skip to main content

Explore Research Products in the NSF-PAR

This content will become publicly available on June 1, 2024

Title: Scaling up uncertainties in allometric models: How to see the forest, not the trees
Quantifying uncertainty in forest assessments is challenging because of the number of sources of error and the many possible approaches to quantify and propagate them. The uncertainty in allometric equations has sometimes been represented by propagating uncertainty only in the prediction of individuals, but at large scales with large numbers of trees uncertainty in model fit is more important than uncertainty in individuals. We compared four different approaches to representing model uncertainty: a formula for the confidence interval, Monte Carlo sampling of the slope and intercept of the regression, bootstrap resampling of the allometric data, and a Bayesian approach. We applied these approaches to propagating model uncertainty at four different scales of tree inventory (10 to 10,000 trees) for four study sites with varying allometry and model fit statistics, ranging from a monocultural plantation to a multi-species shrubland with multi-stemmed trees. We found that the four approaches to quantifying uncertainty in model fit were in good agreement, except that bootstrapping resulted in higher uncertainty at the site with the fewest trees in the allometric data set (48), because outliers could be represented multiple times or not at all in each sample. The uncertainty in model fit did not vary with the number of trees in the inventory to which it was applied. In contrast, the uncertainty in predicting individuals was higher than model fit uncertainty when applied to small numbers of trees, but became negligible with 10,000 trees. The importance of this uncertainty source varied with the forest type, being largest for the shrubland, where the model fit was most poor. Low uncertainties were observed where model fit was high, as was the case in the monoculture plantation and in the subtropical jungle where hundreds of trees contributed to the allometric model. In all cases, propagating uncertainty only in the prediction of individuals would underestimate allometric uncertainty. It will always be most correct to include both uncertainty in predicting individuals and uncertainty in model fit, but when large numbers of individuals are involved, as in the case of national forest inventories, the contribution of uncertainty in predicting individuals can be ignored. When the number of trees is small, as may be the case in forest manipulation studies, both sources of allometric uncertainty are likely important and should be accounted for.  more » « less
Award ID(s):
1637685
NSF-PAR ID:
10491979
Author(s) / Creator(s):
; ; ; ;
Publisher / Repository:
Elsevier B.V.
Date Published:
Journal Name:
Forest Ecology and Management
Volume:
537
Issue:
C
ISSN:
0378-1127
Page Range / eLocation ID:
120943
Subject(s) / Keyword(s):
["Allometric uncertainty","bootstrap","Monte Carlo","Bayesian","forest carbon budget"]
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ( , Ecology and Evolution)
    Abstract

    Soil and litter arthropods represent a large proportion of tropical biodiversity and perform important ecosystem functions, but little is known about the efficacy of different tropical forest restoration strategies in facilitating their recovery in degraded habitats. We sampled arthropods in four 7‐ to 8‐year‐old restoration treatments and in nearby reference forests. Sampling was conducted during the wet and dry seasons using extractions from litter and pitfall samples. Restoration treatments were replicated in 50 × 50‐m plots in four former pasture sites in southern Costa Rica:plantation– trees planted throughout the plot;applied nucleation/islands– trees planted in patches of different sizes; andnatural regeneration– no tree planting. Arthropod abundance, measures of richness and diversity, and a number of functional groups were greater in the island treatment than in natural regeneration or plantation treatments and, in many cases, were similar to reference forest. Litter and pitfall morphospecies and functional group composition in all three restoration treatments were significantly different than reference sites, but island and plantation treatments showed more recovery than natural regeneration. Abundance and functional group diversity showed a much greater degree of recovery than community composition.Synthesis and applications: The less resource‐intensive restoration strategy of planting tree islands was more effective than tree plantations in restoring arthropod abundance, richness, and functional diversity. None of the restoration strategies, however, resulted in similar community composition as reference forest after 8 years of recovery, highlighting the slow rate of recovery of arthropod communities after disturbance, and underscoring the importance of conservation of remnant forests in fragmented landscapes.

     
    more » « less
  2. ; ; ; ( , Ecological Applications)
    Abstract

    In tropical rainforests, tree size and number density are influenced by disturbance history, soil, topography, climate, and biological factors that are difficult to predict without detailed and widespread forest inventory data. Here, we quantify tree size–frequency distributions over an old‐growth wet tropical forest at the La Selva Biological Station in Costa Rica by using an individual tree crown (ITC) algorithm on airborne lidar measurements. The ITC provided tree height, crown area, the number of trees >10 m height and, predicted tree diameter, and aboveground biomass from field allometry. The number density showed strong agreement with field observations at the plot‐ (97.4%; 3% bias) and tree‐height‐classes level (97.4%; 3% bias). The lidar trees size spectra of tree diameter and height closely follow the distributions measured on the ground but showed less agreement with crown area observations. The model to convert lidar‐derived tree height and crown area to tree diameter produced unbiased (0.8%) estimates of plot‐level basal area and with low uncertainty (6%). Predictions on basal area for tree height classes were also unbiased (1.3%) but with larger uncertainties (22%). The biomass estimates had no significant bias at the plot‐ and tree‐height‐classes level (−5.2% and 2.1%). Our ITC method provides a powerful tool for tree‐ to landscape‐level tropical forest inventory and biomass estimation by overcoming the limitations of lidar area‐based approaches that require local calibration using a large number of inventory plots.

