Use of multi-year LiDAR and plot data to assess biomass growth and aboveground net primary production response to fertilization in regeneration and juvenile stands of coastal Douglas-fir.

In managed forests of coastal British Columbia, fertilization with N is used to reduce rotation age and increase merchantable wood and thus a way to increase carbon sequestration. Generally, effects of fertilization are assessed using sample plots pre- and post-treatment, though they monitor only a portion of the treated area. Remote sensing methods using LiDAR have been used for enhanced forest inventories that spatially extrapolate from plots to larger forest areas. Past research in 58-67yrs Douglas-fir stands has shown multi-year LiDAR can be used to determine fertilization effects on tree height, stem volume and biomass increments in paired fertilized and control blocks. This study examined fertilization effects on tree biomass increment, litterfall and ANPP in regeneration (7-16yrs) and juvenile (20-29yrs) forests and developed a random forest (RF) model to estimate stem and total biomass using multi-year LiDAR to extrapolate from plots to stands. The regeneration site was planted in 2000 with 4.2 ha fertilized in 2007 at 60 kg urea-N ha-1 applied to seedlings. In two juvenile sites, planted in 1988 and in 1990, treated areas (18.9ha, 2.2ha) were both aerially fertilized in 2007 at 200 kg urea-N ha-1 . Plots in control (C ) and fertilized (F) areas were measured in 2002, 2006, 2009 and 2017. LiDAR was acquired in 2004, 3 years before fertilization, 2008, 2011 and 2016. The RF model for each year of LiDAR data used the Boruta variable selection which included height, canopy cover and statistical variables plus tree top density. Plot level models, were applied to the wider area and values from 20 raster cells from C vs F areas compared. Stem periodic annual biomass increment (PAI) in juvenile sites was 2.3 and 3.2 Mg ha-1 yr-1 in 2007-09 and 3.19 and 4.68 Mg ha-1 yr-1 in 2010-17 in C vs F plots respectively, an increase in F vs C in each period. In the regeneration site, PAI was 0.2 and 0.07 Mg ha-1 yr-1 in 2007-09 and 2.84 and 2.24 Mg ha-1 yr-1 in 2010-17 for C and F plots respectively, a decrease in F vs C in each period. The RF stem model R2 was 0.94. PAI for F vs C juvenile site raster cells was 91% greater in 2005-11 but similar in 2012-16. PAI for regeneration site F vs C raster cells was 250% greater in 2005-11 and 81% greater in 2012-16. Results suggest multi-year LiDAR can be used to assess fertilization effects beyond sample plots to the treated area.