Aaron R. Weiskittel

Comparison of model forms for estimating stem taper and volume in the primary conifer species of the North American Acadian Region

Abstract• The performance of ten commonly used taper equations for predicting both stem form and volume in balsam fir [Abies balsamea (L.) Mill], red spruce[Picea rubens (Sarg.)], and white pine[Pinus strobus (L.)] in the Acadian Region of North America was investigated.• Results show that the Kozak (2004) and Bi (2000) equations were superior to the other equations in predicting diameter inside bark for red spruce and white pine, while the Valentine and Gregoire (2001) equation performed slightly better for balsam fir.• For stem volume, the Clark et al. (1991) equation provided the best predictions across all species when upper stem diameter measurements were available, while the Kozak (2004) and compatible taper equation of Fang et al. (2000) performed well when those measurements were unavailable.• The incorporation of crown variables substantially improved stem volume predictions (mean absolute bias reduction of 7–15%; root mean square error reduction of 10–15%) for all three species, but had little impact on stem form predictions.• The best taper equation reduced the predicted root mean square error by 16, 39, and 45% compared to estimates from the widely used Honer (1965) regional stem volume equations for balsam fir, red spruce, and white pine, respectively.• When multiple taper equations exist for a certain species, the use of the geometric mean of all predictions is an attractive alternative to selecting the “best” equation.Résumé• Les performances de dix équations de la décroissance de la tige, couramment utilisées pour prédire à la fois la forme du tronc et le volume pour le sapin baumier [Abies balsamea (L.) Mill], l’Épinette rouge [Picea rubens (Sarg.)], et le pin Weymouth [Pinus strobus (L.)] ont été étudiées dans la région de l’Acadie en Amérique du Nord.• Les résultats montrent que les équations de Kozak (2004) et de Bi (2000) étaient supérieures aux autres équations pour la prédiction du diamètre sous écorce pour l’épinette rouge et le pin Weymouth, tandis que l’équation de Valentine et Gregoire (2001) était légèrement meilleure pour la forme du tronc du sapin baumier.• Pour le volume de la tige, l’équation de Clark et al. (1991) fourni les meilleures prévisions pour toutes les espèces lorsque les mesures du diamètre de la partie supérieure de la tige étaient disponibles, tandis que l’équation de Kozak (2004) et l’équation compatible de défilement de Fang et al. (2000) conviennent bien lorsque ces mesures n’étaient pas disponibles.• L’incorporation de variables de couronne a amélioré sensiblement les prédictions du volume des troncs (réduction moyenne des biais absolu de 7–15 % ; réduction de l’erreur quadratique moyenne de 10–15 %) pour les trois espèces, mais avait peu d’impact sur les prédictions de la forme du tronc.• La meilleure équation de décroissance a réduit l’estimation de l’erreur quadratique moyenne de 16, 39, et 45 % par rapport aux estimations largement utilisées avec les équations régionales d’Honer (1965) pour l’estimation du volume de la tige respectivement pour le sapin baumier, l’épinette rouge et le pin Weymouth.• Lorsque plusieurs équations de défilement existent pour certaines espèces, l’utilisation de la moyenne géométrique de toutes les prédictions est une alternative intéressante pour la sélection de la “meilleure” équation.

Sources of within- and between-stand variability in specific leaf area of three ecologically distinct conifer species

