An Assessment of Techniques for Measuring Herbivory: Is Rainforest Defoliation More Intense Than we Thought?

Amounts of leaf material grazed by insect herbivores in Australian rainforest canopies were measured over 3 years, and two methods of assessment were compared. Long-term observations of labelled leaves and shoots measured losses up to five times larger than estimates obtained by the more conventional technique of measuring missing leaf area on single occasions. Long-term observations gave higher but more accurate rates since they incorporated a temporal component to grazing activities and also accounted for leaves totally eaten. Annual leaf area losses of 26, 22, and 14.6 percent were measured in Australian cool temperate, warm temperate, and subtropical rainforests, respectively. Grazing in some plant communities may be higher than previously recorded from measurements of missing-leaf area, resulting in an underestimation of the impact of herbivory. MOST STUDIES OF HERBIVORY in forest communities have estimated grazing by harvesting samples of leaves and estimating proportions of leaf area missing. This method usually indicates leaf area losses of from 3 to 10 percent and is similar in tropical forests (Odum and RuizReyes 1970, Benedict 1976), coral island shrubs (Lowman in press), European beech woods (Nielson 1978), and northern temperate hardwood forests (Bray 1961, 1964; Woodwell and Whittaker 1968, Reichle et al. 1973, and many others). This method-termed discrete since leaves are selected, usually at one point in time, and measured for apparent leaf damage-provides a quick and easy means of estimating herbivory as reflected by partial defoliation of leaves. It may, however, underestimate the real herbivory losses of forest canopies, since it does not account for leaves totally eaten. Two methods of sampling commonly are used: (1) leaf samples removed from litter traps and measured for hole damage (e.g., Odum and Ruiz-Reyes 1970), or (2) leaf samples picked from tree canopies (e.g., Fox and Morrow 1983). Assuming that mature leaves are picked in method 2, then the two sampling techniques are very similar: both measure partial defoliation incurred during the lifespan of a leaf, but the samples include only uneaten or partially eaten leaves and do not account for totally eaten leaves. A more comprehensive method of quantifying herbivory, long-term observations, was developed in this study to incorporate a temporal component into the sampling and also to monitor the number of totally grazed leaves. This technique entails repeated observations and measurements of permanently marked leaves, providing an accurate measure of defoliation and the temporal aspects of grazing activities. In view of the increasing interest in plant-insect relationships, the different methodologies for measuring herbivory are compared in this study for discrepancies that could arise. When similar long-term sampling techniques (albeit only over 3 months) were used in the neotropics (Coley 1983), grazing losses reported were higher (21%) than those in previous studies (7%; Odum and RuizReyes 1970). Similarly, grazing estimates of Australian eucalypts that accounted for leaves totally eaten (although estimated by petiole counts rather than actual observations) were higher than losses measured by sampling (Journet 1981). Discrete samples of coral cay shrubs also revealed lower defoliation losses (2-3%; Lowman, in press) than long-term observations made previously (up to 30%; Heatwole et al. 1981). However, none of these examples featured direct comparisons between leaf samples and long-term observations at the same time and