Mapping multi-scale vascular plant richness in a forest landscape with integrated LiDAR and hyperspectral remote-sensing

Exploring the role of biotic factors in regulating the spatial variability in land surface phenology across four temperate forest sites

Land surface phenology (LSP), the characterization of plant phenology with satellite data, is essential for understanding the effects of climate change on ecosystem functions. Considerable LSP variation is observed within local landscapes, and the role of biotic factors in regulating such variation remains underexplored. In this study, we selected four National Ecological Observatory Network terrestrial sites with minor topographic relief to investigate how biotic factors regulate intra-site LSP variability. We utilized plant functional type (PFT) maps, functional traits, and LSP data to assess the explanatory power of biotic factors for the start and end of season (SOS and EOS) variability. Our results indicate that PFTs alone explain only 0.8-23.4% of intra-site SOS and EOS variation, whereas including functional traits significantly improves explanatory power, with cross-validation correlations ranging from 0.50 to 0.85. While functional traits exhibited diverse effects on SOS and EOS across different sites, traits related to competitive ability and productivity were important for explaining both SOS and EOS variation at these sites. These findings reveal that plants exhibit diverse phenological responses to comparable environmental conditions, and functional traits significantly contribute to intra-site LSP variability, highlighting the importance of intrinsic biotic properties in regulating plant phenology.

LiDAR GEDI derived tree canopy height heterogeneity reveals patterns of biodiversity in forest ecosystems

The "Height Variation Hypothesis" is an indirect approach used to estimate forest biodiversity through remote sensing data, stating that greater tree height heterogeneity (HH) measured by CHM LiDAR data indicates higher forest structure complexity and tree species diversity. This approach has traditionally been analyzed using only airborne LiDAR data, which limits its application to the availability of the dedicated flight campaigns. In this study we analyzed the relationship between tree species diversity and HH, calculated with four different heterogeneity indices using two freely available CHMs derived from the new space-borne GEDI LiDAR data. The first, with a spatial resolution of 30 m, was produced through a regression tree machine learning algorithm integrating GEDI LiDAR data and Landsat optical information. The second, with a spatial resolution of 10 m, was created using Sentinel-2 images and a deep learning convolutional neural network. We tested this approach separately in 30 forest plots situated in the northern Italian Alps, in 100 plots in the forested area of Traunstein (Germany) and successively in all the 130 plots through a cross-validation analysis. Forest density information was also included as influencing factor in a multiple regression analysis. Our results show that the GEDI CHMs can be used to assess biodiversity patterns in forest ecosystems through the estimation of the HH that is correlated to the tree species diversity. However, the results also indicate that this method is influenced by different factors including the GEDI CHMs dataset of choice and their related spatial resolution, the heterogeneity indices used to calculate the HH and the forest density. Our finding suggest that GEDI LIDAR data can be a valuable tool in the estimation of forest tree heterogeneity and related tree species diversity in forest ecosystems, which can aid in global biodiversity estimation.

Delineation of Geomorphological Woodland Key Habitats Using Airborne Laser Scanning

Forest ecosystems provide a range of services and function as habitats for many species. The concept of woodland key habitats (WKH) is important for biodiversity management in forest planning standards and certification schemes. The main idea of the WKH is to preserve biodiversity hotspots in the forest landscape. Current methods used in delineating WKH rely on costly field inventories. Furthermore, it is well known that the surveyor introduces an error because of the subjective assessment. Remote sensing may reduce this error in a cost-efficient way. The current study develops automated methods using airborne laser scanning (ALS) data to delineate geomorphological WKH, i.e., rock walls and stream gorges. The methods were evaluated based on a complete field inventory of WKH in a 1600 ha area in south-eastern Norway. The delineated WKH showed high detection rates, minor omission errors, but high commissions errors. Combining the delineation into a map of potential WKH suitable to guide field surveyors resulted in detecting all field reference WKH, i.e., a detection rate of 100% and a commission error of 25%. It is concluded that a higher degree of automatization might be possible to improve results and increase the efficiency of WKH inventories.

