«C.A. BERGSTROM 1) and T.E. REIMCHEN 2,3) (Department of Biology, University of Victoria, P.O. Box 3020, Victoria, B.C. Canada) Summary Fluctuating ...»
FUNCTIONAL IMPLICATIONS OF FLUCTUATING ASYMMETRY
AMONG ENDEMIC POPULATIONS OF GASTEROSTEUS
C.A. BERGSTROM 1) and T.E. REIMCHEN 2,3)
(Department of Biology, University of Victoria, P.O. Box 3020, Victoria, B.C. Canada)
Fluctuating asymmetry (FA) has been used as a measure of developmental stability across
many taxa, with asymmetric individuals presumed to have reduced tness. FA has also been suggested for use in conservation biology as a measure of the health of populations. Here we assess the suitability of these uses of FA by using a novel measure of asymmetry in the bony lateral plates of threespine stickleback ( Gasterosteus aculeatus) from 60 insular and endemic freshwater populations from the Queen Charlotte Islands. The frequency of asymmetric G. aculeatus individuals among populations varied from 1% to 76% with a mean of 42%. Extreme variation in the frequency of asymmetries among lateral plate positions within samples was also observed. Plates important to the structural integrity of predator defences were least asymmetric, either due to selection against asymmetry at these positions or to variation in the temporal development of the plates. These results emphasize the need for caution when interpreting differential levels of FA among traits in individuals and populations, as the differencesmay be due to variation in the strength or direction of selection for symmetry, and not exclusively to differences in tness.
Keywords: Gasterosteus, uctuating asymmetry, population-wide asymmetry, developmental instability, functional morphology, conservation.
1) E-mail address: email@example.com 2) E-mail address: firstname.lastname@example.org 3) We would like to thank G.E.E. Moodie and P. Hart for helpful comments on this manuscript. Many thanks to S. Douglas for help with the collection of the samples used in this study. This research has been supported by a Natural Sciences and Engineering Council of Canada (A2354) grant awarded to T.E.R. and a King-Platt Fellowship to C.A.B. This study is in partial completion of doctoral dissertation research by C.A.B.
c ® Koninklijke Brill NV, Leiden, 2000 Behaviour 137, 1097-1112 1098 BERGSTROM & REIMCHEN Introduction Quanti cation of the tness of an individual is simultaneously one of the most important tasks in evolutionary ecology and one of the most dif cult. Fitness itself is theoretically simple in de nition, but the practical measurement of such a quality quickly becomes a highly multidimensional and complex task (Endler, 1986). There is need for a simple index of tness that can be measured at one point in time and is relieved of the burden of a complete assessment of the selective factors in an individual’s habitat. Fluctuating asymmetry (FA) has been proposed as such an index of individual tness (Soule’, 1967; Møller, 1994), as it is thought to re ect an organism’s genome-wide ability to buffer against stress during development (Leary et al., 1992). FA is manifested as a population-wide pattern of asymmetry in a bilateral trait which is normally distributed around a mean of zero (Van Valen, 1962; Palmer & Strobeck, 1986). FA is indicative of random errors in the phenotypic development of an organism in response to environmental or genetic stress (Adams & Niswander, 1967; Gest et al., 1986; Clarke & McKenzie, 1992; Imasheva et al., 1997; Campbell et al., 1998).
There is considerable evidence that FA is selected against both in sexually selected traits (Arcese, 1994; Møller, 1994; Swaddle & Cuthill, 1994;
Watson & Thornhill, 1994; Hansen et al., 1999), and in traits important for locomotion (Alexander et al., 1984; Møller, 1991; Balmford et al., 1993). Symmetry in freely moving animals is conceivably the ideal state, as this allows equal between-sides ef ciency in a 3-dimensional, symmetric world where the direction of stimuli is unpredictable (Bradshaw & Rogers, 1993). The symmetry of functional traits used for locomotion or defence would presumably be selected for in a population, and result in reduced numbers of individuals asymmetric at these traits in older age classes.
