Next Article in Journal
Genome-Wide Identification of the Uridine Diphosphate Glucotransferase Gene Family and Expression Profiling Analysis in the Stem Development of Prunus mume
Previous Article in Journal
Construction and Application of Urban Green Space Ecosystem Service Assessment Indicator System and Assessment Method: A Case Study of Chifeng Central Urban Area, China
Previous Article in Special Issue
Land Use Evolution and Its Driving Factors over the Past 30 Years in Luochuan County
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Assessing the Impacts of Selective Logging on the Forest Understory in the Amazon Using Airborne LiDAR

by
Leilson Ferreira
1,2,3,4,*,
Edilson de Souza Bias
4,
Quétila Souza Barros
5,
Luís Pádua
2,3,6,
Eraldo Aparecido Trondoli Matricardi
7 and
Joaquim J. Sousa
3,8
1
Agronomy Department, School of Agrarian and Veterinary Sciences, University of Trás-os-Montes e Alto Douro, 5000-801 Vila Real, Portugal
2
Centre for the Research and Technology of Agro-Environmental and Biological Sciences, University of Trás-os-Montes e Alto Douro, 5000-801 Vila Real, Portugal
3
Engineering Department, School of Science and Technology, University of Trás-os-Montes e Alto Douro, 5000-801 Vila Real, Portugal
4
Applied Geosciences and Geodynamics (Geoprocessing and Environmental Analysis) Institute of Geosciences, Campus Universitário Darcy Ribeiro, University of Brasília, Brasília 70919-970, Brazil
5
National Institute of Amazonian Research (INPA), Acre Research Nucleus, Street Dias Martins, 3868, Rio Branco 69917-560, Brazil
6
Institute for Innovation, Capacity Building and Sustainability of Agri-Food Production, University of Trás-os-Montes e Alto Douro, 5000-801 Vila Real, Portugal
7
Forestry Department, College of Technology, University of Brasilia, Campus Darcy Ribeiro, Brasília 70910-900, Brazil
8
Centre for Robotics in Industry and Intelligent Systems (CRIIS), INESC Technology and Science (INESC-TEC), 4200-465 Porto, Portugal
*
Author to whom correspondence should be addressed.
Submission received: 26 November 2024 / Revised: 7 January 2025 / Accepted: 8 January 2025 / Published: 12 January 2025
(This article belongs to the Special Issue Sustainable Management of Forest Stands)

Abstract

:
Reduced-impact logging (RIL) has been recognized as a promising strategy for biodiversity conservation and carbon sequestration within sustainable forest management (SFM) areas. However, monitoring the forest understory—a critical area for assessing logging impacts—remains challenging due to limitations in conventional methods such as field inventories and global navigation satellite system (GNSS) surveys, which are time-consuming, costly, and often lack accuracy in complex environments. Additionally, aerial and satellite imagery frequently underestimate the full extent of disturbances as the forest canopy obscures understory impacts. This study examines the effectiveness of the relative density model (RDM), derived from airborne LiDAR data, for mapping and monitoring understory disturbances. A field-based validation of LiDAR-derived RDM was conducted across 25 sites, totaling 5504.5 hectares within the Jamari National Forest, Rondônia, Brazil. The results indicate that the RDM accurately delineates disturbances caused by logging infrastructure, with over 90% agreement with GNSS field data. However, the model showed the greatest discrepancy for skid trails, which, despite their lower accuracy in modeling, accounted for the largest proportion of the total impacted area among infrastructure. The findings include the mapping of 35.1 km of primary roads, 117.4 km of secondary roads, 595.6 km of skid trails, and 323 log landings, with skid trails comprising the largest proportion of area occupied by logging infrastructure. It is recommended that airborne LiDAR assessments be conducted up to two years post-logging, as impacts become less detectable over time. This study highlights LiDAR data as a reliable alternative to traditional monitoring approaches, with the ability to detect understory impacts more comprehensively for monitoring selective logging in SFM areas of the Amazon, providing a valuable tool for both conservation and climate mitigation efforts.