Specific leaf area (SLA) is an important ecophysiological variable, but its variability within and between stands has rarely been simultaneously examined and modeled across multiple species. Extensive datasets on SLA in coastal Douglas-fir (Pseudotsuga menziesii var. menziesii (Mirb.) Franco), hybrid spruce (Picea engelmannii Parry × Picea glauca (Moench) Voss × Picea sitchensis (Bong.) Carr.), and ponderosa pine (Pinus ponderosa Dougl. ex P. & C. Laws.) were used to estimate variability of SLA within a canopy and its relationship to tree- and stand-level covariates, and to predict SLA at various locations in tree crowns. Also, in the case of hybrid spruce, variation in SLA due to different relative horizontal lengths from the bole was examined. In all species, SLA systematically increased from tree tip to crown base and decreased with foliage age class. Cardinal direction did not have a highly significant influence in either Douglas-fir or hybrid spruce, but SLA did significantly decrease from branch tip to bole in hybrid spruce. Tree- and stand-level (e.g. density, site index) factors had relatively little influence on SLA, but stand age did have a significant positive influence. For ponderosa pine, a significant relationship between canopy mean current-year SLA and carbon isotope discrimination was also found, suggesting the importance of water stress in this species. An equation was fitted to estimate SLA at various points in the canopy for each species and foliage age class using absolute height in the canopy, relative vertical height in the tree, and stand age.RésuméLa surface spécifique des feuilles (SLA) est un paramètre écophysiologique important mais sa variabilité intra et inter-peuplements n’a jamais été examinée et modélisée sur des gammes larges d’espèces. Des jeux de données très détaillés de SLA de Douglas côtiers [Pseudotsuga menziesii var. menziesii (Mirb.) Franco], d’épicéas hybrides (Picea engelmannii Parry × Picea glauca (Moench) Voss × Picea sitchensis (Bong.) Carr), et de pins ponderosa (Pinus ponderosa Dougl. ex P. & C. Laws.) ont été mobilisés pour évaluer la variabilité de SLA dans une canopée. Les relations entre SLA et des covariables à l’échelle de l’arbre ou du peuplement ont été précisées, un modèle prédictif de SLA à différents niveaux dans les couronnes a été construit. Dans le cas de l’épicéa, l’impact de la distance de branche entre l’aiguille et le tronc a également été testé. Dans toutes les espèces, SLA augmentait systématiquement du sommet des arbres à la base de la couronne, et diminuait avec la classe d’âge des aiguilles. La direction cardinale n’avait guère d’influence sur SLA ni dans le cas du Douglas ni dans celui de l’épicéa; mais SLA diminuait systématiquement depuis l’extrémité des branches vers le tronc. Les facteurs arbre et peuplement (comme la densité, l’indice de productivité de la station) n’avaient que peu d’impact sur SLA alors que l’âge du peuplement avait un effet significatif et positif. Pour le pin ponderosa, une relation significative a été détectée entre la valeur moyenne de SLA des aiguilles de l’année et la discrimination isotopique du carbone, ce qui suggère l’impact des contraintes hydriques pour cette espèce. Un modèle de prédiction de SLA à différentes positions dans la canopée a été ajusté sur les données de chaque espèce et classe d’âge, en se basant sur la hauteur dans la canopée, la hauteur relative dans l’arbre et l’âge du peuplement.

A hybrid model for intensively managed Douglas-fir plantations in the Pacific Northwest, USA

Recent advances in traditional forest growth models have been achieved by linking growth predictions to key ecophysiological processes in a hybrid approach that combines the strengths of both empirical and process-based models. A hybrid model was constructed for intensively managed Douglas-fir plantations in the Pacific Northwest, USA, by embedding components representing fundamental physiological processes and detailed tree allometrics into an empirical growth model for projecting individual tree and stand development. The simulated processes operated at a variety of scales ranging from individual branches to trees and stands. The canopy structure submodel improved predictions of leaf area index at the stand level when compared to allometric and other empirical approaches (reducing mean square error by 30–42%). In addition, the hybrid model achieved accuracy in short-term volume growth prediction comparable to an empirical model. Biases in 4-year stand growth predictions from the hybrid model were similar to those from the empirical model under thinning, fertilization, and the combination of these treatments; however, volume growth predictions in unmanaged plantations averaged approximately 36% less bias. These improvements were attributed to detailed information on crown structure (i.e. size, location, and foliage mass of primary branches), simple representation of key physiological processes, and improved site characterization. Soil moisture, temperature, and nitrogen mineralization predicted by the hybrid model also agreed closely with observed values from several previous studies. Overall, the model framework will be helpful for future analyses as it can lend insight into the influence of weather and site edaphic factors on growth, help identify mechanisms of response to silvicultural treatments, and facilitate the design of sound management regimes for Douglas-fir plantations across the Pacific Northwest region.

Multivariate Spatial Regression Models for Predicting Individual Tree Structure Variables Using LiDAR Data

This study assesses univariate and multivariate spatial regression models for predicting individual tree structure variables using Light Detection And Ranging (LiDAR) covariates. Many studies have used covariates derived from LiDAR to help explain the variability in tree, stand, or forest variables at a fine spatial resolution across a specified domain. Few studies use regression models capable of accommodating residual spatial dependence between field measurements. Failure to acknowledge this spatial dependence can result in biased and perhaps misleading inference about the importance of LiDAR covariates and erroneous prediction. Accommodating residual spatial dependence, via spatial random effects, helps to meet basic model assumptions and, as illustrated in this study, can improve model fit and prediction. When multiple correlated tree structure variables are considered, it is attractive to specify joint models that are able to estimate the within tree covariance structure and use it for subsequent prediction for unmeasured trees. We capture within tree residual covariances by specifying a model with multivariate spatial random effects. The univariate and multivariate spatial random effects models are compared to those without random effects using a data set collected on the U.S. Forest Service Penobscot Experimental Forest, Maine. These data comprise individual tree measurements including geographic position, height, average crown length, average crown radius, and diameter at breast height.