In this paper we assess FA among natural isolated populations of freshwater threespine stickleback (Gasterosteus aculeatus ) from the Queen Charlotte Islands in northern British Columbia, Canada. Stickleback have a defensive apparatus which is composed of a series of heritable bony lateral plates (Hagen, 1973) located in parallel on both sides of the body, as well as 2 large dorsal spines and 2 large pelvic spines (Fig. 1). Freshwater stickleback from the Queen Charlotte Islands exhibit tremendous variation in the
FA IN FUNCTIONAL TRAITS FROM NATURAL POPULATIONSFig. 1. Schematic of lateral plates on the anterior half of a stickleback possessing plates 1 through 10, showing areas of overlap between lateral plates and spine supports. Lateral plate positions are labelled by number. Modi ed from Reimchen (1983).
number of lateral plates, ranging from none to the full set of approximately 30 plates (Moodie & Reimchen, 1976; Reimchen et al., 1985). The lateral plates are functionally important for survival; protecting the underlying integument of stickleback during predator manipulation, as well as providing structural support for the dorsal and pelvic spines (Reimchen, 1983, 1992a;
review in Reimchen, 1994a).
Several studies have found asymmetries in the number of lateral plates in stickleback (Hagen, 1973; Hagen & Gilbertson, 1973; Moodie & Reimchen, 1976; Moodie & Moodie, 1996), although interpretations of functional implications have been limited. Moodie & Reimchen (1976) found a negative correlation between the degree of plate asymmetry among lakes and the presence of certain predators of stickleback, suggesting that asymmetric stickleback are at a disadvantage as prey. Here, we have greatly expanded the investigation of inter-population variation in asymmetry to include 60 populations, and have also used an additional and more rigorous index of lateral plate asymmetry.
Our rst prediction is that asymmetries in the number and positions of lateral plates will compromise their effectiveness as defensive structures, therefore increasing the individual’s chances of predator-mediated mortality.
This will result in overall low levels of asymmetry in the lateral plates among 1100 BERGSTROM & REIMCHEN natural populations of stickleback, as well as a decrease in relative frequency of asymmetric individuals with age.
Second, we will assess variation in the FA of lateral plates among 60 natural populations in order to estimate a baseline level of developmental stability for this species. Other investigators have assessed FA among populations in the wild in order to compare their relative health under varying levels of stress generated by human disturbance (Wayne et al., 1986; Clarke, 1995; Manning & Chamberlain, 1993; Lens et al., 1999).
However, it is dif cult to interpret differences in FA among populations without rst knowing what baseline levels of FA are found naturally (Palmer, 1996). The bene t of using populations from the Queen Charlotte Islands for this purpose is that the habitats are relatively undisturbed and pristine, therefore giving an estimate of ‘natural’ levels of FA with which to compare to disturbed habitats. This is the rst study to our knowledge that looks at levels of FA among such a large number of wild populations.
Thirdly, we will ascertain whether there is variation in the incidence of asymmetry among lateral plate positions. The assumption that FA is an indication of overall developmental stability predicts that we should nd roughly equal frequencies of asymmetry among the plate positions.
However, other studies have not found high correlations in asymmetry between traits within individuals (Van Valen, 1962; Palmer & Strobeck, 1986; Dufour & Weatherhead, 1996). One explanation for this lack of correlation is that different traits will vary in their susceptibility to stress as a result of different developmental pathways (Møller & Swaddle, 1997).
The lateral plates of stickleback are ideal to test this prediction with, as they provide a series of structures which presumably are under similar developmental constraints and which develop in close temporal proximity of each other. Therefore, variance in developmental processes among plates should be minimal.
Materials and methods Sampling and study area Samples of stickleback from lakes and streams throughout the archipelago were obtained during multiple expeditions between 1975 and 1990. Habitat descriptions and general collecting methods are published elsewhere(Reimchen, 1989, 1992b, 1994b; Reimchen et al., 1985). Brie y, the majority of stickleback samples were collected from April to June using standard-mesh minnow traps placed in the littoral zones. Fish were xed in 10% formalin and stored in 95% ethanol. Sixty of these samples were used for this study.
FA IN FUNCTIONAL TRAITS FROM NATURAL POPULATIONS
Morphometrics were completed on 50 sub-adults and 50 adults randomly selected from each sample, sample size allowing. Morphometrics involved the following traits: standard body length (SL), sex, position of each lateral plate on the left, position of each lateral plate on the right, total plate number on the left and total plate number on the right. Each lateral plate overlies a single myomere, and can be identi ed by a number (P1, P2, P3,..., P30) which is consistent between individuals(Reimchen, 1983).
