National food production stabilized by crop diversity
Delphine Renard & David Tilma, 2019.

ncreasing global food demand, low grain reserves and climate change threaten the stability of food systems on national to global scales. Policies to increase yields, irrigation and tolerance of crops to drought have been proposed as stability-enhancing solutions. Here we evaluate a complementary possibility—that greater diversity of crops at the national level may increase the year-to-year stability of the total national harvest of all crops combined. We test this crop diversity–stability hypothesis using 5 decades of data on annual yields of 176 crop species in 91 nations. We find that greater effective diversity of crops at the national level is associated with increased temporal stability of total national harvest. Crop diversity has stabilizing effects that are similar in magnitude to the observed destabilizing effects of variability in precipitation. This greater stability reflects markedly lower frequencies of years with sharp harvest losses. Diversity effects remained robust after statistically controlling for irrigation, fertilization, precipitation, temperature and other variables, and are consistent with the variance-scaling characteristics of individual crops required by theory for diversity to lead to stability. Ensuring stable food supplies is a challenge that will probably require multiple solutions. Our results suggest that increasing national effective crop diversity may be an additional way to address this challenge.

 Determinants of national caloric yield stability. Regression coefficients (±s.e.) show the magnitude of the effect of each variable in a multiple regression of log_e(national yield stability). a, Regression coefficients using effective crop group diversity (n = 437). b, Regression coefficients using effective crop species diversity (n = 437). Each predictor variable was standardized to zero mean and unique variance across all nations and time periods (Methods) to enable the comparison of effects with a change of 1σ in each predictor on the log(stability) values. Asterisks indicate the significance of each predictor. ***P < 0.001; NS, not significant (P > 0.05).

.

https://www.nature.com/articles/s41586-019-1316-y

.