     
    more » « less
  3. ; ; ; ( , Earth Surface Processes and Landforms)
    Abstract

    Fluvial processes strongly influence riparian forests through rapid and predictable shifts in dominant species, tree density and size that occur in the decades following large floods. Modelling riparian forest characteristics based on the age and evolution of floodplains is useful in predicting ecosystem functions that depend on the size and density of trees, including large wood delivered to river channels, forest biomass and habitat quality. We developed a dynamic model of riparian forest structure that predicts changes in tree size and density using floodplain age derived from air photos and historical maps. Using field data and a riparian forest chronosequence for the 160‐km middle reach of the Sacramento River (California, USA), we fit Weibull diameter distributions with time‐varying parameters to the empirical data. Species were stratified into early and late successional groups, each with time‐varying functions of tree density and diameter distributions. From these, we modelled how the number and size of trees in a stand changed throughout forest succession, and evaluated the goodness‐of‐fit of model predictions.

    Model outputs for the early successional group, composed primarily of cottonwoods and willows, accounted for most of the stand basal area and large trees >10 cm DBH for the first 50 years. Post‐pioneer species with slower growth had initially low densities that increased slowly from the time of floodplain creation. Within the first 100 years, early successional trees contributed the most large wood that could influence fluvial processes, carbon storage, and instream habitat. We applied the model to evaluate the potential large wood inputs to the middle Sacramento River under a range of historical bank migration rates. Going forward, this modelling approach can be used to predict how riparian forest structure and other ecosystem benefits such as carbon sequestration and habitat quality respond to different river management and restoration actions.

     
    more » « less
  4. ; ; ; ; ; ; ; ; ; ( , Remote Sensing in Ecology and Conservation)
    Abstract

    Measuring forest biodiversity using terrestrial surveys is expensive and can only capture common species abundance in large heterogeneous landscapes. In contrast, combining airborne imagery with computer vision can generate individual tree data at the scales of hundreds of thousands of trees. To train computer vision models, ground‐based species labels are combined with airborne reflectance data. Due to the difficulty of finding rare species in a large landscape, many classification models only include the most abundant species, leading to biased predictions at broad scales. For example, if only common species are used to train the model, this assumes that these samples are representative across the entire landscape. Extending classification models to include rare species requires targeted data collection and algorithmic improvements to overcome large data imbalances between dominant and rare taxa. We use a targeted sampling workflow to the Ordway Swisher Biological Station within the US National Ecological Observatory Network (NEON), where traditional forestry plots had identified six canopy tree species with more than 10 individuals at the site. Combining iterative model development with rare species sampling, we extend a training dataset to include 14 species. Using a multi‐temporal hierarchical model, we demonstrate the ability to include species predicted at <1% frequency in landscape without losing performance on the dominant species. The final model has over 75% accuracy for 14 species with improved rare species classification compared to 61% accuracy of a baseline deep learning model. After filtering out dead trees, we generate landscape species maps of individual crowns for over 670 000 individual trees. We find distinct patches of forest composed of rarer species at the full‐site scale, highlighting the importance of capturing species diversity in training data. We estimate the relative abundance of 14 species within the landscape and provide three measures of uncertainty to generate a range of counts for each species. For example, we estimate that the dominant species,Pinus palustrisaccounts for c. 28% of predicted stems, with models predicting a range of counts between 160 000 and 210 000 individuals. These maps provide the first estimates of canopy tree diversity within a NEON site to include rare species and provide a blueprint for capturing tree diversity using airborne computer vision at broad scales.

     
    more » « less
  5. ; ; ; ( , Forests)
    Tree mortality is a key process in forest dynamics. Despite decades of effort to understand this process, many uncertainties remain. South American broadleaf species are particularly under-represented in global studies on mortality and forest dynamics. We sampled monospecific broadleaf Nothofagus pumilio forests in northern Patagonia to predict tree mortality based on stem growth. Live or dead conditions in N. pumilio trees can be predicted with high accuracy using growth rate as an explanatory variable in logistic models. In Paso Córdova (CO), Argentina, where the models were calibrated, the probability of death was a strong negative function of radial growth, particularly during the six years prior to death. In addition, negative growth trends during 30 to 45 years prior to death increased the accuracy of the models. The CO site was affected by an extreme drought during the summer 1978–1979, triggering negative trends in radial growth of many trees. Individuals showing below-average and persistent negative trends in radial growth are more likely to die than those showing high growth rates and positive growth trends in recent decades, indicating the key role of droughts in inducing mortality. The models calibrated at the CO site showed high verification skill by accurately predicting tree mortality at two independent sites 76 and 141 km away. Models based on relative growth rates showed the highest and most balanced accuracy for both live and dead individuals. Thus, the death of individuals across different N. pumilio sites was largely determined by the growth rate relative to the total size of the individuals. Our findings highlight episodic severe drought as a triggering mechanism for growth decline and eventual death for N. pumilio, similar to results found previously for several other species around the globe. In the coming decades, many forests globally will be exposed to more frequent and/or severe episodes of reduced warm-season soil moisture. Tree-ring studies such as this one can aid prediction of future changes in forest productivity, mortality, and composition. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email