Measurement error is often disguised as FA, and thus can in ate asymmetry estimates beyond what is actually present (Palmer, 1994). It is crucial to determine if measurement error is equal to, greater than or less than the FA of a trait. Therefore, after initial measurements were complete, 3 localities were randomly chosen and 20 individuals from each were remeasured for plate position. The measurement error of lateral plate position was 0.8% of the total number of plates scored on the 60 sh which were re-measured. Of the total number of lateral plates scored in the 60 samples ( N = 57,487), 4.0% of them were asymmetric for position. In other words, the incidence of asymmetry was 5 times greater than that of the measurement error. The % error did not differ signi cantly among populations (x 2 = 4.86;
df = 2; 0.10 p 0.05). Since this study is a comparative assessment of asymmetry among populations, and the measurement error was similar among populations, we did not remove measurement error from our FA indices. If there is a difference in our FA index among populations, this will be due almost exclusively to a true FA differential since the measurement error is relatively minimal and constant among localities. All measurements were made by one individual to eliminate investigator bias.
Asymmetry was calculated in two ways: plate number asymmetry (PNUM) and plate position asymmetry (PPOS). PNUM was calculated by subtracting the total number of plates on the left from the total number of plates on the right for each individual. This gave a signed asymmetry value for each sh (0 being symmetric), allowing us to test statistically for the presence of FA. We tested whether mean plate number asymmetry (PNUM) differed signi cantly from zero by performing a 2-tailed t-test, and found that none of the 60 populations had signi cant non-zero PNUM means after Bonferroni corrections.
Distributionsof PNUM scores for each population were examined for normality on frequency histograms. Non-normality was detected graphically in 7 of the 60 localities and were thus excluded from further analysis.
In order to calculate plate position asymmetry (PPOS), each plate along the trunk was given an asymmetry score by subtracting its presence (1) or absence (0) on the left side from the right side. Plate positions that had no plates on either side were given a null value for that position. PPOS was then calculated by summing the absolute asymmetry scores for each plate position.
Previous published studies have used PNUM as their only measure of lateral plate asymmetry in stickleback (Hagen, 1973; Hagen & Gilbertson, 1973; Moodie & Reimchen, 1976;
Moodie & Moodie, 1996). PPOS provides a more informative assessment of asymmetry than PNUM. In the context of this study, it allows for the inclusion of sh into the asymmetric group that have a plate present at a designated position on one side but not the other even if total plate number is symmetric (e.g. plates 2 to 7 present on the left, and 3 to 8 on the right).
1102 BERGSTROM & REIMCHEN For this study we considered a sh to be asymmetric if it was asymmetric at one or more plate positions.
We compared PPOS between sexes from each locality using 2-tailed t-tests. None of the differences were signi cant at the p 0.05 level after Bonferroni corrections. The sexes were therefore pooled for further analysis.
The relative frequency of asymmetric individuals was calculated among sampled populations. We also compared the frequency of asymmetric sub-adults and adults at each location to determine if the asymmetric individuals were being selected out of the populations. Some localitiescontained a mixture of sticklebackwith lateral plates and those that were completely naked. The validity of scoring a sh as symmetric for a trait that is absent on both sides is questionable. Twelve of the remaining 53 samples contained at least one naked stickleback, and were therefore excluded from population comparisons of frequencies of asymmetric individuals.
Results Distribution of asymmetries among populations The frequency of stickleback asymmetric for PNUM varied from 1% to 60% among localities (mean = 39%), while the frequency of those asymmetric for PPOS varied from 1% to 76% (mean = 42%; Fig. 2). As expected, the frequency of stickleback asymmetric for PNUM were in all cases equal to or less than the frequency of stickleback asymmetric for PPOS. The majority of the populations contained between 35-50% asymmetric individuals. Mean PPOS was not signi cantly different between sub-adults and adults for any locality at the p 0.05 level after Bonferroni corrections.
Distribution of asymmetries among lateral plate positions
The frequency of asymmetric plates among the 30 possible plate positions for the pooled samples was highly variable ( x 2 = 211.41; df = 29; p 0.001;