1. Introduction

Deforestation and degradation within the Amazon rainforest account for approximately 15% to 35% of anthropogenic carbon emissions, impacting the global climate substantially [1,2,3]. To address these impacts, signatories of the Paris Agreement have committed to reducing emissions associated with deforestation and forest degradation through initiatives such as REDD (reducing emissions from deforestation and forest degradation) [4,5,6]. In the Amazon, unsustainable logging practices, whether conducted illegally or through poorly regulated means, contribute to forest degradation [7,8,9]. In 2006, the Brazilian Forest Service (SFB) introduced a system of forest concessions to establish a legal framework for sustainable timber production and to combat illegal logging activities in the Amazon [10].
Brazil’s public forests cover 327.3 million hectares, with 1.3 million hectares currently under logging concessions in the Amazon region. The government aims to expand this coverage to 4 million hectares by 2026 [11]. Approximately 35 million hectares are potential concession areas, with 20 million hectares considered sufficient to sustainably meet the timber industry’s demands [7,12].
The Brazilian Forest Code [13] and the Public Forest Management Law [10] establish guidelines to balance conservation with forest use, mandating practices that protect biodiversity and enable regeneration on public and private lands. The Ministry of the Environment (MMA) has issued normative instructions [14], which require the implementation of inventories and continuous monitoring of forest species impacts. These laws require concessionaires to adopt reduced-impact logging (RIL) practices, which aim to reduce forest ecosystem damage by up to 50% compared to conventional logging methods, making RIL an important component for sustainable forest management [15,16,17,18,19,20,21,22].
Despite these existing regulations, studies reveal that current concession contracts fail to fully adhere to sustainability principles. An analysis of sustainable forest management plans (SFMPs) in the Amazon reveals technical and legal irregularities, with 82.3% of plans authorized for single-year production units, potentially facilitating unauthorized activities and highlighting the need for better logging practices [23,24,25]. Effective monitoring of selective logging impacts remains a priority as the demand for tropical wood rises and logging expands within public areas in the Amazon [7,13,26].
The majority of logging impacts occur in the forest understory, induced by falling trees, the removal of trees from the forest via skid trails, and structures such as roads and log landings [27,28,29,30]. The understory of tropical forests is essential for species regeneration, biodiversity maintenance, carbon storage, and nutrient cycling. This forest layer is characterized by low light, which regulates the growth of young plants that will eventually occupy the upper canopy, controls the regeneration of shade-tolerant and slow-growing species, and facilitates interactions essential to ecosystem dynamics [31,32,33]. The herbaceous species of the understory are linked to the forest’s environmental conditions, where they regulate the microclimate, light, humidity, and nutrients, providing resources for the survival of animals, fungi, and microorganisms [34,35].
After the creation of canopy gaps in selective logging, the availability of light increases, favoring forest regeneration and the development of successional stages typical of tropical forests, where recovery is generally rapid [36,37]. Under natural conditions, only 1%–2% of solar radiation reaches the understory, but in logged gaps, this proportion can reach 20%–35% [32,38]. This promotes the initial growth of light-dependent and fast-growing species, such as invasive plants (e.g., Guadua spp.), as they are more efficient than native species at using nutrients and sunlight [39,40,41,42,43]. These invasive species alter forest dynamics, intensify herbivory, and endanger the regeneration of shade-tolerant and slow-growing species [31,44]. Forest fragmentation and selective logging may amplify damage to biomass, reduce functional diversity and carbon reserves, and increase emissions and temperatures in the understory. In addition, these practices intensify the spread of fires, often promoted by invasive plant species, which increase the frequency and intensity of wildfires [39,45,46,47]. All of these impacts can last for decades, changing the composition and structure of the understory [33]. Therefore, monitoring the understory is crucial for obtaining an accurate assessment of forest disturbances.
Current monitoring methods typically use a combination of optical imagery and global navigation satellite system (GNSS) data to detect large-scale tree loss. Despite their widespread use in Earth observation, satellite platforms such as Landsat-8 and Sentinel-2 are not effective in identifying minor disturbances caused by selective logging [48,49,50,51]. This represents a significant challenge, as the forest’s closed canopy and understory regrowth obscure these impacts. Alternative remote sensing methods, such as unmanned aerial vehicles (UAVs), despite providing more detailed data due to its higher-spatial-resolution imagery for monitoring selective logging activities compared to coarser satellite imagery [52,53], can also suffer from the same canopy cover-related issues. The dense canopy can obstruct the view of the understory, making it difficult for UAVs to capture imagery of the forest floor [54]. Additionally, the survey area covered in a single flight is restricted by limited flight times due to battery life, and the deployment of UAVs in remote and rugged terrains can be costly and logistically challenging, especially for fixed-wing UAVs [55].
Traditional approaches that rely on field inventories and GNSS mapping are often restricted to small spatial extents, costly, and time-consuming [15,56,57]. Furthermore, these methods are frequently impeded by the arduous nature of accessing remote forested regions, particularly during the rainy season [58,59,60,61], which makes it unfeasible for teams to monitor forests with their equipment in the field to perform inventories and georeferencing in the Amazon rainforest. Additionally, these techniques often fail to capture the spatial distribution of logging impacts on the forest understory, resulting in an underestimation of the extent of selective logging disturbances [62,63,64].
To address these limitations, airborne light detection and ranging (LiDAR) technology comes as a viable technology as part of airborne laser scanners (ALSs), particularly the relative density model (RDM), for monitoring understory impacts from selective logging over large spatial extents. ALSs provide comprehensive data on vertical vegetation structures and have been used in multiple forest applications, including biomass measurement, carbon stock estimation, and the detection of forest disturbances [19,56,57,65,66,67,68,69]. D’Oliveira et al. [15] pioneered the use of RDM for mapping logging infrastructure, demonstrating its potential to capture detailed understory changes and improve selective logging monitoring.
This article aims to complement previous research [15] by using LiDAR data collected across approximately 5500 hectares, validated through fieldwork. The main objective is assessing the impact of logging operations on the understory in logging concession areas in the Amazon rainforest. Specifically, it is intended to (1) determine the efficacy of airborne LiDAR data in detecting and quantifying selective logging disturbances in the understory; (2) evaluate the accuracy of RDM to map primary and secondary roads, skid trails, log landings, and tree gaps; and (3) improve monitoring capabilities in sustainable management zones.

