Current Biology 18, 1572���1575, October 28, 2008 ��2008 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2008.08.066 Report Long-Term Global Trends in Crop Yield and Production Reveal No Current Pollination Shortage but Increasing Pollinator Dependency Marcelo A. Aizen,1,* Lucas A. Garibaldi,1,2 Saul A. Cunningham,3 and Alexandra M. Klein4,5 1Laboratorio Ecotono INIBIOMA-CONICET and Centro Regional Bariloche Universidad Nacional del Comahue Quintral 1250 8400 Bariloche R�� ��o Negro Argentina 2Catedra �� de Metodos �� Cuantitativos Aplicados Facultad de Agronom����a Universidad de Buenos Aires Av. San Mart��n �� 4453 1417 Buenos Aires Argentina 3CSIRO Entomology GPO Box 1700 Canberra ACT 2601 Australia 4Environmental Sciences Policy and Management 137 Mulford Hall University of California, Berkeley Berkeley, California 94720-3114 5Agroecology University of Goettingen Waldweg 26 37073 Goettingen Germany Summary There is evidence that pollinators are declining as a result of local and global environmental degradation [1���4]. Because a sizable proportion of the human diet depends directly or in- directly on animal pollination [5], the issue of how decreases in pollinator stocks could affect global crop production is of paramount importance [6���8]. Using the extensive FAO data set [9], we compared 45 year series (1961���2006) in yield, and total production and cultivated area of pollinator-depen- dent and nondependent crops [5]. We investigated temporal trends separately for the developed and developing world because differences in agricultural intensification, and so- cioeconomic and environmental conditions might affect yield and pollinators [10���13]. Since 1961, crop yield (Mt/ha) has increased consistently at average annual growth rates of w1.5%. Temporal trends were similar between pollina- tor-dependent and nondependent crops in both the devel- oped and developing world, thus not supporting the view that pollinator shortages are affecting crop yield at the global scale. We further report, however, that agriculture has become more pollinator dependent because of a dispro- portionate increase in the area cultivated with pollinator- dependent crops. If the trend toward favoring cultivation of pollinator-dependent crops continues, the need for the ser- vice provided by declining pollinators will greatly increase in the near future. Results and Discussion Although many crops have a long history of human selection, this has not, in most cases, circumvented their pollinator de- pendency. Pollinators can even enhance seed production of genetically engineered crops such as canola, soybean, and sunflower [14���16]. Interestingly, the incidence of pollination limitation (i.e., the percent of species that do not express their full reproductive potential in terms of fruit or seed set when pollination is limited) is relatively high among cultivated plants, w60%, similar to the incidence observed in wild spe- cies [17, 18]. For these reasons, we expect that many crops could be susceptible to current declines in pollinator abun- dance and diversity of wild pollinators [1, 19, 20] and increas- ingly frequent collapses of managed honey bee (Apis mellifera) populations [2, 3, 21, 22]. Using FAO data [9], we examined temporal trends in yield, total production, and cultivation of crops over the past five de- cades in relation to pollinator dependency at a global scale. Particularly, we compared the group that depends to some extent on pollinators to produce the parts we consume (fruits or seeds) to the group of crops that does not depend on animal pollination, mostly because they are wind- or autogamously pollinated or cultivated for their vegetative parts (e.g., leaves, stems, tubers, etc.). This latter nondependent group also includes crops, such as potatoes and other vegetables, which do not depend directly on pollinators for the production of the parts we consume but for which pollinators are still important for propagation via seed or in breeding programs. The animal- pollinated crops represent a phylogenetically diverse group of species for which production, in terms of number or size of the seeds or fruits, is influenced by the presence of pollinators. The degree of dependence, however, varies greatly, such that absence of effective pollinators would reduce production by 100% in the extreme case, and by only a few percent for low-dependence species [5, 23]. Temporal trends were ana- lyzed separately for the developed and developing world (see justification in the Supplemental Data available online). The most stringent expectation from the hypothesis that global agriculture is experiencing a pollination shortage [6���8, 24] is that, all else being equal, pollinator-dependent crops should show declining average yield (i.e., metric tonnes per hectare) during at least the last part of the 45 year study period, a trend that would not be observed among the nondependent crops. This is probably a highly naive prediction considering that nothing has remained equal agriculture is substantially more intensified today than half a century ago, and there has been increasing use of selectively bred or genetically modified high-yielding crop varieties [10]. We might then expect an increase over time in yield for most crops, regardless of their degree of pollinator dependency. However, under a pollina- tion-shortage scenario we could predict a lower relative yield growth among pollinator-dependent than that among nonde- pendent crops, a trend exacerbated in recent decades if polli- nators had become an increasingly limiting resource [1]. Contrary to expectation, we found little evidence of differ- ences in relative yield between pollinator dependent and nondependent crops. Yield has increased since 1961 by *Correspondence: marcelo.aizen@crub.uncoma.edu.ar

