Consequences of decreasing the number of cored trees per plot on chronology statistics and climate–growth relationships: a multispecies analysis in a temperate climate

Existing literature investigates the effect of the number of cored trees per plot (N) on chronology statistics. The present study sought to highlight (i) the effect of N on the reliability of both chronology and climate–growth relationships and (ii) its variability across five European tree species with differentiated ecophysiological patterns. Fifty-eight pure, even-aged forests were sampled across France. For each plot, dendroecological investigations were carried out using chronologies built from 28 to three trees. Chronology reliability was studied using the mean intertree correlation (rbt) and the expressed population signal (EPS), whereas the climate–growth relationships were evaluated through the bootstrapped correlation coefficients (BCC). The accuracy of the dendroecological investigations decreased with decreasing N: EPS and BCC approached zero, implying that the signal common to all trees weakened. Thus, most of the significant correlations became nonsignificant when the sample size decreased from 28 to three trees per plot. Differences were found between species. For a given sample size, the shade-intolerant species Quercus petraea and Pinus sylvestris displayed lower intertree differences in growth-index series (higher rbt) and higher EPS than the shade-tolerant species Abies alba, Fagus sylvatica, and Picea abies. These latter species also displayed a greater sensitivity to sample size decrease, with a stronger BCC weakening and a higher proportion of changes in correlation significance. The EPS threshold of 0.85 was reached for around six to 10 trees for shade-intolerant species versus 20–30 for the shade-tolerant ones and generally corresponded to a mean correlation precision of around 0.06. We finally propose a general method to estimate this precision.