2.1. Sampling and storing specimens for analysis
The scope of any sampling largely depends on the aim of the study. In case of identification, it may just comprise the samples of interest, e.g. from a specific apiary or bee line. However, because of inter-colony differences, a reasonable number of at least three colonies should be sampled. If one wants to investigate an entire population covering an area to detect regional differences, about 5 colonies should be sampled per location to keep sampling errors within tolerable limits (Radloff et al., 2003). In particular, attention should be given to the sample coverage of a region, ideally with balanced spacing between locations and no apparent gaps in the sampling to avoid artificial groupings due to local sample clustering. Indeed, resolution of type changes strongly depends on sample density, where at low density large-scale differences tend to be more pronounced, but small-scale regional differences might get blurred (Radloff and Hepburn, 1998). Therefore, it may be necessary to differentiate between mesoscale and macroscale studies and adapt the sampling regime accordingly (Radloff et al., 2010).
Ideally, samples should be collected in a way that ascertains their origin from a given hive under analysis. While the sampling of pupae would be ideal for this purpose, they are not suitable for morphometric analysis and their collection is time consuming. Therefore, we recommend sampling workers from inside the colonies, most desirably from the brood area. However, this is also not always possible, since hives may not easily be opened for a variety of reasons (traditional hives, natural comb, defensive bees, etc.). In these cases sampling from the flight entrance will have to suffice. However, one sample should always represent one colony, i.e. only contain workers sampled from one hive. The sampling of drones is more difficult, since they are only seasonally available and more prone to drifting than workers (Currie and Jay, 1988; Jay, 1969; Neumann et al., 2000).