Native bees have received increased attention in recent years. Their populations have been declining due to invasive plants (Hanula and Horn 2011), competition from the European honey bee, Apis mellifera L. (Goulson 2003, National Research Council 2007, Paini 2004), introduced pathogens (National Research Council 2007), and habitat loss and fragmentation (Bommarco et al. 2010, Brown and Paxton 2009, Hines and Hendrix 2005, National Research Council 2007). These insects provide an estimated $2–3 billion in pollination services each year to agriculture as well as act to maintain plant communities in natural areas (National Research Council 2007).

Effective conservation efforts for bees should include sampling and monitoring of communities with standardized and efficient methods. Kirk (1984) proposed that color could be a viable tool for attracting and collecting pollinating insects using a so-called flower surrogate. This idea soon became “pan” trapping and has been widely used due to the fact that it requires little technical training, is time and cost effective, and allows for equal sampling effort in variable habitats (Brosi et al. 2007, Campbell and Hanula 2007, Hanula and Horn 2011, Hostetler and McIntyre 2001, Hudson et al. 2012, Romey et al. 2007, Russell et al. 2005, Toler et al. 2005, Westphal et al. 2008). This method generally includes a colored, plastic bowl filled with soapy water suspended above the ground with metal wire or simply placed on the ground. The soap acts to decrease surface tension so insects sink and are trapped (Droege 2008).

Although now widely used, this method has some potential drawbacks. Cane et al. (2000) tested pan traps versus traditional netting techniques. Over multiple years of net surveys, they found that using pan traps alone to measure communities may misrepresent the presence of some bees resulting in bias towards certain families, mainly Halictidae. Other studies have generated similar results, and all recommend netting surveys to accompany pan traps in the field when feasible (Cane et al. 2000, Grundel et al. 2011a, Roulston et al. 2007, Wilson et al. 2008). However, for some habitats, particularly forests with a heavy shrub layer, it is difficult to use netting effectively and consistently across habitats (Campbell and Hanula 2007, Hanula and Horn 2011, Hudson et al. 2012). Additionally, some researchers have speculated that pan traps may reduce local bee abundance over time (Shepherd et al. 2003, Tepedino et al. 2015), but studies have shown this may not be the case (Gezon et al. 2015). Pan traps also seem to exhibit an inverse relationship between effectiveness and floral resource availability (Baum and Wallen 2011 and sources therein), thus rendering them subjective based on what is flowering at the time of sampling. Despite the concern over these attributes, pan traps remain the standard technique for sampling pollinators in most environments.

Some studies have attempted to maximize capture rates of pan traps by using different colors and trap arrangements (Campbell and Hanula 2007, Droege et al. 2010, Leong and Thorp 1999, Toler et al. 2005). Efforts have also used video observations from focal flowers to record pollinator visitation and suggested that video be used as an enhancement technique to accompany surveys using pan traps (Lortie et al. 2012). However, to our knowledge, video has not been utilized to record the behavior of bees when approaching pan traps or the percentage of bees attracted to bowls but not collected. In this study, we analyzed video recordings of pan traps in the field to determine what portion of the bee community attracted to pan traps is not captured.

Materials and Methods

White and blue pan traps were deployed within open habitats at various times throughout the trapping season (April–October) of 2012 at three locations near Athens, GA (State Botanical Gardens of Georgia, University of Georgia's Horticultural Farm, and the University of Georgia's Whitehall Experimental Forest). These two colors were selected because prior experience has shown them to be highly attractive and our objective was focused on measuring the efficiency of pan traps at capturing approaching bees. Two Sony Handycam HDR-XR500V^® cameras were used for each viewing period with each placed over a single pan trap. Each camera was supported on a tripod so it was approximately 30 cm above the ground and 30 cm from the pan trap (Fig. 1A). Cameras and pan traps were left out for approximately 2 h each trapping session, which was the limit of the camera batteries. To maximize possible observations, camera and trap use was limited to sunny days between 10 a.m. and 2 p.m.

View Figure

• Download as Powerpoint
• Download Figure

Video footage was analyzed using Adobe Premiere Elements^® editing software, which could be slowed or stopped for accurate bee identification and determination of behaviors. Videoed bees were classified as either (a) captured, (b) contacting the trap or liquid surface but not captured, or (c) approaching the trap but not contacting it (Fig. 1B). Because the trap filled up most of the camera viewing area, we assumed bees that approached the trap were attracted to it and not simply caught on video by chance. These criteria are similar to previous studies evaluating trap efficiency using video for other insects (McGeachie 1988, Packer and Brady 1990). Bees captured were identified to species using Mitchell (1960) and Gibbs (2010), while those bees that contacted or approached the bowl were identified to the lowest taxonomic level possible, typically genus.

Results and Discussion

A total of 83 bees were recorded during more than 52 h of video observation. We captured 16 bees in our traps, had 25 that made contact but were not captured, and 42 that approached the trap but departed without making contact (Table 1). Lasioglossum was the most commonly observed genus while some genera were represented by only one individual. Species captured made up 19.3% of bees videoed while 30.1% made contact with the trap or liquid and 50.6% approached the trap but made no contact.