2. Materials and Methods

2.1. Study Area

This study was conducted within the Jamari National Forest (Jamari Flona), located in Rondônia, Brazil. Jamari Flona spans the municipalities of Itapuã do Oeste, Cujubim, and Candeias do Jamari, defined by the geographical coordinates S 09°00′00″ to 09°30′00″ and W 62°44′05″ to 63°16′54″. It is characterized by Ombrophilous Dense Forests with portions of Open Ombrophilous Forest formations that are commonly found across the Amazon. This protected area serves as an important resource for sustainable forest management under the SFB, providing a controlled setting for testing RIL practices. In terms of study areas, this study focused on forest management units (FMUs) I, II, and III, where data were collected from 25 selectively logged areas managed as a forest concession. Annual production units (APUs) within these areas were subjected to selective logging by using mechanical harvesting with forest tractors combined with RIL techniques. The mean logging intensity across the study area was 1.8 trees per hectare (±0.54), corresponding to a harvested volume of 13.8 m3 per hectare (±3.5). Figure 1 presents the spatial location of the study sites within Jamari Flona, delineating each FMU and highlighting areas selected for ALS data collection. APUs within the FMUs went through standard silvicultural treatments, and RIL practices aimed to minimize logging impacts on the forest understory. This configuration facilitated a controlled analysis of understory disturbances and allowed for accurate LiDAR-based assessments of post-harvest forest structure.