Mapping Potential Plant Species Richness over Large Areas with Deep Learning, MODIS, and Species Distribution Models

The spatial patterns of species richness can be used as indicators for conservation and restoration, but data problems, including the lack of species surveys and geographical data gaps, are obstacles to mapping species richness across large areas. Lack of species data can be overcome with remote sensing because it covers extended geographic areas and generates recurring data. We developed a Deep Learning (DL) framework using Moderate Resolution Imaging Spectroradiometer (MODIS) products and modeled potential species richness by stacking species distribution models (S-SDMs) to ask, “What are the spatial patterns of potential plant species richness across the Korean Peninsula, including inaccessible North Korea, where survey data are limited?” First, we estimated plant species richness in South Korea by combining the probability-based SDM results of 1574 species and used independent plant surveys to validate our potential species richness maps. Next, DL-based species richness models were fitted to the species richness results in South Korea, and a time-series of the normalized difference vegetation index (NDVI) and leaf area index (LAI) from MODIS. The individually developed models from South Korea were statistically tested using datasets that were not used in model training and obtained high accuracy outcomes (0.98, Pearson correlation). Finally, the proposed models were combined to estimate the richness patterns across the Korean Peninsula at a higher spatial resolution than the species survey data. From the statistical feature importance tests overall, growing season NDVI-related features were more important than LAI features for quantifying biodiversity from remote sensing time-series data.

Estimating the Growing Stem Volume of Coniferous Plantations Based on Random Forest Using an Optimized Variable Selection Method

Forest growing stem volume (GSV) reflects the richness of forest resources as well as the quality of forest ecosystems. Remote sensing technology enables robust and efficient GSV estimation as it greatly reduces the survey time and cost while facilitating periodic monitoring. Given its red edge bands and a short revisit time period, Sentinel-2 images were selected for the GSV estimation in Wangyedian forest farm, Inner Mongolia, China. The variable combination was shown to significantly affect the accuracy of the estimation model. After extracting spectral variables, texture features, and topographic factors, a stepwise random forest (SRF) method was proposed to select variable combinations and establish random forest regressions (RFR) for GSV estimation. The linear stepwise regression (LSR), Boruta, Variable Selection Using Random Forests (VSURF), and random forest (RF) methods were then used as references for comparison with the proposed SRF for selection of predictors and GSV estimation. Combined with the observed GSV data and the Sentinel-2 images, the distributions of GSV were generated by the RFR models with the variable combinations determined by the LSR, RF, Boruta, VSURF, and SRF. The results show that the texture features of Sentinel-2’s red edge bands can significantly improve the accuracy of GSV estimation. The SRF method can effectively select the optimal variable combination, and the SRF-based model results in the highest estimation accuracy with the decreases of relative root mean square error by 16.4%, 14.4%, 16.3%, and 10.6% compared with those from the LSR-, RF-, Boruta-, and VSURF-based models, respectively. The GSV distribution generated by the SRF-based model matched that of the field observations well. The results of this study are expected to provide a reference for GSV estimation of coniferous plantations.

Tree canopy cover constrains the fertility-diversity relationship in plant communities of the southeastern United States

The goal of elucidating the primary mechanisms constraining the assembly and distribution of biodiversity remains among the central unresolved challenges facing the field of ecology. Simulation studies and experimental manipulations have focused on how patterns in community assembly result from bivariate relationships along productivity or environmental gradients. However, the joint influence of multiple resource gradients on the distribution of species richness in natural communities remains understudied. Using data from a large network of multiscale vegetation plots across forests and woodlands of the southeastern United States, we find significant evidence for the scale-dependent, joint constraints of forest structure and soil resources on the distribution of vascular plant species richness. In addition to their significant partial effects on species richness, understory light levels and soil fertility positively interact, suggesting a trade-off between the two limiting resources with species richness peaking both in high-light, low-fertility conditions as well as low-light, high-fertility settings. This finding provides a novel perspective on the biodiversity-productivity relationship that suggests a transition in limiting resources from soil nutrients to light availability when enhanced productivity results in reduced light resources for subordinate individuals. Results likewise have meaningful implications for our understanding of scale-dependent community assembly processes as size-asymmetric competition replaces environmental filtering as the primary assembly mechanism structuring temperate forest communities along an increasing soil fertility gradient.

