Analysing >18,000 effect sizes from recent ecology studies published in five leading journals, the authors identify widespread under-powered designs, exaggeration bias, selective reporting and few corrections for multiple hypotheses among statistically sig…