Cutpoint analysis for models with binary outcomes: a case study on branch mortality

Models of binary outcomes are commonly used in forestry, but the predictions errors of these types of models are difficult to present effectively. In addition, most studies generally use a fixed value of 0.5 as the separation between events and non-events. The use of cutpoint analysis has been widely utilized in the health sciences and other fields, while it is relatively uncommon in the forestry literature. Cutpoint analysis involves locating the optimal value that minimizes prediction errors associated with binary outcomes. This case study illustrates the use of cutpoint analysis to improve a dynamic model of individual branch mortality. In this study, the use of cutpoint analysis increased the model specificity (prediction of events) from 77.8% (standard cutpoint of 0.5) to 90.3% (optimal cutpoint of 0.672). At the same time, the sensitivity of the model decreased only slightly and the false positive rate (non-event predicted as an event) was greatly decreased from 22.2 to 9.7%. In addition, the use of receiver operating characteristics (ROC) curves was an effective approach for evaluating prediction errors of models of binary outcomes. Cutpoint analysis is a simple yet effective method for improving predictions of binary outcomes and should be used more regularly, particularly when modelling the binary outcome of rare events such as mortality.

Branch surface area and its vertical distribution in coastal Douglas-fir

Wood area index (WAI; total surface area of branches and bole per unit of land area) is an important yet often neglected forest structural attribute. Branchwood surface area, in particular, has significant implications for many ecophysiological processes including total respiration and interception of radiation and rainfall. Branch surface area was estimated at the branch-, tree-, and stand-level for 33 Douglas-fir (Pseduotsuga menziesii [Mirb.] Franco) plantations in the Oregon Coast Range. Patterns in WAI, leaf area index (LAI; total surface area of needles per unit of land area), tree area index (TAI=WAI+LAI) and various ratios of these dimensions were then investigated. The main axes of primary branches (those attached to the main stem) comprised 82±13% of total branchwood surface area. Tree surface area (needles + woody tissue) increased with increasing tree size and crown length, and decreased with greater intensity of Swiss needle cast (SNC). At the stand-level, woody surface area increased with greater stand density and decreased with more severe SNC, but on average it constituted 29±12% of total tree surface area. Branchwood surface area and bole surface area contributed equally to WAI. The variation in WAI for a given LAI has important implications for radiation and rainfall attenuation in these stands and for accurate partitioning of intercepted radiation between photosynthetic and non-photosynthetic tissues.

Leaf mass per area relationships across light gradients in hybrid spruce crowns

This study examined the distribution and variation of mass to projected area ratio of foliage (LMA, g/m2) within hybrid spruce (Picea engelmannii Parry×Picea glauca (Moench) Voss×Picea sitchensis (Bong.) Carr) crowns across three stages of stand development (20, 60, and 140 years). Variability of LMA was assessed at different heights and branch positions. LMA decreased with distance from top of tree (p<0.0001) at rates that varied among stand development stages (p<0.0001). A multi-level mixed effect analysis indicated that distance from the tip of the first-order branch (p=0.0002) had a significant influence on LMA. In general, LMAs decreased towards the base of the tree and increased towards tree apex.LMA differed significantly across stand development stages. Older trees (140 years) showed the highest LMA, while younger trees (20 years) showed the lowest LMA values. LMA also increased with foliage age, suggesting a developmental change in leaf area and mass with increasing foliage age. Multiple linear regression (MLR) relationships were developed to predict LMA at various positions within tree crowns. The precision of the models was slightly greater when branch position (from the apex of the tree) was described relative to tree height, rather than relative to crown length. The MLR function resulted in precise representations of LMA within tree crowns.

Comparing strategies for modeling individual-tree height and height-to-crown base increment in mixed-species Acadian forests of northeastern North America

Various methods for predicting annual tree height increment (∆ht) and height-to-crown base increment (∆hcb) were developed and evaluated using remeasured data from permanent sample plots compiled across the Acadian Forest of northeastern North America. Across these plots, 25 species were represented upon which total height (ht) measurements were collected from mixed-species stands displaying both single- and multi-cohort structures. For modeling ∆ht, development of a unified equation form was found to result in higher accuracy and less bias compared to a maximum-modifier approach. Incorporating species as a random effect resulted in predictions that were not significantly different compared to predictions from species-specific equations for nine of the ten most abundant species examined. For ∆hcb, equations that modeled changes in hcb over two time periods (i.e., an incremental approach) were either not significantly different from or significantly closer to zero compared to predictions that estimated hcb at two time periods (i.e., a static approach). Results highlight the advantages of incorporating species as a random effect in individual-tree models and demonstrate the effectiveness of modeling tree crown recession directly for application in mixed-species forest growth and yield models.

