The color of environmental noise

Biological populations are strongly influenced by the random variation in their environment. The spectrum of frequencies in noise is particularly important to dynamics and persistence. Here we present an analysis of the variance spectra of a wide variety of long-term time series of environmental variables. Spectra were well approximated by the inverse power law 1/f(beta) within the appropriate range of frequencies f; however, the majority of spectra were "flattened" at low frequencies. With some qualification we found the spectral exponents (beta) to corroborate an earlier suggestion that terrestrial noise tends to be "white" (P < 0.5), while marine environments tend to be "red" (beta approximate to 1) or "brown" (beta approximate to 2). As well, we found a tendency for whiter noise in temperate latitudes than in either high or low latitudes. These results have wide-ranging consequences for ecosystem fragility and species conservation.