NASS statistics also do not include an assessment of pollination-relevant characteristics such as colony quality (strength, presence of parasites or pathogens), number of colonies lost over the winter (trade journal reports of winter losses are largely anecdotal), number of colonies rented for pollination, pollination rental fees, crops pollinated, and the numbers of queens and packages produced. At present, NASS statistics provide data only from beekeepers; such information from growers as rental price paid or the ease of obtaining the appropriate pollination service would provide a useful comparison with data collected from beekeepers to allow for an assessment of demand and shortages.
Several contemporary investigators have visited historical field sites where earlier pollinator surveys, particularly of native bees, had been conducted to determine if landscape changes during intervening years had resulted in changes in bee guild composition or losses of species from the area. Carlinville, Illinois, was sampled from 1884 to 1916 by Robertson (1929), who collected 214 bee species on over 400 plant species. Of the 214 species, 157 were found on only 24 of the plant species sampled. Approximately three-quarters of a century later (1970–1972), Marlin and LaBerge (2001) repeated that survey in Carlinville, concentrating their sampling effort on the 24 plant species that provided the bulk of the bee species reported by Robertson (1929). They collected 140 species of bees representing 82 percent of the species found by Robertson (as well as 14 species not recorded on those plants in the earlier survey). The relatively high degree of similarity in the bee community, despite the passage of 75 years and extensive landscape changes, was not the anticipated result. The authors suggested that patches of diverse habitats embedded within the agricultural matrix (for example, rural grasslands, forests, and open woods) have maintained bee diversity over time despite major changes in land use patterns.
Kevan and his colleagues have been analyzing data for pollinator diversity and abundance (Kevan et al., 1997) on New Brunswick blueberry fields for a period of about 8 years and find that the Sørensen, and other indices, of similarity between years but on the same fields are typically low (about 0.2) (unpublished). Turnock et al. (in preparation) have analyzed 8-year-long patterns of abundance in bumble bees in Manitoba, and noted changes by orders of magnitude from one year to the next. Javorek has some longer term studies ongoing in New Brunswick and Nova Scotia (Javorek et al., 2002). Sheffield (2006) has compared the data of Brittain (1933)