Info

forced through the origin at extreme morph frequencies (Gendron 1987), but it fits most data well and is appropriate to pollinator-plant experiments where apparent morph frequencies do not change.

I have reanalyzed published data sets, using the Greenwood & Elton model where possible. I have found it essential to log-transform data before analysis, due to non-normality of variances. Analyses can be greatly influenced by outliers, so bootstrap estimates of parameters and standard errors are appropriate. With small data sets, power is limited, but I have examined correlations between morph frequency and deviations of observed from expected values.

Bumble bee behavior on single artificial flowers

Both Real (1990) and I (Smithson & Macnair 1996, 1997a, b) explored frequency-dependent behavior by bumble bees using a "bee-board" - a large, rigid, plastic sheet drilled with small wells to hold "nectar." Colored discs representing flowers are placed under selected holes. Flower color, density, positioning, and reward content are easily manipulated, and the experiments are easily replicable. One shortcoming is the lack of inflorescence structure, which may affect behavior. Table 12.1 summarizes results from this method.

Significant frequency dependence emerged in 11 of 13 experiments with flowers of two colors offered to bumble bees at different frequencies (Table 12.1). In 9 of 14 experiments, bees expressed significant preferences for one color. In 6 of the 8 experiments where both colors were equally rewarding, bees preferred the common color. Altering the relative amount ofreward provided by each color did not significantly affect frequency-independent preference, although variances were greatly increased (Smithson 1995). However, a significant change in frequency-independent preference was recorded when the reward provided by one color was more variable than the other (Table 12.1, t = 8.242, p < 0.001). Other authors have found

Fig. 12.1 Opposite. Diagram showing various types of frequency-dependent pollinator preferences amongst two morphs (A and B) and the resultant selection regimes that could be induced in plant populations assuming a simple and positive relationship between preference and plant fitness. Lines a and b contrast different strengths of common morph preference and resultant positive FDS. Lines c, d, and e show the effects of FDS and FIS acting simultaneously. Line f shows FIS only, with line g showing FIS increasing with morph frequency - this could be considered as FDS in a broad sense (see text). Lines h, i, and j show rare-morph preference and resultant negative FDS. The dotted line represents the line of equal preference or fitness for the two morphs, and the dashed line shows 50% morph A.

Table 12.1. Results of experiments testing for frequency-dependent andfrequency-independent choice in artificialflower experiments using bumble bee workers; all experiments used flowers placed on a bee-board, and all experiments contained two flower colors, yellow and blue, which were tested for frequency-dependence (except for Smithson & Macnair 1997a, where three flower colors were present but only two, blue and purple, were tested)

Table 12.1. Results of experiments testing for frequency-dependent andfrequency-independent choice in artificialflower experiments using bumble bee workers; all experiments used flowers placed on a bee-board, and all experiments contained two flower colors, yellow and blue, which were tested for frequency-dependence (except for Smithson & Macnair 1997a, where three flower colors were present but only two, blue and purple, were tested)

Reward"

Reward scheduleb

Frequency dependencec

Frequency independencec

Source

Was this article helpful?

0 0

Post a comment