Deep Learning Approaches for the Mapping of Tree Species Diversity in a Tropical Wetland Using Airborne LiDAR and High-Spatial-Resolution Remote Sensing Images

The monitoring of tree species diversity is important for forest or wetland ecosystem service maintenance or resource management. Remote sensing is an efficient alternative to traditional field work to map tree species diversity over large areas. Previous studies have used light detection and ranging (LiDAR) and imaging spectroscopy (hyperspectral or multispectral remote sensing) for species richness prediction. The recent development of very high spatial resolution (VHR) RGB images has enabled detailed characterization of canopies and forest structures. In this study, we developed a three-step workflow for mapping tree species diversity, the aim of which was to increase knowledge of tree species diversity assessment using deep learning in a tropical wetland (Haizhu Wetland) in South China based on VHR-RGB images and LiDAR points. Firstly, individual trees were detected based on a canopy height model (CHM, derived from LiDAR points) by the local-maxima-based method in the FUSION software (Version 3.70, Seattle, USA). Then, tree species at the individual tree level were identified via a patch-based image input method, which cropped the RGB images into small patches (the individually detected trees) based on the tree apexes detected. Three different deep learning methods (i.e., AlexNet, VGG16, and ResNet50) were modified to classify the tree species, as they can make good use of the spatial context information. Finally, four diversity indices, namely, the Margalef richness index, the Shannon–Wiener diversity index, the Simpson diversity index, and the Pielou evenness index, were calculated from the fixed subset with a size of 30 × 30 m for assessment. In the classification phase, VGG16 had the best performance, with an overall accuracy of 73.25% for 18 tree species. Based on the classification results, mapping of tree species diversity showed reasonable agreement with field survey data (R2Margalef = 0.4562, root-mean-square error RMSEMargalef = 0.5629; R2Shannon–Wiener = 0.7948, RMSEShannon–Wiener = 0.7202; R2Simpson = 0.7907, RMSESimpson = 0.1038; and R2Pielou = 0.5875, RMSEPielou = 0.3053). While challenges remain for individual tree detection and species classification, the deep-learning-based solution shows potential for mapping tree species diversity.

Potential biodiversity map of understory plants for Nothofagus forests in Southern Patagonia: Analyses of landscape, ecological niche and conservation values

The role of understory plants in native forests is critical for ecosystem function, wildlife protection and ecosystem productivity. The interest to estimate biodiversity increased during the last decades at landscape level. The objective was to elaborate a map of potential biodiversity (MPB) of understory species of Nothofagus forest using potential habitat suitability maps (PHS) of 15 plants in Santa Cruz province, Argentina. Additionally, we asked the following questions: (i) Were plant species differentially distributed according to the forest types?, (ii) do forest types represent different plant species assemblage with specific ecological niche requirements?, and (iii) is it possible to detect hotspots in the MBP according to the forest types? We used 721 plots database of vascular plants, from where 15 indicator species were identified. The assemblage species for different forests (Nothofagus antarctica, N. pumilio and evergreen mixed) were analysed using a detrended correspondence analysis. Also, we explored 41 potential explanatory variables to develop PHS, and combined these maps to obtain one MPB (1-100%). Finally, we analysed the outputs into a GIS through different landscapes alternatives to detect hotspot areas. Marginality and specialization values allowed identifying species assemblage that presented similar variability in the habitat requirements. MPB varied across the landscape, with higher values in the south and lower values near glaciers. MPB had the highest values in N. antarctica forest with >50% cover at landscape level. N. antarctica present more hotspots than N. pumilio forests, mainly in the south, compared to mixed evergreen forests which present few hotspots near glaciers. These results can be used as a tool to design new management and conservation strategies at landscape level.

