Ecological studies with Scarabaeinae dung beetles have increased exponentially over the past 30 years, using lethal pitfall traps baited with mammal feces or carrion as the preferred sampling method. Different studies have determined the distance between pitfall traps for effective sampling, but the number of traps is often subjective, leading to excessive or poor sampling. This study provides quantitative guidelines for establishing the sample size for optimal completeness of dung beetle diversity by systematically reviewing the relationship between sampling intensity and sampling coverage, habitat type, and the journal impact factor in peer-reviewed research. We gathered 94 studies covering a range from México to Argentina. Sampling was conducted mainly in forested habitats, followed by treeless agriculture and agroforestry systems, with a median value of 50 pitfall traps per sampled habitat. Sampling completeness was above 0.9 in 95% of the studies. Oversampling ranged from 1 to more than 96,000 individuals, and sampling deficit varied between 2 and 3,300 specimens. Sampling intensity and the journal impact factor were significantly and positively correlated with oversampling, but these variables did not explain the sampling deficit. The positive correlation between journal impact factor and oversampling may reflect a publication bias where high-impact journals and researchers seek more generalizable information obtained with a higher sampling intensity. Dung beetle oversampling was not homogeneous between habitats, being highest in old-growth forests and lowest in disturbed habitats such as pastures and forest edges. Our results show that the collection intensity used in dung beetle studies should be reconsidered carefully. By incorporating ethical principles used in animal science, we suggest sampling guidelines for a robust sampling scheme of dung beetle diversity, which would also prevent oversampling. Consciously reducing sampling intensity will make resource use more cost-effective. We suggest increasing the number of independent sampling units rather than intensifying subsampling, thereby increasing the predictive power of statistical models to obtain more robust evidence of the phenomena under study.
Read full abstract