Abstract.  1. Community level oak–tannin–insect patterns have been largely unexplored since Paul Feeny's ground‐breaking research. Two hypotheses were tested for Quercus velutina and Q. alba in the Missouri Ozarks: abundance and richness of leaf‐chewing herbivores are negatively correlated with foliar condensed tannin concentrations and variation in condensed tannin concentrations explains variation in herbivore community structure. 2. In 2001, foliar condensed tannins in the understorey and canopy of these two oak species were quantified simultaneously with censuses of herbivores in May, during leaf expansion, and in June and August, when leaves were fully expanded. Thirty‐eight of the 134 species encountered had densities sufficient to be analysed individually (n = 10). Of those, Acronicta increta (Noctuidae) and Attelabus sp. (Curculionidae), both oak specialists, were negatively correlated with condensed tannins in the canopy of Q. alba. One additional specialist, Chionodes pereyra (Gelechiidae), was marginally negatively correlated with condensed tannins in the understorey of Q. velutina. Understorey species richness of May Q. velutina herbivores was negatively correlated with condensed tannins, as were total canopy insect density and species richness of August herbivores on Q. alba. 3. Principal component analysis (PCA) of insect abundances indicated that understorey and canopy Q. velutina and Q. alba had different communities of leaf‐chewing insects. Furthermore, condensed tannin levels contributed significantly to variation in PCA scores for Q. velutina, explaining 25% of the total variation. 4. Overall, these results indicate that specialists were more likely than generalists both to correlate negatively with condensed tannins and to occur in lower tannin habitats; abundance and richness of both early and late season fauna correlated negatively with tannins; and species were more likely to correlate negatively with condensed tannins when feeding on Q. alba than on Q. velutina and when feeding in the canopy than in the understorey. Future studies of tannin–insect interactions should manipulate leaf quality in combination with manipulations of other factors that likely influence community structure.

A widely used method for analyzing hydrolyzable tannins afer reaction with KIO(3) has been modified to include a methanolysis step followed by oxidation with KIO(3). In the new method, hydrolyzable tannins (gallotannins and ellagitannins) are reacted at 85 degrees C for 20 h in methanol/sulfuric acid to quantitatively release methyl gallate. Dried plant samples can be methanolyzed under the same conditions to convert hydrolyzable tannins to methyl gallate. Oxidation of the methyl gallate by KIO(3) at pH 5.5, 30 degrees C, forms a chromophore with lambda(max) 525 nm, which is determined spectrophotometrically. The detection limit of the method is 1.5 microg of methyl gallate, and with plant samples, relative standard deviations of less than 3% were obtained.

Current research suggests that the strength of top‐down forces in communities increases with resource availability to primary producers. We examined the relative impacts of top‐down and bottom‐up forces on oak–herbivore communities by factorial manipulation of predation pressure and plant growth and quality. Plant growth and quality (phenolics, nitrogen) and avian predation intensity were altered through the addition of fertilizer to and the exclusion of birds from Quercus prinus and Q. rubra saplings. We censused the densities of insect herbivore guilds and predaceous arthropods on experimental and control trees in the summers of 1996 and 1997. We assessed changes in oak foliage quality using chemical assays for nitrogen, gallotannins, proanthocyanidins, and foliar astringency (protein‐binding capacity). In general, leaf chewers, phloem feeders, and leaf miners were more frequent on fertilized than on unfertilized trees. Predaceous arthropods were also more frequent on fertilized trees and were positively correlated with herbivore densities on Q. prinus saplings. Bird exclusion by itself had no detectable effects on the presence or absence of herbivores or predaceous arthropods. However, avian predation intensity on phloem feeders on Q. rubra did increase in the presence of nutrient addition to the bottom trophic level in July 1997. Cumulative 1997 yearly densities of herbivores and predaceous arthropods showed similar patterns of increasing avian predation pressure on fertilized Q. rubra saplings. Nonetheless, fertilization and bird exclusion treatments did not alter seasonal defoliation levels. At this early stage in the development of our oak–herbivore community, bottom‐up forces appear to influence the impact of top‐down forces.