(mean 6 1 SE) 1.30% 6 0.32%/year in the developed world and 1.61% 6 0.17%/year in the developing world. Overall the mean relative yield of pollinator-dependent crops showed a similar increase over time as the mean relative yield of the nondependent crops for both the developed and developing world (Figure 1). We found no significant differences in the average growth rate in relative yield between dependent and nondependent crops in the developed world (mean 6 1 SE = 1.31 6 0.50 versus 1.28% 6 0.35%/year, t test, t = 0.05, Df = 68, p = 0.96) or developing world (mean 6 1 SE = 1.53 6 0.24 versus 1.72% 6 0.23%/year, t test, t = 20.59, Df = 80, p = 0.59). Because many tropical, pollinator-dependent crops are exclusively or predominantly cultivated in the developing world (e.g., Brazil nut, cocoa beans, oil palm, etc.), we also compared trends in relative yield for a subset of ten common pollinator-dependent crops and ten common nondependent crops (see the Supplemental Data for details on crop selec- tion). Analysis of this crop subset revealed no significant effect of pollinator dependency despite a trend toward a lower growth rate in average yield among pollinator-dependent than in that among nondependent crops in both the developed and developing world (Figure 1, Figure S1, and Table S1). Thus, these results do not support the view that a pollination short- age affects agriculture at a global scale. Further distinction among dependent crops does suggest, however, that high pollinator dependence might impose a limit to the rate of yield growth. This was particularly apparent in the case of the highly pollinator-dependent crops from the developing world that exhibited an average growth rate in rel- ative yield that was half the rate shown by the crops with low pollinator dependence (Figure S2 and Table S2). A similar pat- tern was found in the developed world once we excluded a crop complex, Cucurbita spp., a putative outlier in terms of Figure 1. Temporal Trends in Mean Crop Yield from 1961 to 2006 For each crop grown in the developed world and developing word, we estimated the percent change of yield (D yield) at year t with respect to its value in 1961. The depicted means (61 SE) in relative yield were estimated from all pollinator- dependent and nondependent crops included in our data set and from a subset of ten pollinator- dependent and ten nondependent crops widely cultivated in temperate and subtropical regions worldwide. For instance, this figure shows that the yield of commonly cultivated crops has doubled over the study���s 45 year period. its rapidly increasing annual growth rate in relative yield. These results imply that the availability of effective pollina- tors may be imposing an upper limit to many crops��� yield increase that has not been completely offset by agricultural intensification, selective breeding, or ge- netic engineering. If growth in yield of highly pollinator-dependent crops is limited by pollinator availability, there will be an even greater demand for agri- cultural land to meet growing global consumption. The subset of 20 shared crops also illustrates differences in yield according to the extent of development in the region. Despite the fact that productivity of both pollinator-dependent and nondepen- dent crops increased almost 2-fold during the study 45 year period, the differences between the developed and developing world in absolute yield in 1961 (mean 6 1 SE = 5.8 6 1.63 ver- sus 3.3 6 0.94 Mt/ha, paired t test, t = 3.14, Df = 19, p = 0.0053) still persisted in 2006 (mean 6 1 SE = 10.5 6 2.88 versus 7.6 6 2.51Mt/ha, paired t test, t = 2.75, Df = 19, p = 0.012). Higher ab- solute crop yields observed in the developed world compared with the developing world presumably reflect disparate socio- economic conditions and differences in agricultural intensifi- cation and subsidy policies [12, 13]. However, these contrasts, as well as environmental differences between these two re- gions of the world, did not seem to influence to any large extent differences in the growth rate in relative yield between pollina- tor-dependent and nondependent crops (Figure 1, Figure S1, and Table S1). The absence of a positive correlation in the annual growth rate in relative yield across this shared crop subset between the developed and developing world (r = 20.135, N = 20, p = 0.57), further implies that growth in yield is influenced quite idiosyncratically, to at least some extent, by socioeconomic or environmental factors rather than by intrin- sic crop traits. Thus, although differences between the devel- oped and developing world appear to be important drivers of change in agriculture, none of these changes is so strongly linked to the pollination problem that a clear signal emerges. Unlike yield, aggregate production of pollinator-dependent and nondependent crops showed strikingly different temporal trends. In both the developed and developing world there has been a steady increase in the production of pollinator-depen- dent crops that surpassed the increase rate in the production of nondependent crops (Figure 2). Pollinator-dependent crops contributed 8.4% to total agriculture production in the Pollinator Decline and Crop Yield 1573