2.2. Data Acquisition and Processing

LiDAR data were collected using manned aircraft equipped with airborne laser scanning systems by specialized contractors under SFB supervision from 2011 to 2020 (Table 1). The data were provided in the form of point cloud files (*.las format) with a Universal Transverse Mercator (UTM) projection, Zone 20S, referenced to the SIRGAS 2000 datum (EPSG:31980).
The quality of the LiDAR data collected over the different periods (2011–2020) can be affected by factors such as variations in flight altitude, scan frequency, and return density, which may have an impact on subsequent analyses. To ensure consistency across datasets, a series of preprocessing steps were applied, including noise filtering, ground point classification, return density normalization, and spatial-resolution standardization to one meter. This process was conducted using FUSION software (version 4.61) [71] with commands executed via the Windows command line.
Field data were collected using a measuring tape and a random sampling approach to record the widths of forest exploitation infrastructure. The geolocation and trajectories of forest exploitation infrastructure were mapped in the field using GNSS equipment (GPSMAP® 76CSx, Garmin International, Inc., Olathe, KS, USA). The georeferencing of all roads within the forest was performed by recording GNSS coordinates during car-based field surveys. The trajectory of the skidder (machine used to drag logs) and the geolocation of the skid trails were also mapped by recording the GNSS coordinates during log dragging operations. For the log landing yards, the coordinates of the center of each log landing were recorded. Infrastructure that was already in place, such as roads, was excluded from consideration in the impact calculations. The GNSS receivers used in this study have an accuracy of between five to ten meters under ideal conditions. However, as with all GNSS receivers, their accuracy is unknown and highly variable in densely forested areas such as the Amazon rainforest [72].
Data were from the Commercial Forest Inventory (IF100%), including registration number, tree identification, tree diameter at breast height (DBH), and geolocations of felled trees, carried out by the concessionaires for the preparation of the annual operational plan (AOP) for each APU under examination. Technical guidelines from the SFMPs, as established by the Brazilian Institute of Environment and Renewable Natural Resources (IBAMA), Execution Standard No. 1/2007 [73], were used to ensure methodological consistency. The SFB provided supplementary data regarding the volume of timber harvested for each of the UPAs. These data were generated through the measurement (cubing) of the logs. The GNSS receiver used in this study was the Garmin 76MAP CSX model. The data were formatted in both vector and spreadsheet formats for subsequent analysis.
Given that the study areas were logged at different times, it was necessary to use LiDAR data collected at different times (Table 2), always after each area had been logged, in order to assess the impact of the forestry activity.
To map logging infrastructure and detect understory impacts caused by selective logging, RDMs were generated for all LiDAR point clouds using a one-meter spatial resolution. This approach was adapted from the methodology developed by D’Oliveira et al. [15]. Point clouds were processed in FUSION (version 4.61), which facilitated the calculation of relative vegetation density across defined height intervals, with resulting raster products being manipulated and analyzed in ArcGIS software (version 10.5) [74].
In the initial phase, noisy returns were filtered from the raw LiDAR point cloud data. These returns were identified as outliers due to their height, which did not correspond to vegetation (e.g., birds) or were random returns occurring in the atmosphere [75]. The GroundFilter command was used to filter the LIDAR returns. These points were then classified and filtered using the weighted least squares (WLS) algorithm [76]. The WLS algorithm integrates filtering and interpolation procedures [69]. In this algorithm, the deviation from the reference surface is calculated for each point using weights for each degree of deviation. Consequently, returns in which the deviations ( v i ) exceeded the tolerance value (g + w) were excluded. Those that fell between the threshold (g) and the tolerance value (g + w) received 100 weights ranging from 0 to 1, and those that were below the threshold value (g) received maximum weight and were the support for the new reference surface, as shown in (1). The components of the weighting function were determined as follows: g = −2, w = 2.5, a = 1, and b = 4.
p i = 1 1 1 + ( a     v i g b ) 0 v i g   g < v i g + w g + w < v i
The output of the GroundFilter is a file containing only the points classified as ground, stored in LAS format. This process enables the construction of a surface model using only the LiDAR points that reached the ground surface. Thus, a digital terrain model (DTM) was created using FUSION’s GridSurfaceCreate command. This tool employs random points to create a grid surface where the value of each cell is the average elevation in all the points within it [71]. In the absence of points within a given output cell, the cell is populated with the heights of adjacent cells [69]. The DTMs were generated with a resolution of one meter, ensuring compatibility with the point density of the LiDAR data collected.
The Cover command in FUSION software was used to calculate the proportion (%) of returns with elevations close to the ground, assessing the laser’s penetration into the forest [71]. Using the bare-ground surface model as a reference, all return elevations were normalized, allowing the computation of the proportion of returns within specific height ranges, thereby enabling the estimation of the relative vegetation density across different height strata. The RDM was calculated by dividing the number of returns in a selected stratum by the total number of points from the surface to the upper limit of that stratum, as illustrated in Figure 2. In this study, height limits were set between one and five meters, capturing the gaps caused by logging activities. Output values for density estimates range from 0.0 to 100.0 percent, representing the proportion of returns with elevations within each cell of the selected layer. Pixels with high RDM values indicate an intact understory, while low values represent vegetation loss.
Several density threshold simulations were performed until a result was visually consistent with the impact of logging for each area. Testing began with thresholds ranging from 0 to 5%, gradually increasing to 10% and beyond until a threshold was found that represented both linear infrastructure and continuous patches. Care was also taken to avoid selecting a threshold that introduced excessive noise, which could overestimate the impact on the understory. For this study, a density range of 0%–20% was chosen. This means that any 1 × 1 m cell with a relative density below 20% was classified as understory loss. This approach allowed for the identification and digitization of logging infrastructure (e.g., log landing, roads, trails, and gaps from harvested trees in the understory) as well as total impacted areas of the forest cover of the understory.

