Phenotypic variability within and between fish shoals

Ecology, July, 1996 by Jens Krause, Jean-Guy J. Godin, Rolf Myller

INTRODUCTION

Phenotypic homogeneity among members of social groups is expected from functional models of group formation (Lindstrom and Ranta 1993, Ranta et al. 1994) and has been observed under laboratory conditions (Pitcher et al. 1986, Ranta and Lindstrom 1990, Krause 1994a, Krause and Godin 1994a), but to date reported only anecdotally in field studies (Springer 1957, Parrish 1989). Few studies have systematically investigated and compared the variability of a given morphological or behavioral character within and between free-ranging social groups (e.g., Watkins et al. 1992). Theory predicts that, given a certain amount of phenotypic variability in a population, individuals within groups should be less variable (i.e., more similar in phenotype) than individuals from different groups (Ranta et al. 1994). Here, we quantitatively tested this general hypothesis using fish shoals occurring naturally in the littoral zone of a north temperate lake. First, we ascertained whether sampled fish species or fish size classes were ecologically (and therefore spatially) segregated by relating observed distributions of fish to certain habitat characteristics. Second, we attempted to identify the phenotypic factors that characterize the composition of fish shoals.

METHODS

Fish were collected in Morice Lake (near Sackville, New Brunswick, Canada; 45 [degrees] 55[minutes] N, 64 [degrees] 21[minutes] W) with a 5-m beach seine (mesh size: 3 x 2 mm) between 6 and 25 September 1994. Morice Lake is a small ([approximately equal to]130 ha) impounded lake. The study area (on the southeast side of the lake) consisted of [approximately equal to]30 m shoreline and up to 7 m offshore. This part of the littoral zone was shallow throughout (water depth [less than]55 cm) and included a patch of emergent reeds and submerged vegetation (mainly algae). The substratum was variable and mainly consisted of gravel, sand, and silt.

The same net was used for catching all shoals to standardize the collection procedure. Shoals were located visually while slowly wading through the water. Each shoal was caught by carefully encircling it with the seine, which was then lifted out of the water and carried onto shore. Care was taken not to allow any fish belonging to the same shoal to escape nor to include any individuals that did not belong to the shoal being collected. Because the water was shallow and very clear, we could reliably determine whether or not we had caught the entire shoal. We analyzed only shoals that were caught in their entirety.

For each shoal member, we identified the species, measured its standard body length (from the snout to the beginning of the caudal fin) to the nearest millimetre and counted the number of black spots visible through its skin. Fish body size was always determined by measuring standard length of the fish. These black spots represent cysts containing metacercariae of the parasitic trematode worm Crassiphiala bulboglossa that are embedded in the skeletal musculature and fins (Krause and Godin 1994b). We chose the presence or absence of parasitism (due to C. bulboglossa) and per capita parasite load as phenotypic characters potentially underlying assortative shoaling, because C. bulboglossa is externally visible, one of the main fish parasites in Morice Lake, and is known to influence fish shoaling behavior and shoal choice (Krause and Godin 1994b, Krause and Godin, in press). We counted the black cysts individually up to 30, and estimated their numbers in categories of tens (40, 50, etc.) when they exceeded 30 on individual fish. After the fish were measured and screened for parasites, they were preserved in 10% formalin. Collected fish were not returned to the lake, thereby avoiding problems associated with pseudo-replication (re-sampling the same fish). The fish that were collected represent a relatively small proportion of the population in the study area, and it is likely that new fish immigrated from other sites more or less continuously.

Sampling was carried out between 0900 and 1600. Water temperature was 17 [degrees] [ or -] 1 [degree] C (measured in mid-water). We also recorded water depth and the distance from the nearest aquatic vegetation at the location of each shoal capture. A total of 34 shoals was collected. Repeatability of measurements of fish body length was determined by randomly selecting 100 fish from our collection and independently measuring each individual twice using the same method. The resulting error (paired t test: n = 100, mean difference [ or -] 1 SD = 0.011 [ or -] 0.096 mm, df = 99, t = 1.14, P [greater than] 0.26) between the first and second measurements for individual fish was well below the level at which body length was measured (i.e. the nearest 1 mm) to investigate size-assortativeness.

The first analysis is designed to identify the primary sources of assortativeness in the shoal samples using a contingency table analysis (G test, Sokal and Rohlf 1981). Data on body size and per capita parasite load were first converted to categories using the following category boundaries; body size (represented by x): x [less than] 12 mm, 12 [less than] x [less than] 16 mm, 16 [less than] x [less than] 20 mm, 20 [less than] x [less than] 24 mm, 24 mm [less than] x; parasite load (y): y [less than] 10 cysts per fish, 10 [less than] y [less than] 50, 50 [less than] y. A four-dimensional contingency table comprising the four factors or categories, shoal (Sh), species (Sp), body size (s), and parasite load (P), was then constructed. We tested for each factor separately whether or not a significant assortative influence of Sp, S, and P existed, after the effects of the other two factors (among these three) had been included. This was done by fitting log-linear models (McCullagh and Nelder 1989) relating the frequency of occurrence in each cell (of the contingency table) to all possible combinations of these four factors and their interactions. This analysis was restricted to the three most abundant species (comprising 99.7% of fish collected) and the 26 shoals that contained [greater than] 10 individuals (to avoid artefacts).