Summary
Correlations were sought between insect species richness and abundance, and the chemical traits of foliage of 33 species of broad-leaved deciduous trees in southern Ontario. Number of Leipdopteran species collected by the Canadian Forest Insect Survey was not significantly correlated with any of the foliar traits after the effects of sampling intensity on species richness had been removed statistically. Relative abundance of Lepidopteran larvae was significantly related only negatively to the percentage of leaf dry mass as available carbohydrate. The distribution of four abundant, generalist moth species among host trees were considered in detail. Number of col- lections per host was associated with much of the distributional pattern for the fall webworm Hyphantria cunea and spotless webworm H. textor, whereas foliar traits were unimportant. Collecting effort was unrelated to relative abundance of the fall cankerworm Alsophila pometaria and linden looper Erannis tiliaria. For these latter two species relative abundance and a nutritional index, which includes nitrogen, water, lignin and polyphenol content of the leaves, were positively related. Our results suggest that general characteristics of communities of phytophagous insects on broad-leaved deciduous trees are not related to the levels of major nutrient components of the foliage.
Methodology
We use data collected by the Canadian Forest Insect Survey (CFIS) on Lepidopteran larvae associated with 33 species of broad-leaved, deciduous trees in southern Ontario. Karban and Ricklefs (1983) describe the collecting area and sampling methods. The CFIS samples trees in approximate proportion to their abundance, but collecting is often focussed on trees sustaining out- breaks of one or more species of phytophagous insect. Also, a "collection", which may be obtained by beating branches of foliage over a collecting cloth or by hand- picking specimens from vegetation, enumerates the number of each species of insect obtained from one or more individuals of a single species of tree at a particular location or date (McGugan 1958, Harris et al. 1972). Hence collections may be biased in favor of including species that are either gregarious or frequently achieve "outbreak" abundances. All species found are, however, recorded and, because we base our analyses on collections rather than individuals, differences in relative abundance do not introduce bias into the relationship between species richness and sampling intensity. We used CFIS data only from the region between Kinston and Ottawa, Ontario, to provide a small, silvically homogeneous area that included the locality at which foliage was sampled. Most of the species of Lepidoptera are native to Ontario.
Foliage from 33 species of broad-leaved deciduous trees was collected, as described by Ricklefs and Matthew (1982), in the vicinity of Lake Opinicon, Ontario, which is within the area sampled by CFIS and within the natural geographical ranges of all the species of trees except Malus spp. (feral apple), Populus deltoides, and Acer negundo. For each tree species, an index of leaf toughness, mass per unit area, % water, and % of dry mass as available carbohydrate, crude fiber, hemicellulose, holocellulose, lignin, polyphenols, nitrogen, phosphorus and total ash were determined for three sampling periods during 1977 (30 May-9 June, 4-8 July, and 2-4 August) (Ricklefs and Matthew 1982). An analysis of replicated samples demonstrated that variation among tree species was significant (R2, 0.73-0.92) with respect to dry weight per unit area, toughness, and percentages of lipid, available carbohydrate, crude fiber, holocellulose, lignin, polyphenols, and nitrogen. Variation among species was marginally significant for percent water and was not significant for hemicellulose, phosphorus and ash (Ricklefs and Matthew 1982). When species and sampling period were analyzed in an unreplicated two-way ANOVA, species was a highly significant effect for all foliar traits. Because seasonal samples were not collected from the same individuals and because some replicate samples were collected at distances of several km and on different soil types, we conclude that variation in foliar traits among species greatly overshadows that within species, at least at the level of the individual tree, local geographic area, and season. We could not assess the importance of variation within trees with respect to exposure to sun or other factors.