2.3. Study Area Infrastructure Mapping in the Understory

The mapping of logging infrastructure in the understory was conducted using RDMs as the base layer. Manual vectorization of all skid trails, roads, and log landings was performed for each area of interest at a scale of 1:3500. Roads and skid trails were distinguished by cross-referencing the planned road networks provided by concessionary companies and were approved by IBAMA.
Although automated linear feature extraction tools are available [77,78,79], manual digitization was preferred. Human interpreters are particularly skilled at integrating diverse spatial cues to form meaningful patterns, improving the reliability of understory mapping [80]. The primary interpreter possessed in-depth familiarity with the local landscape, gained through extensive fieldwork in logging operations. Manual digitization allowed for the exclusion of false signals in the RDM by cross-referencing additional data sources, including the digital terrain model (DTM), management plan, AOP, and other files. A secondary review by a senior analyst was conducted to ensure quality control and reduce false positives, providing reliable spatial data for subsequent analysis.
The area occupied by each type of logging infrastructure was estimated by multiplying the field-measured mean widths by the lengths digitized in the RDM. Density metrics for roads and trails were calculated by dividing the total length obtained from RDM by the total area sampled, expressed in meters per hectare (m.ha−1). Log landing areas were expressed in square meters per hectare (m2.ha−1). Validation of LiDAR mapping accuracy for understory infrastructure was achieved by comparing RDM-derived measurements to GNSS field data, with comparisons focused on the location and dimensions of skid trails, roads, and log landings. Positional accuracy was assessed by calculating the percentages of each infrastructure type within buffer zones defined by horizontal projections of 5, 10, and 20 m around the GNSS data points, as described by Goodchild and Hunter [81].

2.4. Total Understory Impact

Logging operations create gaps in the understory due to both tree felling and incidental damage that extends beyond the areas directly occupied by infrastructure. Impacted areas in the understory are defined here as zones of the forest cover exhibiting low relative vegetation density (<20%), particularly in the one-to-five-meter height range, close to the origin of the impact [82], and the proportion (%) of the cover removed in relation to the total area of the plot is calculated.
For this study, “impact zones” were generated to represent areas most likely affected by logging activities. In the RDM, disturbances from logging infrastructure, including roads, skid trails, log landings, and felled trees, are detectable alongside natural understory gaps created by environmental factors such as wind-felled trees [15]. These zones allow for a more nuanced analysis of anthropogenic impacts on the forest understory.
Buffers were created around each mapped RDM feature to define the extent of these impact zones, with buffer distances set to match field-work-observed dimensions for each type of infrastructure. Buffers started from the central axis and extended four meters for skid trails, six meters for secondary roads, 10 m for primary roads, 20 m for log landings, and 25 m for felled tree gaps, based on D’Oliveira et al. [15]. This buffered delineation approach aligns with previous studies and provides consistency in measuring infrastructure-related disturbances [81].
To prevent overlap between classifications, each infrastructure and felled tree gap was individually identified, ranked, and systematically ordered within the attribute table of the merged buffer file. Each intersecting zone was assigned a hierarchy: areas within log landings were classified as such; everything intercepted by the primary roads, except for landings, was considered primary roads; everything intercepted by the secondary roads, except landings and primary roads, was considered as secondary roads; everything intercepted by skid trails, except landings, primary roads, and secondary roads, was considered skid trails; and areas not superimposed on any of these sites was considered as felled tree gaps.

2.5. Statistical Analysis

Statistical analyses were conducted to evaluate the accuracy of LiDAR-derived RDMs in mapping logging infrastructure and understory impacts, as well as to assess correlations between understory impact metrics and various logging parameters. All analyses were performed using R software (version 4.3.3) [83]. Performance metrics for the RDM were evaluated using statistical parameters, including the root-mean-square error (RMSE) and the coefficient of determination (R2), were selected as metrics for evaluating the accuracy of LiDAR-derived measurements. RMSE was used to quantify the absolute differences between LiDAR and field-based GNSS measurements, providing a measure of the magnitude of errors in infrastructure mapping. The R2 value was chosen to assess the proportion of variance explained by the LiDAR-derived data relative to GNSS measurements, offering insights into the model’s predictive reliability. Together, these metrics provide complementary perspectives on the accuracy and robustness of the mapping approach.
Simple linear regression was performed for comparisons between field-based GNSS measurements (observed), and LiDAR (estimated) RDM data were conducted for each logging infrastructure type, including primary roads, secondary roads, skid trails, log landings, and canopy gaps created from felled trees. For each infrastructure category, RMSE and R2 values were computed to quantify the accuracy of RDM mapping relative to GNSS-based field observations, further supporting the evaluation of LiDAR as an effective tool for monitoring forest management impacts.
Statistical significance was assessed using the F-test, providing a basis for comparing RDM-derived measurements with those obtained via GNSS [84]. Pearson’s correlation test was employed to assess relationships between logging intensity, time since logging, and the extent of understory disturbance as indicated by RDM-derived values. Correlations were calculated to determine the strength and direction of associations, with all intervals reflecting a 95% confidence level [85]. These correlations were important for identifying patterns and implications of logging practices, particularly in assessing the temporal and spatial dynamics of the impacts on the understory.

