A Comparison of Pan Trap and Intensive Net Sampling Techniques for Documenting a Bee (Hymenoptera: Apiformes) Fauna

Abstract

Recent interest in pollinator sampling, often motivated by concerns about pollinator decline, has led to the increased use of standardized sampling protocols based on passive insect traps. Compared with netting insects at host plants, these protocols have the potential to limit some types of sampling bias, such as those associated with a researcher's observational and netting skills. The most commonly deployed recent protocol is the use of colored pan traps (bowls) filled with soapy water (Aguiar and Sharkov, 1997; Calbuig, 2001; Cane et al, 2000; Leong and Thorp, 1999; Toler et al, 2005). Insects approach the bowls, land on the water, and drown. The method is particularly good at catching numerous species of bees, but can also be effective for capturing various flower-visiting flies, skippers and a wide range of other insect taxa. As a bee sampling device, pan traps have several known biases: they catch bumble bees, honey bees and bees in the genus Colletes much less frequently than expected by their perceived abundance (e.g., Toler et al, 2005). Pan traps are especially good at catching halictid bees (Family Halictidae). Because of these biases, researchers sometimes use a modest amount of net collecting on flowers to accompany pan traps. Other potential biases remain to be studied, such as whether the effectiveness of pan traps is inversely related to flower abundance, a bias suspected by several researchers but not yet studied. Based on their pan-trapping study, Toler et al (2005) concluded that the predominant flower color in the plant community did not influence the relative attractiveness of particular pan trap colors, but they did not study the effect of floral abundance per se. Depending on a researcher's interest in deploying pan traps, these biases may or may not generate concern. Pan traps seem very well suited to testing for the presence of particular bee species in the community (Leong and Thorp, 1999), including many parasitic bee species that are seldom caught at flowers. They also provide a very valuable tool for augmenting other collecting methods, especially when there are few host plants to sample (e.g., early spring). Work by Cane et al. (2000) included a comparison of pan traps versus intensive netting to sample the very diverse and well known flower-visiting insect fauna of creosote bush. These workers concluded that pan trapping was of limited use in detecting the creosote bush fauna because of numerous species caught by netting but not by pan traps. Pan traps did catch many bee species not netted at creosote bush, however. Because netting in that study was restricted to creosote bush, there was no basis to compare the relative effectiveness of netting versus pan trapping for sampling a local bee fauna in the wider flower-rich community. In 2002 in northern Virginia, we carried out an intensive netting/observational survey of floral visitors within one hectare plots at Blandy Experimental Farm, an ecological experiment station of the University of Virginia. The sampled habitat was an open field and the sampling protocol comprised net sampling on the six most prominent plant species in the community from 8:00-16:00 hrs Eastern Daylight Time. Each plant species was surveyed for a total of 2.25 hrs, spread equally across the sampling period. Each of three researchers sampled all plant species in succession, one species at a time, three times across the sampling period (8:00-10:00, 11:00-13:00, 14:00-16:00). Survey periods lasted 15 mins per plant species per researcher, with researchers moving through the entire plot during that period. Bee species that could be recognized on the wing (e.g., Apis mellifera) were noted but not captured. To compare the effectiveness of pan trapping to such an intensive netting protocol at one of our sites, we placed a line of 30 pan traps (6 oz Solo bowls painted fluorescent blue, fluorescent yellow, or left white) in a diagonal line across the plot,