Keywords: Multinomial modelling
Seed predators and herbivores can operate as strong selective agents in the evolution of plant defence. In this context, Delpino (1886) posed the "ant-guard" hypothesis to explain the role of extrafloral nectaries on plants. Extrafloral nectaries (EFN), distributed on species in over 80 plant families, occur on vegetative organs and "outer floral parts" not directly associated with pollination. Basically, the hypothesis states that extrafloral nectar production attracts pugnacious "bodyguards" (usually ants) which by their foraging activities deter the activities of herbivorous insects and seed predators.
Since its inception, the ant-guard hypothesis has remained controversial. A few careful studies have experimentally demonstrated that ants attending EFN protect plants (von Wettstein, 1889; Inouye and Taylor, 1979; Schemske, 1980) while several recent studies showed no effect (ODowd and Catchpole, 1983; Tempel, 1983; Boecklen, 1984). ODowd and Catchpole (1983), for example, found that attendance of ants at EFN deterred other insects from developing flowerheads but that their presence decreased neither the numbers of seed predators nor damage to developing flowerheads. The object of this paper is to describe the ant-insect interactions by means of a simple probability model.
Full experimental detail is provided by O'Dowd and Catchpole (1983) but an outline is as follows. The plants studied were helichrysum bracteatum. Three sites were chosen in clearings in the Tallaganda State forest, 40 km. southeast of Canberra, and at each site ten pairs of plants were studied. Plants within each pair were of similar initial size and less than 1 metre apart. Within each pair, ants were excluded from one plant, while the other served as a control. The plants were censused once a week for 17 weeks over the reproductive season (from initiation of flowerheads through the postflowering phase). The data recorded for each plant included the number of flowerheads (capitula), the number of capitula with ants, and the total number of other insects. Different species of ants (predominantly Iridomyrmex spp.) and other insects were observed, but in the data here are pooled within each general category.
To clarify: the first column (Week) lists the week the observation was made, the second (Index) lists the index given to the pair of plants observed, the third (AntCap) is the number of capitula on the plant with ant access, the fourth (ExcCap) is the number of capitula on the plant excluded from ant access, the fifth (Ants) is the number of capitula that have ants present on them, the sixth column (AntIns) is the number of insects on the plant with ant access, and the seventh (ExcIns) is the number of insects on the plant excluded from ant access. Index number 1-10 refer to Site 1, 11-20 to Site 2 and 21-30 to Site 3.
Data file (tab-delimited text)
|Catchpole, E.A., Morgan, B.J.T., O'Dowd, D.J. (1987). Modelling ant-insect interactions on flowerheads of Helichrysum Bracteatum. Biometrics 43, 767-782|