322 
Psyche 
[Vol. 88 
In this paper we use data for insects foraging on plantings of 
commercial sunflowers ( Helianthus annuus L.) and onions {Allium 
cepa L.) to test these hypotheses. 
An additional question of interest is whether bee species differ in 
their distribution across flowers. For example, Benest (1976) has 
suggested that honeybees {Apis mellifera L.) are more tolerant of 
joint foraging than are bumblebees {Bombus sp.) and Kalmus (1954) 
reported that honeybees tend to form clusters at artificial feeding 
sites. Group foraging, leading to clumped distributions on flowers 
has also been reported for several tropical bee species (Frankie and 
Baker 1974). To ascertain if the distribution of the multispecies 
assemblage obscured differences among the component species, we 
compared the distributions of the more abundant species with the 
balance of other foraging individuals on the inflorescences. 
Methods 
Five cultivars of sunflower and 2 of onions were grown at the 
Greenville Farm Agricultural Research Station in North Logan, 
Utah. Sunflowers were planted in 5 adjacent 40m rows, 1 row per 
cultivar. The 2 onion cultivars were planted alternately in 4 adjacent 
rows, 2 rows per cultivar. 
Counts of floral visitors were made several times during the 
flowering period as 1 observer (FDP) walked along each row. A 
tape recorder facilitated observations. Only heads with some open 
flowers were censused. 
The data were transcribed to number of flower heads with zero, 
one, two, etc. insects and then compared with values expected on 
the assumption of a Poisson distribution (Southwood 1978). The 
Poisson series describes a random distribution and is written P x (k) = 
e" x (x k /K!) where e = base of Napierian logarithms, and P x is the 
expected number of flower heads with k insects (k = 0, 1, 2,—). The 
parameter x is estimated by the mean number of insects per flower 
head. For the Poisson distribution, the mean and variance are 
equal, and an indication of the dispersion of insects across flowers is 
given by the coefficient of dispersion (C.D. = s 2 /x). When C.D. is 
>1.0 the dispersion is clumped or contagious; and when <1.0 
dispersion is regular or repulsed (Southwood 1978). The expected 