3. Results

3.1. Infrastructure Mapping in the Understory

An RDM map illustrating the pathways of primary and secondary roads, skid trails, and landings in the study area, demonstrating the characteristic RIL pattern, is presented in Figure 3. Using RDM as a reference, a total of 117.4 km of secondary roads, 35.1 km of primary roads, 595.6 km of skid trails, and 323 log landings were mapped across the 5504.5-hectare sample area. Among the mapped infrastructure, skid trails showed the highest density, with a mean of 110.0 m.ha−1 (±7.44 m.ha−1), followed by secondary roads with a mean density of 20.55 m.ha−1 (±0.87 m.ha−1) and primary roads at 6.1 m.ha−1 (±0.86 m.ha−1). Log landings occupied a mean density of 33.58 m2.ha−1 (±2.75 m2.ha−1).
Variability in skid trail density was observed across harvesting sites. In areas where LiDAR data were collected within two years post-logging, RDM estimates of skid trail density were generally higher, with mean values surpassing 119 m.ha−1 (±7.20 m.ha−1). In areas where the time since logging exceeded two years, the average skid trail density was lower (90 m.ha−1), aligning with the expected forest regeneration over time.
A weak positive correlation was detected between logging intensity and skid trail density (R2 = 0.30, p = 0.004). Although the coefficient was low, the relationship between the two variables was significant, at a 5% probability. This correlation increased significantly (R2 = 0.61, p = 0.00003) when excluding areas where logging had occurred over two years prior to LiDAR data collection.
The mean area occupied by logging infrastructure in the understory was 6.0% ± 0.32%, most of which resulted from the opening of skid trails at 3.9% ± 0.27%, followed by the construction of secondary roads at 1.1% ± 0.05%, primary roads at 0.5% ± 0.07%, and log landings at 0.3% ± 0.03%.

3.2. Validation of Understory Structure Mapping

The accuracy of understory infrastructure mapping using LiDAR-derived RDMs was validated by comparing results to GNSS-based field measurements. Overall, the quantitative differences between RDM-derived and GNSS-mapped logging infrastructure were minimal, except for skid trails, which showed greater variability. LiDAR-based mapping estimated a mean skid trail density 27% higher than field-based GNSS measurements, with RDM estimates showing a mean density of 98.3 m.ha−1 compared to the GNSS-measured density of 77.4 m.ha−1 (p = 0.27). Despite this difference, statistical analysis indicated no significant variance between GNSS and LiDAR mean estimates for skid trails (p > 0.05), confirming the general reliability of RDMs in representing skid trail density. Moreover, the statistical analysis confirmed that variability in LiDAR acquisition parameters from the different epochs did not significantly influence the results (p > 0.05).
For other infrastructure types, such as primary roads, secondary roads, and log landings, RDM-based and GNSS-based mappings showed a closer alignment. RDM slightly underestimated infrastructure by 0.5% for secondary roads, 2.1% for primary roads, and 0.7% for log landings, with no statistically significant differences in these measurements (p > 0.05). Table 3 presents a comparison of the GNSS measurements and RDM estimations for each type of logging infrastructure.
In terms of positional accuracy, primary roads mapped using RDM showed 98% of the variance explained by GNSS measurements (R2 = 0.98), with an RMSE of 0.53 m.ha−1. Secondary roads displayed 75% of the variance explained by GNSS (R2 = 0.75), with an RMSE of 2.33 m.ha−1. Skid trails showed 68% of the variance explained by GNSS data (R2 = 0.68), with an RMSE of 26.8 m.ha−1. For log landings, 92% of the variance was explained by GNSS data (R2 = 0.92), with an RMSE of 0.003 n.ha−1.
The spatial overlap between RDM- and GNSS-mapped infrastructure was assessed across different buffer zones, considering GNSS receiver errors of up to 20 m (Table 4 and Figure 4). With a 5 m buffer, primary roads exhibited a 58% overlap, while secondary roads showed 55%, skid trails 44%, and log landings 55%. Increasing the buffer to 10 m improved overlap rates to 79% for primary roads, 82% for secondary roads, 60% for skid trails, and 65% for log landings. At a 20 m buffer, overlap further increased to 98% for primary roads, 96% for secondary roads, 74% for skid trails, and 92% for log landings.
The gaps present in the understory left by the felled trees were validated separately, as they are not part of the exploration infrastructure but are important elements for mapping the impact on the understory. The gaps agreed by 81% with the location (GNSS) of the trees logged, assuming an error of 20 m.

