Modelling resin production distributions for Pinus pinaster Ait. stands in NW Spain

Abstract

Pine resin, a viscous material secreted as a defensive response to biotic or abiotic damage, is a highly valuable non-wood forest product with multiple uses in the industrial sector. Resin production can be induced by tapping live trees, but not all pine species produce resin of suitable quality and/or in profitable quantities. Maritime pine (Pinus pinaster Ait.) is currently the only species tapped in Spain, where resin tapping activity has been recovered in the last few decades. Most studies on resin production focus on the mean production per tree or per area, and less attention is given to determining how the production is distributed across individuals or production classes. We modelled the distribution of resin production in Pinus pinaster stands in Galicia (NW Spain) by using the Weibull function and the moment-based parameter recovery method. We observed a high level of variance in resin production between plots (different sites, stimulants used, tapping method or year) and within plots, between trees. All resin production distributions modelled using the arithmetic mean resin production (x¯) and the variance of the distribution (σ2) per plot satisfied the Kolmogorov-Smirnov (KS) test, in which critical values were obtained by Monte Carlo simulation. The variance of the distribution (σ2) was positively correlated with x¯, and the relationship was described by an exponential model. When resin production distributions were modelled using x¯ and estimated variance (σˆ2), 7% of the distributions (corresponding to trees in which chemical stimulants were not used) did not satisfy the KS test. The mean production (x¯) can be easily determined by dividing the stand production by the number of trees. However, x¯ could also be estimated before commercial tapping by sampling a representative number of trees. We conclude that in order to estimate x¯, a minimum sample of 50–60 trees should be tapped, to yield a relative standard error (RSE) below 10%; 10–15 trees should be considered for RSE < 20% and 5–10 for RSE < 30%.