Rachael Winfree and colleagues consider that we have misinterpreted their paper (Winfree et al. 2007, Ecology Letters 10: 1105); and that we are incorrect when we suggest that their claim that pollination services from wild, non-honeybee species provide insurance against potential honeybee losses is misleading (Allsopp et al. 2008, PLoS ONE 3: e3128).

In response, we are confident that we have not misinterpreted Winfree et al. 2007 in any way, and that it is indeed ‘misleading to suggest wild pollination services could replace managed services’ (Allsopp et al. 2008, page 2). We do not wish to criticise Winfree et al. 2007 exhaustively to justify our contention, but briefly point out the following:

• The methods used by Winfree et al. 2007 are substantially biased towards the importance of non-honeybee native bees by collecting data only on sunny, relatively windless days. A major part of the renting of millions of honeybee colonies for pollination purposes by growers around the world is ‘bad weather insurance’. In perfect weather conditions, growers know that there are often enough wild pollinators available, pollinators that are largely absent when the weather is less than perfect. Hence the need to flood fields and orchards with introduced honeybee colonies. A ‘fair’ assessment of the contribution of wild bees would not be limited to good days, but rather all weather conditions.

• We suspect that the 23 watermelon fields used in Winfree et al. 2007 do not represent typical agricultural crop fields requiring insect pollination in the study area. While we would not like to accuse Winfree and colleagues of ‘cherry-picking’ fields that just happen to have a sizeable native pollinator contribution, the reality is that the reader has no idea how representative these 23 fields are, and hence how general the conclusions reached. Areas of more intensive agriculture would tend to have a lower native pollinator component, a point acknowledged by Winfree et al. in their discussion, and different regions and different crops will quite obviously have different pollinator profiles.

• It is really not possible to evaluate the contribution of managed bees in Winfree et al. 2007, as no information is presented as regards the introduction of managed pollinators onto the 23 farms in question (i.e. number of hives), or onto neighbouring farms, or of the presence of hived honeybees within foraging distance of crop fields, in the season in question or in previous seasons. This lack of information is accentuated by the data presentation in Winfree et al. 2007 which does not allow for farm-by-farm assessments, but does suggest that the honeybee data is variable. Such variability in the presence of managed pollinators and their subsequent contribution to crop pollination, may be due to failure in recording managed honeybee hives in the study system.

Taken together, it is our contention that Winfree et al. 2007 does not provide an accurate assessment of the contribution of native versus managed pollinators in the pollination of the watermelon fields studied, and furthermore, that the data acquired is not representative of the general situation and cannot be used to extrapolate in this regard. We suggest that while Winfree and colleagues might demur to the contrary, they have presented generalized claims on the basis of their data. That is, they have used these data to espouse that ‘native bees alone provide sufficient pollination at 90% of the farms studied’ and that ‘native bees provide insurance against ongoing honey bee losses’ (the title of the article). Any qualifications that the authors feel should be attached to their results are presented in the final paragraph, and not in the title or abstract, and hence we remain of the opinion that the conclusions reached are not supported by the data and are misleading.

It is perhaps trite to mention, but it should be obvious that if the conclusions reached by Winfree et al. 2007 were indeed accurate and general, then beekeepers must be the greatest salesmen on the planet, growers the most gullible buyers, and the furore around Colony Collapse Disorder (CCD) an unnecessary overreaction. We don’t believe any of this to be true, and the point of our paper (Allsopp et al. 2008, PLoS ONE 3: e3128) was to try to emphasize the need for real-world, economically defendable assessments of the value of pollination services, both managed and wild, so that informed policy decisions to protect them may be reached.


Mike Allsopp, Honeybee Research Section, ARC-Plant Protection Research Institute
Willem de Lange, Environmental and Resource Economics Group, CSIR
Ruan Veldtman, Applied Biodiversity Research, SANBI