characteristics (Griffiths and Armstrong, 2000) and that there may be temperature-dependent changes in temporal patterns of competition for different resources (Fraser et al., 1993), it is likely that highly complex interactions of processes affect the distribution of kin in natural habitats. This may explain, at least to some extent, why, despite the apparent advantages of kin association in simple laboratory environments (Brown and Brown, 1996), genetic analyses of wild populations have failed to find a high incidence of kin association (Fontaine and Dodson, 1999; Garant et al., 2000; Mjølnerød et al., 1999).
|
ACKNOWLEDGEMENTS |
|---|
We thank Felicity Huntingford, Tom Quinn, Graeme Ruxton, Neil Trudgill, and the staff of the F.R.S. Freshwater Lab for helpful comments and advice. We also thank three anonymous referees for their constructive comments. S.W.G. received financial support from a Natural Environment Research Council Postdoctoral Freshwater Biology Fellowship.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Armstrong JD, Braithwaite VA, Huntingford FA, 1997. Spatial strategies of wild Atlantic salmon parr: exploration and settlement in unfamiliar areas. J Anim Ecol 66: 203-211.
Armstrong JD, Griffiths SW, 2001. Density-dependent refuge use among over-wintering wild Atlantic salmon juveniles. J Fish Biol 58: 1524-1530. [CrossRef]
Bateson P, 1979. How do sensitive periods arise and what are they for? Anim Behav 27: 470-486. [CrossRef]
Bateson P, 1980. Optimal outbreeding and the development of sexual preferences in the Japanese quail. Z Tierpsychol 53: 231-244. [Web of Science]
Blaustein AR, Waldman B, 1992. Kin recognition in anuran amphibians. Anim Behav 44: 207-221. [CrossRef]
Brown GE, Brown JA, 1993. Do kin always make better neighbours?: the effects of territory quality. Behav Ecol Sociobiol 33: 225-231. [CrossRef][Web of Science]
Brown GE, Brown JA, 1996. Kin discrimination in salmonids. Rev Fish Biol Fish 6: 201-219.
Chapman D, Bjorn T, 1969. Distribution of salmonids in streams, with special reference to food and feeding. In: Proceedings of a symposium on salmon and trout in streams (Northcote T, ed). Vancouver: University of British Columbia; 153–176.
Crisp DT, 1995. Dispersal and growth rate of 0-group salmon (Salmo salar L.) from point-stocking together with some information from scatter stocking. Ecol Freshwater Fish 4: 1-8.
Cunjak RA, 1988. Behaviour and microhabitat of young Atlantic salmon (Salmo salar) during winter. Can J Fish Aquat Sci 45: 2156-2160.
Fellowes MDE, 1998. Do non-social insects get the (kin) recognition they deserve? Ecol Entomol 23: 223-227.
Fletcher DJC, Mitchener CD, 1987. Kin recognition in animals. Chichester: Wiley.
Fontaine P-M, Dodson JJ, 1999. An analysis of the distribution of juvenile Atlantic salmon (Salmo salar) in nature as a function of relatedness using microsatellites. Mol Ecol 8: 189-198.
Fraser NHC, Heggenes J, Metcalfe NB, Thorpe JE, 1995. Low summer temperatures cause juvenile Atlantic salmon to become nocturnal. Can J Zool 73: 446-451. [CrossRef]
Fraser NHC, Metcalfe NB, Thorpe JE,
1993. Temperature-dependent switch between diurnal and nocturnal foraging in salmon.
Proc R Soc Lond B
252:
135-139.
Garant D, Dodson JJ, Bernatchez L, 2000. Ecological determinants and temporal stability of the within-river population structure in Atlantic salmon (Salmo salar L.). Mol Ecol 9: 615-628. [CrossRef][Medline]
Greenwood PJ, 1980. Mating systems, philopatry and dispersal in birds and mammals. Anim Behav 28: 1140-1162. [CrossRef]
Gregory JS, Griffith JS, 1996. Aggressive behaviour of underyearling rainbow trout in simulated winter concealment habitat. J Fish Biol 49: 237-245. [CrossRef]
Griffiths S, Armstrong J, 2000. Differential responses of kin and nonkin salmon to patterns of water flow: does recirculation influence aggression? Anim Behav 59: 1019-1023. [CrossRef][Web of Science][Medline]
Hamilton WD, 1964. The genetical evolution of social behaviour. J Theor Biol 7: 1-52. [CrossRef][Web of Science][Medline]
Hepper PG, 1991. Kin recognition. Cambridge: Cambridge University Press.
Hiscock MJ, Brown JA, 2000. Kin discrimination in juvenile brook trout (Salvelinus fontalis) and the effect of odour concentration on kin preferences. Can J Zool 78: 278-282. [CrossRef][Web of Science]
Holmes WG, Sherman PW, 1982. The ontogeny of kin recognition in two species of ground squirrels. Am Zool 22: 491-517.
Hurst JL, Barnard CJ, 1991. Kinship and social behavior in wild mice: effect of social group membership and relatedness on the responses of dominant males toward juveniles. Behav Ecol 3: 196-206. [CrossRef][Web of Science]
Jasienski M, Korzeniak U, Lomnicki A, 1988. Ecology of kin and nonkin larval interactions in Tribolium beetles. Behav Ecol Sociobiol 22: 277-284. [CrossRef][Web of Science]
Kasuya E, 2000. Kin-biased dispersal behaviour in the mango shield scale, Milviscutulus mangiferae. Anim Behav 59: 629-632. [CrossRef][Web of Science][Medline]
Metcalfe NB, Fraser NHC, Burns MD, 1999. Food availability and the nocturnal vs diurnal foraging trade-off in juvenile salmon. J Anim Ecol 68: 371-381. [CrossRef]
Mjølnerød IB, Refseth UH, Hindar K, 1999. Spatial association of genetically similar Atlantic salmon juveniles and sex bias in spatial patterns in river. J Fish Biol 55: 1-8.
Olsén KH, Järvi T, Löf A-C,
1996. Aggressiveness and kinship in brown trout.
Behav Ecol
7:
445-450.
Prentice EF, Flagg TA, McCutcheon CS, 1990. Feasibility of using implantable passive integrated transponder (PIT) tags in salmonids. In: Fish-marking techniques (Parker NC, Giorgi AE, Heidinger RC, Jester JD, Prince BED, Winans GA, eds). Seattle: University of Washington; 7:317–322.
Riddell BE, Leggett WC, 1981. Evidence of an adaptive basis for geographic variation of body morphology and time of downstream migration of juvenile Atlantic salmon (Salmo salar). Can J Fish Aquat Sci 38: 308-320.
Smirnov YA, Shastov YA, Khrennikov VV, 1976. On the behaviour and feeding of juvenile Onega salmon (Salmo salar morph sebago) in winter. J Ichthyol 16: 503-506.
Sokal RR, Rohlf FJ, 1995. Biometry: the principles and practice of statistics in biological research, 3rd ed. New York: Freeman.
Stabell OB, 1982. Detection of natural odorants by Atlantic salmon parr using positive rheotaxis olfactometry. In: Proceedings of the Salmon and Trout Migratory Behavior Symposium, June 1981 (Brannon EL, Salo EO, eds). Seattle: University of Washington; 238–246.
Steck N, Wedekind C, Milinski M,
1999. No sibling odor preference in juvenile three-spined sticklebacks.
Behav Ecol
10:
493-497.
Valdimarsson SK, Metcalfe NB, 1998. Shelter selection in juvenile Atlantic salmon, or why do salmon seek shelter in winter? J Fish Biol 52: 42-49.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||