Simple remedy for pitfalls in detecting negative density dependence

Conspecific negative density dependence (CNDD) is one of the processes that can maintain high species diversity by decreasing population growth rates at high densities, and can thereby favour locally less common species over common ones. But the methods for detection of CNDD can produce false signals, in particular, overestimate CNDD, due to error prone predictors causing regression dilution and underestimation of regression slope. Using simulated and real observed data from tropical forest plot in Barro Colorado Island, we showed that major axis regression can considerably decrease the effects of errors in predictors where classical regression methods did not succeed. The best major axis method correctly identified (1) 93% of no CNDD cases in simulated data, and (2) no CNDD in real species observed data in concordance with direct assessment using survival between censuses. The errors were mostly higher if artificial/virtual adults were introduced in the quadrats with saplings, but without adults. Although major axis methods can be used as a simple remedy for the reductions of these biases, to properly identify dynamic processes like CNDD, repeated census of the plot and identification of parent’s offspring still provide the most relevant data.