3.3. Total Understory Impact

The mean impact of logging on the forest understory was calculated to be 17.8% (±1.4%), with individual impact values ranging from 5.06% to 31.8% across the sampled areas, as shown in Table 5. Gaps created by felled trees represented the highest percentage of impact, contributing an average of 8.7% (±0.85%) to the total affected area, followed by skid trails, which accounted for 6.17% (±0.54%) of the impact.
An analysis of understory impact based on time elapsed since logging (Table 6) revealed higher mean impacts in areas surveyed within 12 months post-harvest. In these recently logged areas, the mean total impact was 19.8% (±1.8%), with gaps from felled trees comprising the largest disturbance (9.47% ± 1.12%), followed by skid trails (7.14% ± 0.63%). In areas surveyed 13 to 24 months after logging, the mean impact decreased to 15.3% (±3.0%), with felled tree gaps again being the most significant contributor to disturbance (6.75% ± 1.65%) and skid trails representing 5.77% (±1.0%). For areas with logging activities more than 24 months prior to LiDAR data collection, the mean impact further decreased to 14.0% (±3.0%), with felled tree gaps accounting for the highest percentage of disturbance (8.61% ± 2.24%), followed by skid trails at 3.13% (±0.8%). Further analysis based on logging intensity also demonstrated variations in understory impact, as summarized in Table 6.
Areas with a lower logging volume (8.0 to 12.0 m3.ha−1) had a mean impact of 13.1% (±0.95%), primarily driven by tree gaps (5.55% ± 0.61%) and skid trails (5.09% ± 0.39%). In areas with a logging intensity of 12.1 to 16.0 m3.ha−1, the mean impact increased to 17.5% (±1.9%), with tree gaps contributing 8.74% (±1.28%) and skid trails 5.62% (±0.75%). Areas with logging intensities exceeding 16.0 m3.ha−1 demonstrated the highest mean impact at 23.6% (±3.06%), with tree gaps and skid trails contributing 12.16% (±1.59%) and 8.18% (±1.32%), respectively.
The impact distribution also varied by FMU. FMU I suffered the greatest mean impact at 21.3% (±2.16%), driven primarily by felled tree gaps (11.08% ± 1.28%) and skid trails (6.90% ± 0.94%). In contrast, FMU II demonstrated a lower mean impact of 13.9%, with felled tree gaps contributing 6.06% and skid trails 5.46%. FMU III showed an intermediate impact level, with a mean of 14.6% (±1.48%), where gaps created by tree felling constituted the majority of disturbance (6.49% ± 0.81%), followed by skid trails (5.50% ± 0.58%).
Statistical analysis identified a significant negative correlation between total understory impact and the time elapsed since logging activities (r = −0.42, p = 0.03), as presented in Figure 5. This trend suggests that as the time interval between logging and LiDAR data collection increases, the observable impact on the understory decreases, likely due to natural forest regeneration processes. Skid trail impact exhibited a moderate negative correlation with time (r = −0.59, p = 0.002), along with similar correlations for log landings (r = −0.56, p = 0.003) and secondary roads (r = −0.59, p = 0.002). However, no significant correlation was detected for felled tree gaps (p = 0.28) or primary roads (p = 0.79), indicating that these impacts may persist longer within the understory.
Additionally, a positive correlation was observed between logging intensity and total understory impact (r = 0.61, p = 0.001). Within this analysis, tree gaps demonstrated a moderate positive correlation with logging volume (r = 0.66, p = 0.0003), while skid trails exhibited a weaker positive correlation (r = 0.43, p = 0.029). Log landings and secondary roads showed no significant correlation with logging volume (p = 0.052 and p = 0.10, respectively), while primary roads also showed no correlation (p = 0.59). A very strong positive correlation between total impact and tree gaps (r = 0.95, p < 0.0001) was found along with a strong positive correlation between impact and skid trails (r = 0.87, p < 0.0001) and the log landings (r = 0.76, p < 0.0001), a moderate correlation was found between the total impact and secondary roads (r = 0.55, p = 0.004), and a weak correlation was found with primary roads (r = 0.44, p = 0.02). As for the felled tree gaps, the correlation was strongly positive for trails (r = 0.69, p < 0.0001) and log landings (r = 0.69, p < 0.0001) and weak for secondary roads (r = 0.43, p = 0.031) and primary roads (r = 0.42, p = 0.034). The correlation between trails with log landings (r = 0.64, p = 0.0007) and secondary roads (r = 0.51, p = 0.008) was moderate, and the correlation with primary roads was not significant (p = 0.33).