Hyperspectral and LiDAR Data Fusion Classification Using Superpixel Segmentation-Based Local Pixel Neighborhood Preserving Embedding

A new method of superpixel segmentation-based local pixel neighborhood preserving embedding (SSLPNPE) is proposed for the fusion of hyperspectral and light detection and ranging (LiDAR) data based on the extinction profiles (EPs), superpixel segmentation and local pixel neighborhood preserving embedding (LPNPE). A new workflow is proposed to calibrate the Goddard’s LiDAR, hyperspectral and thermal (G-LiHT) data, which allows our method to be applied to actual data. Specifically, EP features are extracted from both sources. Then, the derived features of each source are fused by the SSLPNPE. Using the labeled samples, the final label assignment is produced by a classifier. For the open standard experimental data and the actual data, experimental results prove that the proposed method is fast and effective in hyperspectral and LiDAR data fusion.

Do High-Voltage Power Transmission Lines Affect Forest Landscape and Vegetation Growth: Evidence from a Case for Southeastern of China

The rapid growth of the network of high-voltage power transmission lines (HVPTLs) is inevitably covering more forest domains. However, no direct quantitative measurements have been reported of the effects of HVPTLs on vegetation growth. Thus, the impacts of HVPTLs on vegetation growth are uncertain. Taking one of the areas with the highest forest coverage in China as an example, the upper reaches of the Minjiang River in Fujian Province, we quantitatively analyzed the effect of HVPTLs on forest landscape fragmentation and vegetation growth using Landsat imageries and forest inventory datasets. The results revealed that 0.9% of the forests became edge habitats assuming a 150 m depth-of-edge-influence by HVPTLs, and the forest plantations were the most exposed to HVPTLs among all the forest landscape types. Habitat fragmentation was the main consequence of HVPTL installation, which can be reduced by an increase in the patch density and a decrease in the mean patch area (MA), largest patch index (LPI), and effective mesh size (MESH). In all the landscape types, the forest plantation and the non-forest land were most affected by HVPTLs, with the LPI values decreasing by 44.1 and 20.8%, respectively. The values of MESH decreased by 44.2 and 32.2%, respectively. We found an obvious increasing trend in the values of the normalized difference vegetation index (NDVI) in 2016 and NDVI growth during the period of 2007 to 2016 with an increase in the distance from HVPTL. The turning points of stability were 60 to 90 meters for HVPTL corridors and 90 to 150 meters for HVPTL pylons, which indicates that the pylons have a much greater impact on NDVI and its growth than the lines. Our research provides valuable suggestions for vegetation protection, restoration, and wildfire management after the construction of HVPTLs.

Predictive Ecosystem Mapping of South-Eastern Australian Temperate Forests Using Lidar-Derived Structural Profiles and Species Distribution Models

Modern approaches to predictive ecosystem mapping (PEM) have not thoroughly explored the use of ‘characteristic’ gradients, which describe vegetation structure (e.g., light detection and ranging (lidar)-derived structural profiles). In this study, we apply a PEM approach by classifying the dominant stand types within the Central Highlands region of south-eastern Australia using both lidar and species distribution models (SDMs). Similarity percentages analysis (SIMPER) was applied to comprehensive floristic surveys to identify five species which best separated stand types. The predicted distributions of these species, modelled using random forests with environmental (i.e., climate, topography) and optical characteristic gradients (Landsat-derived seasonal fractional cover), provided an ecological basis for refining stand type classifications based only on lidar-derived structural profiles. The resulting PEM model represents the first continuous distribution map of stand types across the study region that delineates ecotone stands, which are seral communities comprised of species typical of both rainforest and eucalypt forests. The spatial variability of vegetation structure incorporated into the PEM model suggests that many stand types are not as continuous in cover as represented by current ecological vegetation class distributions that describe the region. Improved PEM models can facilitate sustainable forest management, enhanced forest monitoring, and informed decision making at landscape scales.