Parsing variation in long-term patterns underlying insect abundances and assigning mechanisms are critical in light of recent reports of dramatic insect declines.

Grasshopper abundances in a North American prairie exhibited both 5-y cycles and >2%/y declines over the past 20 y. Large-scale climate oscillations predicted the cycles in grasshopper abundances. Moreover, plant biomass doubled over the same period—likely due to changes in climate and increasing atmospheric CO2—diluting the concentrations in plant tissue of key nutrients which in turn predicted the declines of a dominant herbivore. Nutrient dilution, like CO2 enrichment, is likely a global phenomenon, posing a challenge to Earth’s herbivore populations.




Evidence for global insect declines mounts, increasing our need to understand underlying mechanisms. We test the nutrient dilution (ND) hypothesis—the decreasing concentration of essential dietary minerals with increasing plant productivity—that particularly targets insect herbivores. Nutrient dilution can result from increased plant biomass due to climate or CO2 enrichment. Additionally, when considering long-term trends driven by climate, one must account for large-scale oscillations including El Niño Southern Oscillation (ENSO), North Atlantic Oscillation (NAO), and Pacific Decadal Oscillation (PDO). We combine long-term datasets of grasshopper abundance, climate, plant biomass, and end-of-season foliar elemental content to examine potential drivers of abundance cycles and trends of this dominant herbivore. Annual grasshopper abundances in 16- and 22-y time series from a Kansas prairie revealed both 5-y cycles and declines of 2.1–2.7%/y. Climate cycle indices of spring ENSO, summer NAO, and winter or spring PDO accounted for 40–54% of the variation in grasshopper abundance, mediated by effects of weather and host plants. Consistent with ND, grass biomass doubled and foliar concentrations of N, P, K, and Na—nutrients which limit grasshopper abundance—declined over the same period. The decline in plant nutrients accounted for 25% of the variation in grasshopper abundance over two decades. Thus a warming, wetter, more CO2-enriched world will likely contribute to declines in insect herbivores by depleting nutrients from their already nutrient-poor diet. Unlike other potential drivers of insect declines—habitat loss, light and chemical pollution—ND may be widespread in remaining natural areas.





Figure 1:

Mean grasshopper abundance/sample (±SE) exhibited a linear decline from 1996 to 2017 in ungrazed KNZ watersheds (AF(1,20) = 7.5, R2 = 0.27, P = 0.01, slope = −0.022) and from 2002 to 2017 in bison-grazed KNZ watersheds (BF(1,14) = 9.7, R2 = 0.41, P = 0.008, slope = −0.033). The time series of grasshopper abundances in ungrazed watersheds was decomposed into a 5-y cycle (C) and decycled (E) trend. The time series of grasshopper abundances in grazed watersheds was decomposed into a 5-y cycle (D) and decycled (F) trend. Black dashed lines indicate linear regression trends of the raw abundances (A and B); solid black lines indicate linear regression of decycled trends (E and F). Declines persisted after removing the cycle for grasshopper abundances in both ungrazed time series (EF(1,20) = 8.9, R2 = 0.31, P = 0.007, slope = −0.021) and grazed time series (FF(1,14) = 16.4, R2 = 0.54, P = 0.001, slope = −0.027). The presence of the 5-y cycle had minimal impact on the slope of grasshopper abundance over time in ungrazed watersheds. In the shorter time series from grazed watersheds, the decomposed estimate of a 2.7% annual decline was ∼20% less steep than the raw estimate of a 3.3% annual decline, supporting the axiom that, in a cyclical system, shorter time series increase the likelihood of slope estimation error (SI Appendix, Fig. S12). Mean grasshopper abundance/sample in ungrazed watersheds (A) was positively correlated with the combined climate index of spring ENSO + summer NAO – winter PDO (G). Mean grasshopper abundance/sample in grazed watersheds (B) was positively correlated with the climate index of summer NAO − spring PDO in grazed watersheds (H).