Female-limited polymorphism occurs in different animal taxa but is particularly abundant among species of damselflies (Insecta: Odonata), most likely as a consequence of selection to avoid excessive male harassment. Recent work on the damselfly Nehalennia irene indicated that within year spatial variation in female morph frequencies was limited in nearby populations (i.e. intra-regional scale), but large at a continental scale. As anticipated, some of the observed variation in morph frequency was correlated with variation in the estimated degree of male harassment towards female morphs, measured by male density and operational sex ratio. Here, we extended earlier work by quantifying variation in morph frequency over two to three years, allowing us to elucidate how morph frequencies vary temporally at both intra-regional and continental scales (data for 8 populations over three years and for 33 populations over two years, respectively). Annual variation in morph frequencies was relatively high at the intra-regional scale, but was never large enough to obscure the underlying spatial pattern at the continental scale. At both geographic scales, male density and operational sex ratio were highly variable between years. The estimated degree of male harassment correlated with variation in morph frequency within some regions, but not all. Together, the observed natural variation in female morph frequencies may be partly explained by variation in male harassment, but it appears that a complete understanding will require considering the role of other environmental factors.
Sampling procedures and study sites
First, we focussed on variation in morph frequencies on a continental scale, looking for differences in morph frequency between three regional clusters of sites in Western Canada (British Columbia), Central Canada (Ontario) and Eastern Canada (New Brunswick), with clusters separated by a maximum distance of 4000 km (see table 1, fig. 1 ). All sites within a regional cluster were >5 km apart and not part of the same lake or river. For all of these study sites latitude and longitude was determined from topographic maps (scale 1:250 000). Each site was sampled in 2004 and 2007 at comparable sample dates. Th e rational for this was to minimize differences due to seasonal variation in morph frequencies (Bots et al., 2007 ; Van Gossum et al., 2007), although we acknowledge that this approach may not entirely control for between year variations in phenology. An additional site was sampled near Quebec City, between the Central and Eastern region, which was not included in any analysis, but results on temporal variation are reported separately. Second, we determined temporal variation in population parameters on an intra-regional scale, for which eight sites within a regional cluster were sampled at three years: 1992 (see Forbes et al., 1995 ), 2004 and 2007. These populations were all located within 15 km of the Queen’s University Biological Field Station (Chaff ey’s Locks, Ontario, 79°15’W longitude / 44°34’N latitude). Sample dates were again comparable among years; 1992: 6-7 July; 2004: 11-14 July; 2007: 1-5 July.
In both continental and intra-regional surveys, the numbers of male and female N. irene were recorded at the site of reproduction and used to estimate female morph frequency (proportion of andromorphs to total females), population density, male density and operational sex ratio (OSR; proportion of adult males to all adult individuals in the population). Animals were collected by sweeping an insect net within 5m of the water’s edge, where mating of N. irene typically occurs (Forbes et al., 1995 ). We transcribed ‘eight-shaped’ figures with the insect net, while walking slowly through the vegetation. Sex and morph of each individual N. irene netted was noted, and all individuals were marked prior to their release to preclude multiple counts. For the continental survey, sweeps were made while time was recorded. Several sets of sweeps were made per site until at least thirty adult females were caught (Van Gossum et al., 2007). For the continental survey, the number of damselflies caught per unit time spent sampling (individuals/min) was used as an estimate of density. For the intra-regional survey we adhered to the methodology described by Forbes et al. (1995) and made a total of five standardised samples, each sample consisting of eight separate sweeps. An identical number of sweeps was made for each site and year, such that relative densities could be compared directly based on the average counts of individuals in each sweep (individuals/sweep). Furthermore, continental and intra-regional surveys differed in that for the former age class was noted (adult or teneral), while for the latter sampling was restricted to adults. For both surveys we report only results based on adult individuals to allow direct comparison between both methods. Results based on adults and tenerals in the continental survey are only mentioned when these are at variance with analyses based on adults only.