Search for: All records

Abstract Douglas‐fir [Pseudotsuga menziesii(Mirb.) Franco] is the predominant forest plantation species in the Pacific Northwest (PNW), with site productivity and fertilizer response influenced by climate and soil variations. This study investigates the utility of in situ 12‐week supply measurements of nitrogen (N), calcium (Ca), and phosphorus (P) to ion‐exchange resins (specifically Plant Root Simulator [PRS] probes) to estimate carbon (C):N ratios, soil nutrient contents (0–1 m), foliar nutrient concentrations, Douglas‐fir productivity (site index and basal area mean annual increment), and fertilizer volume response. PRS nutrient supply rates were correlated with N, Ca, and P soil nutrient contents (0–1 m), C:N ratios, and foliar nutrient concentrations. Low PRS NO₃supply rates (<25 mg N·m⁻²·burial period⁻¹) were correlated with lower Douglas‐fir productivity and greater fertilizer volume response. PRS NO₃supply rates performed as well as total soil N contents and foliar N concentrations at estimating volume growth response to fertilizer. Twelve weeks after fertilization, PRS NO₃, NH₄, and Ca supply rates were significantly elevated compared to the unfertilized treatment. This research found that PRS probes were an effective in situ tool and are recommended for understanding N, Ca, and P nutrient availabilities, site productivity, and fertilizer response in Douglas‐fir plantations and for developing fertilizer prescriptions. 

Patterns in foliar nitrogen (N) stable isotope ratios (δ15N) have been shown to reveal trends in terrestrial N cycles, including the identification of ecosystems where N deficiencies limit forest ecosystem productivity. However, there is a gap in our understanding of within-species variation and species-level response to environmental gradients or forest management. Our objective is to examine the relationship between site index, foliar %N, foliar δ15N and spectral reflectance for managed Douglas-fir (Pseudotsuga menziesii) and loblolly pine (Pinus taeda) plantations across their geographic ranges in the Pacific Northwest and the southeastern United States, respectively. Foliage was measured at 28 sites for reflectance using a handheld spectroradiometer, and further analyzed for δ15N and N concentration. Unlike the prior work for grasslands and shrubland species, our results show that foliar δ15N and foliar %N are not well correlated for these tree species. However, multiple linear regression models suggest a strong predictive ability of spectroscopy data to quantify foliar δ15N, with some models explaining more than 65% of the variance in the δ15N. Additionally, moderate to strong explanations of variance were found between site index and foliar δ15N (R2 = 0.49) and reflectance and site index (R2 = 0.84) in the Douglas-fir data set. The development of relationships between foliar spectral reflectance, δ15N and measures of site productivity provides the first step toward mapping canopy δ15N for these managed forests with remote sensing. 

Deep learning approaches have been adopted in Forestry research including tree classification and inventory prediction. In this study, we proposed an application of a deep learning approach, Temporal Convolution Network, on sequences of radial resistograph profiles to identify non-thrive trees and to predict wood density. Non-destructive resistance drilling measurements on South and West orientations of 274 trees in a 41-year-old Douglas-fir stand in Marion County, Oregon, USA were used as input series. Non-thrive trees were defined based on their changes in social status since establishment. Wood density was derived by X-ray densitometry from cores obtained by increment borers. Data was split for cross validation. Optimal models were fine-tuned with training and validation datasets, then run with test datasets for model evaluation metrics. Results confirmed that the application of the Temporal Convolution Network on resistograph profiles enables non-thrive tree identification with the probability, represented by the area under the Receiver Operator Characteristic curve, equal to 0.823. Temporal Convolution Network for wood density prediction showed a slight improvement in accuracy (RMSE = 18.22) compared to the traditional linear (RMSE = 20.15) and non-linear (RMSE = 20.33) regression methods. We suggest that the use of machine learning algorithms can be a promising methodology for the analysis of sequential data from non-destructive devices.