Predicting wood density of growth increments of Douglas-fir stands in New Zealand

Background: Douglas-fir comprises 6% of New Zealand's planted forest area and contributes to the national carbon sequestration estimate. Carbon stock changes in Douglas-fir stem wood can be calculated by multiplying the increment in stem volume under bark by the density of the growth sheath and the carbon fraction. This paper describes a new model developed to predict variation in wood density of Douglas-fir growth sheaths from known wood density drivers. Methods: Datasets used to parameterise the wood density model contained: (1) mean breast height (1.4 m above ground) outerwood (based on 50 mm long cores) density of 30 trees per stand from 32 semi-mature and mature stands with soil and climate data, for predicting wood basic density from environmental factors; (2) basic density of wood samples taken at breast height and at regular height intervals along the stem of 75 trees from 10 stands, for predicting the weighted mean wood basic density of pre-defined growth sheaths; and (3) breast height pith-to-bark radial density profiles based on 500 trees from 47 stands, for predicting wood density of individual annual growth rings at breast height. Linear and non-linear mixed models were developed using these data to explain the variation in wood density of growth sheaths. Results: Breast height outerwood density was positively related to mean annual air temperature and negatively related to soil nitrogen fertility. This environmental model jointly explaining 83% of the variation in wood basic density at breast height of ring 30 from the pith. The radial pattern of wood density variation of annual growth rings at breast height was calibrated to a site using predictions from the environmental model. The ratio of growth sheath density to breast height ring density was applied to the predicted density of annual growth rings at breast height. This ratio decreased from 1.05 close to the pith to 0.95 in outerwood rings of mature trees, with ring number from pith explaining 49% of the variation in this ratio. Conclusions: A wood density model that incorporates the important drivers of variation in density of stem wood growth sheaths in New Zealand-grown Douglas-fir will improve the accuracy of carbon stock and stock change estimates in NZ's planted forest estate. The new wood density model has been linked to the Forest Carbon Predictor which predicts carbon stocks in live and dead biomass pools from inventory plot measurements, site mean annual temperature, and soil nitrogen fertility information.