Response of Douglas-fir leaf area index and litterfall dynamics to Swiss needle cast in north coastal Oregon, USA

Sources of variation in leaf area index (LAI; m2 of projected leaf area per m2 of ground area) and its seasonal dynamics are not well known in managed Douglas-fir stands, despite the importance of leaf area in forecasting forest growth, particularly in stands impacted by insects or disease. The influence of Swiss needle cast (SNC) on coastal Douglas-fir (Pseudotsuga menziesii var. menziesii [Mirb] Franco) LAI and litterfall dynamics was quantified by destructively sampling 122 stems from 36 different permanent plots throughout north coastal Oregon, USA, and by monitoring litterfall for 3 years in 15 of these plots. LAI, total annual litterfall, and the seasonal distribution of foliage and fine woody litterfall were all influenced by stand structural attributes, physiographic features, and SNC severity. Mean LAI in this study was 5.44 ± 2.16. The relatively low LAIs were attributed primarily to the effects of SNC on foliage retention, and secondarily to its direct measurement by hierarchical foliage sampling in contrast to indirect measurement by light interception or tree allometry. For a given stand structure and SNC severity, LAI was 36% greater in the fall after current year foliage was fully developed and older aged classes had not yet senesced. Annual litterfall expressed as a proportion of LAI at the start of the growing season varied from 0.13 to 0.53 and declined with increasing initial LAI. SNC also shifted more of the annual foliage litterfall to earlier in the spring. Fine woody litterfall experienced a different seasonal shift as the peak occurred later in the year on sites with high SNC, but this only occurred on northerly aspects. Defoliation from the endemic SNC pathogen can drastically reduce LAI and change both total and seasonal foliage litterfall patterns.RésuméLes sources de variation de l’index foliaire (LAI, m2 de surface projetée des feuilles par m2 de surface de sol) et sa dynamique saísonnière ne sont pas bien connues dans les peuplements aménagés de Douglas, malgré l’importance de la surface foliaire dans les prévisions de la croissance des forêts, particulièrement dans les peuplements touchés par des insectes ou les maladies. L’influence de la rouille suisse (SNC) sur l’index foliaire et la dynamique de chute de litière de Pseudotsuga menziesii var. menziesii [Mirb.] Franco ont été quantifiées grâce à un échantillonnage destructif de 122 tiges dans 36 placeaux permanents dans la zone côtière du Nord Oregon (USA) et le suivi pendant 3 ans des chutes de litière dans 15 de ces placeaux. L’index foliaire, la chute annuelle totale de litière, et la distribution du feuillage et la litière ligneuse fine ont tous été influencés par les attributs structuraux, les caractéristiques physiographiques et la gravité de SNC. Dans cette étude la moyenne de l’index foliaire était de 5,44 ± 2,16. Les index foliaires relativement faibles ont été essentiellement attribués aux effets de SNC sur le maintien du feuillage, et secondairement sur ses mesures directes par un échantillonnage hiérarchisé par opposition aux mesures indirectes par interception de la lumière ou par des méthodes d’allométrie au niveau des arbres. Pour une structure de peuplement et une gravité de SNC données, l’index foliaire a été 36 % plus élevé à l’automne après le plein développement du feuillage de l’année en cours et avant la sénescence des classes plus âgées. La chute annuelle de litière exprimée en proportion de l’index foliaire au début de la saison de croissance a varié de 0,13 à 0,53 et a baissé avec l’augmentation de l’index foliaire initial. La SNC a aussi enlevé plus que la chute annuelle de feuillage de la litière plus tôt au printemps. La litière ligneuse fine a été rencontrée à différents moments dans la saison alors que le pic s’est produit plus tard dans l’année dans les sites présentant une SNC élevée, mais ceci s’est seulement produit dans les expositions au nord. La défoliation par le pathogène endémique SNC peut réduire considérablement l’index foliaire et change à la fois les modèles de chute totale et de chute saisonnière de litière.

Layer Stacking: A Novel Algorithm for Individual Forest Tree Segmentation from LiDAR Point Clouds

Abstract As light detection and ranging (LiDAR) technology advances, it has become common for datasets to be acquired at a point density high enough to capture structural information from individual trees. To process these data, an automatic method of isolating individual trees from a LiDAR point cloud is required. Traditional methods for segmenting trees attempt to isolate prominent tree crowns from a canopy height model. We here introduce a novel segmentation method, layer stacking, which slices the entire forest point cloud at 1-m height intervals and isolates trees in each layer. Merging the results from all layers produces representative tree profiles. When compared to watershed delineation (a widely used segmentation algorithm), layer stacking correctly identified 15% more trees in uneven-aged conifer stands, 7%–17% more in even-aged conifer stands, 26% more in mixedwood stands, and 26%–30% more (with 75% of trees correctly detected) in pure deciduous stands. Overall, layer stackings commission error was mostly similar to or better than that of watershed delineation. Layer stacking performed particularly well in deciduous, leaf-off conditions, even those where tree crowns were less prominent. We conclude that in the tested forest types, layer stacking represents an improvement in segmentation when compared to existing algorithms.