Table 1 Number of bees (large- and small-bodied) observed in each video category.

View Fullscreen

We summed bees within each observation criteria as either Apidae or other. Because most Apidae are large bodied (as were the ones encountered in this study), this provided a means to separate relatively larger bees from smaller-bodied bees. Families classified as “other” included Andrenidae, Megachilidae, and Halictidae. Apidae comprised 18.7% of bees captured while other families made up 81.3%. Similarly, 16.0% of the total number of bees that made contact with the trap were Apidae, while 84% of the contacts were other families. Conversely, Apidae made up the majority of bees that approached the bowl but failed to land (61.9%).

Pan traps collected <20% of bees that were attracted to them (Table 1). Of the 16 bees captured during this study, 81% were Lasioglossum spp. and 80% of those that made contact were in the same genus, but they accounted for only 21% of approaches. We only identified species that were captured because this is the only reliable way for accurate determinations. The only large-bodied bee captured was the honey bee, Apis mellifera L. (n = 3), while four species of sweat bee were sampled by our pan traps; Lasioglossum bruneri (Crawford) (n = 3), Lasioglossum hitchensi Gibbs (n = 2), Lasioglossum oblongum (Lovell) (n = 5), and Lasioglossum tegulare (Robertson) (n = 3). If our observations were representative of the bee community in the local area, then our data suggest that pan traps are biased toward Lasioglossum spp. These data are consistent with other studies which showed that pan traps may be biased toward Lasioglossum spp. leading to overestimates of their abundance in nature (Cane et al. 2000, Grundel et al. 2011b, Wilson et al. 2008).

We observed seven genera of bees approaching pan traps, five that made contact, and two that were captured in 52 h of observation. These data suggest that pan traps may underestimate species richness of bees in an area, particularly larger-bodied bees, which is consistent with previous findings (Cane et al. 2000, Roulston et al. 2007, Wilson et al. 2008). Apidae made up only 18.75% of the total number of species captured and only 16% of species that contacted the trap but they represented 62% of the approaches (Table 1).

Harrison and Roberts (2000) reviewed flight, its requirements, and relationships among animal groups and reported that flight efficiency increases as body size increases, probably due to increased metabolism. Other studies have examined the relationship between body size and foraging distances of bees, and all agree that foraging distance increases with body size (Gathmann and Tscharntke 2002, Greenleaf et al. 2007, Guedot et al. 2009). Therefore, larger-bodied bees are likely more efficient fliers than smaller-bodied bees, which may result in more controlled flight patterns and better hovering capabilities. If true, these attributes may allow them to avoid capture by approaching traps more cautiously and use other sensory cues to differentiate the trap from a flower before committing to land (Dobson et al. 1999, Goulson et al. 2001).

Despite the limitation of pan traps, they still catch a large proportion of the bee community when deployed in sufficient numbers in both space and time. It is important to acknowledge and take into account their tendency to be more efficient at capturing Lasioglossum spp. and other small-bodied bees; however, other sampling methods have biases as well. Video techniques that utilize focal flowers show promise, but would be difficult and expensive to set up on a larger scale. Additionally, the lack of specimen collection makes species identification nearly impossible, especially in areas of increased diversity. Netting is another commonly used method and when possible can complement pan trap sampling (Wilson et al. 2008). However, netting can be difficult if flowers are unavailable or vegetation is so thick that the method is not an option (i.e., forests invaded with nonnative shrubs). Thus, pan traps remain the best alternative for studies comparing different habitats, particularly when those habitats vary widely in vegetative structure or when needing to sample pollinator communities in the same habitat over extended periods of time.

Future studies wishing to improve collections should focus on lubricants to make the edges of traps more “slippery” or pursue adding odiferous compounds to potentially attract and capture more bees. Studies conducted to increase captures of forest insects using Lindgren funnel and flight-intercept panel traps showed that an aerosol lubricant can significantly increase trap catch (Allison et al. 2011). Perhaps application of these lubricants to pan trap rims could have the same effect on bee captures. Adding odors or essences to traps as is done to attract orchid bees in South America (Storck-Tonon et al. 2013) and might be considered here. Additionally, studies have shown that bumblebees and carpenter bees are attracted to some floral lures, such as geraniol (Sipolski et al. 2019), although relatively little information exists on the complex of floral scent attractants that flowers use to lure bees (Dötterl and Vereecken 2010).

The results of this study show that pan traps are far from a perfect technique, with less than 20% of bees attracted to the traps actually being collected. In addition, many smaller-bodied bees (i.e., Lasioglossum) may be overestimated while the reverse is true for larger-bodied bees (i.e., Bombus or Melissodes). Although these biases should be taken into consideration anytime pan traps are used, due to their positive attributes (inexpensive, easy to deploy, standardized in space and time), they remain one of the most efficient and repeatable methods available to sample bee communities.