Predicting forest site productivity in temperate lowland from forest floor, soil and litterfall characteristics using boosted regression trees

Aertsen, Wim; Kint, Vincent; Vos, Bruno De; Deckers, Jozef; Orshoven, Jos Van; Muys, Bart

doi:10.1007/s11104-011-1052-z

Cited by 65 publications

(53 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also during data exploration all predictor variables were tested for ecologically acceptable level of collinearity (i.e. individual variance inflation factor (VIF) of less than five) between predictor variables (Zuur et al, 2010;Aertsen et al, 2012). In construction of BRT models, flea indices were used as response variable and individual land uses were used as predictor variables.…”

Section: Discussionmentioning

confidence: 99%

Contribution of land use to rodent flea load distribution in the plague endemic area of Lushoto District, Tanzania

Hieronimo

Kihupi

Kimaro

et al. 2014

Tanzania J Hlth Res

View full text Add to dashboard Cite

Fleas associated with different rodent species are considered as the major vectors of bubonic plague, which is still rampant in different parts of the world. The objective of this study was to investigate the contribution of land use to rodent flea load distribution at fine scale in the plague endemic area of north-eastern Tanzania. Data was collected in three case areas namely, Shume, Lukozi and Mwangoi, differing in plague incidence levels. Data collection was carried out during both wet and dry seasons of 2012. Analysis of Variance and Boosted Regression Tree (BRT) statistical methods were used to clarify the relationships between fleas and specific land use characteristics. There was a significant variation (P ≤ 0.05) of flea indices in different land use types. Fallow and natural forest had higher flea indices whereas plantation forest mono-crop and mixed annual crops had the lowest flea indices among the aggregated land use types. The influence of individual land use types on flea indices was variable with fallow having a positive effect and land tillage showing a negative effect. The results also demonstrated a seasonal effect, part of which can be attributed to different land use practices such as application of pesticides, or the presence of grass strips around fields. These findings suggest that land use factors have a major influence on rodent flea abundance which can be taken as a proxy for plague infection risk. The results further point to the need for a comprehensive package that includes land tillage and crop type considerations on one hand and the associated human activities on the other, in planning and implementation of plague control interventions.

show abstract

Section: Discussionmentioning

confidence: 99%

Contribution of land use to rodent flea load distribution in the plague endemic area of Lushoto District, Tanzania

Hieronimo

Kihupi

Kimaro

et al. 2014

Tanzania J Hlth Res

View full text Add to dashboard Cite

show abstract

“…Geocentric approaches have been used as a basis for estimating SI or other measures of forest productivity and involve relating SI to various direct and/or indirect environmental factors [1]. Many studies have revealed environmental predictors of SI, using edaphic [2][3][4][5], topographic [6][7][8], and/or climatic [9][10][11] variables. Geocentrically-based (biophysical) SI models are independent from stand age and structure, usually have satisfactory prediction power and, therefore, provide tools that can effectively inform forest management.…”

Section: Introductionmentioning

confidence: 99%

High Resolution Site Index Prediction in Boreal Forests Using Topographic and Wet Areas Mapping Attributes

Bjelanovic

Comeau

White

2018

Forests

View full text Add to dashboard Cite

Abstract:The purpose of this study was to evaluate the relationships between environmental factors and the site index (SI) of trembling aspen, lodgepole pine, and white spruce based on the sampling of temporary sample plots. LiDAR generated digital elevation models (DEM) and wet areas mapping (WAM) provided data at a 1 m resolution for the study area in Alberta. Six different catchment areas (CA), ranging from 0.5 ha to 10 ha, were tested to reveal optimal CA for calculation of the depth-to-water (DTW) index from WAM. Using different modeling methods, species-specific SI models were developed for three datasets: (1) topographic and wet area variables derived from DEM and WAM, (2) only WAM variables, and (3) field measurements of soil and topography. DTW was selected by each statistical method for each species and, in most cases, DTW was the strongest predictor in the model. In addition, differences in strength of relationships were found between species. Models based on remotely-sensed information predicted SI with a root mean squared error (RMSE) of 1.6 m for aspen and lodgepole pine, and 2 m for white spruce. This approach appears to adequately portray the variation in productivity at a fine scale and is potentially applicable to forest growth and yield modeling and silviculture planning.

show abstract

“…The main advantages of this method are that the BRT scheme can analyze different types of variables and interaction effects among variables, and is applicable to nonlinear relationships. In recent years, the BRT technique has been used to examine the distribution of soil characteristics at a regional scale (Aertsen et al, 2011;Cools et al, 2014;Martin et al, 2011). Major outputs from BRT analyses can identify the following: (1) the relative importance (percentage of influence or contribution) of predictor variables (explanatory variables) on the basis of the weighted and scaled number of times a variable is selected for splitting (Elith et al, 2008) and (2) the relationships among variables and the explained variable shown in partial dependence plots.…”

Section: Boosted Regression Trees (Brt) Schemementioning

confidence: 99%

Data-mining analysis of the global distribution of soil carbon in observational databases and Earth system models

et al. 2017

View full text Add to dashboard Cite

Abstract. Future climate change will dramatically change the carbon balance in the soil, and this change will affect the terrestrial carbon stock and the climate itself. Earth system models (ESMs) are used to understand the current climate and to project future climate conditions, but the soil organic carbon (SOC) stock simulated by ESMs and those of observational databases are not well correlated when the two are compared at fine grid scales. However, the specific key processes and factors, as well as the relationships among these factors that govern the SOC stock, remain unclear; the inclusion of such missing information would improve the agreement between modeled and observational data. In this study, we sought to identify the influential factors that govern global SOC distribution in observational databases, as well as those simulated by ESMs. We used a data-mining (machine-learning) (boosted regression trees -BRT) scheme to identify the factors affecting the SOC stock. We applied BRT scheme to three observational databases and 15 ESM outputs from the fifth phase of the Coupled Model Intercomparison Project (CMIP5) and examined the effects of 13 variables/factors categorized into five groups (climate, soil property, topography, vegetation, and land-use history). Globally, the contributions of mean annual temperature, clay content, carbon-to-nitrogen (CN) ratio, wetland ratio, and land cover were high in observational databases, whereas the contributions of the mean annual temperature, land cover, and net primary productivity (NPP) were predominant in the SOC distribution in ESMs. A comparison of the influential factors at a global scale revealed that the most distinct differences between the SOCs from the observational databases and ESMs were the low clay content and CN ratio contributions, and the high NPP contribution in the ESMs. The results of this study will aid in identifying the causes of the current mismatches between observational SOC databases and ESM outputs and improve the modeling of terrestrial carbon dynamics in ESMs. This study also reveals how a data-mining algorithm can be used to assess model outputs.

show abstract

Predicting forest site productivity in temperate lowland from forest floor, soil and litterfall characteristics using boosted regression trees

Cited by 65 publications

References 55 publications

Contribution of land use to rodent flea load distribution in the plague endemic area of Lushoto District, Tanzania

Contribution of land use to rodent flea load distribution in the plague endemic area of Lushoto District, Tanzania

High Resolution Site Index Prediction in Boreal Forests Using Topographic and Wet Areas Mapping Attributes

Data-mining analysis of the global distribution of soil carbon in observational databases and Earth system models

Contact Info

Product

Resources

About