4. Discussion

4.1. Infrastructure Mapping in the Understory

This study quantified and mapped logging infrastructure in the forest understory using LiDAR-derived RDMs, providing a comprehensive analysis of roads, skid trails, and log landings across selectively logged areas in the Amazon when compared to existing maps of infrastructure. RDM analysis highlighted the spatial distribution of all logging-related impacts on the understory, effectively capturing damage typically obscured by the forest canopy. The use of RDMs allowed for detailed mapping of understory disturbance, confirming LiDAR’s effectiveness in capturing the spatial distribution of selective logging impacts even in dense forest environments. These results align closely with previous research highlighting LiDAR’s potential as a robust monitoring tool, offering high-resolution data to support sustainable management practices by detailing fine-scale vegetation disturbances in tropical forests [56,62,86].
Elli et al. [87] reported slightly higher densities for skid trails (175 ± 32 m.ha−1) in areas where reduced-impact logging techniques were also applied; however, their study areas experienced a higher mean logging intensity (37 ± 9 m3.ha−1), more than double the intensity observed in this study (Table 2). The mean area occupied by logging infrastructure in the understory found here aligns closely with previous field-based surveys of tropical forests subjected to RIL practices [58,59,88,89,90].
The percentage of area occupied by forest exploitation was similar to other studies carried out in tropical regions where low-impact logging practices were also adopted, such as Arevalo et al. [88] and Pereira et al. [90]. Asner et al. [58] found a similar impacted area of 5.8% ± 1.9% in the eastern Amazon, while Pereira et al. [90] found values of 7.9% of the affected area in the same region. Additionally, Arevalo et al. [88] estimated that 7% of the directly affected area was impacted in Belize, Central America. In contrast, Jackson et al. [91] found that conventional logging affected approximately 25% of the understory in Bolivia. Although differing methodologies were used in these studies, they provide valuable estimates that reinforce the reliability of LiDAR-based RDM assessments for comparative understory impact evaluation.
In this study, skid trails occupy 3.9% (±0.27%) of the evaluated areas, representing approximately 67% of the total area impacted by logging infrastructure in the understory. This result aligns with recent studies where skid trails accounted for the largest portion of altered understory area due to logging infrastructure. Locks and Matricardi [86] reported that skid trails comprised around 4.8% (±1.2%) of the affected area, followed by secondary roads at 1.0% (±0.2%), primary roads at 0.6% (±0.5%), and log landings at 0.5% (±0.2%).
Similarly, Carvalho et al. [41] observed that skid trails covered 3.2% of the impacted area in a selectively logged forest, with a total of 5.1% of the understory affected one year after logging activities. Braz and d’Oliveira [92] further recommended that the maximum area occupied by secondary roads, storage landings, and skid trails should not exceed 1%, 0.75%, and 6% of the total managed area, respectively, to ensure sustainable logging practices. These findings suggest that skid trails contribute consistently to the majority of logging-related impacts on the understory, reflecting common patterns across tropical forest studies. The infrastructure density values observed in this study remain within these recommended limits and align with legal guidelines for sustainable forest management [14,93].
Additionally, the validation of RDM-based mapping showed that LiDAR data overestimated skid trail density compared to GNSS field measurements (Table 3), a result similar to that reported by Locks and Matricardi [86], who found a 9.4% overestimation using RDM mapping. Elli et al. [87] also observed a 14% higher value when mapping skid trails and tree-fall areas with LiDAR relative to field data obtained via GNSS. These results indicate that LiDAR-derived RDMs provide reliable yet slightly conservative estimates of skid trail impacts, making them a valuable tool for accurately assessing logging infrastructure within Amazonian forest understory environments.
Surveys of logging infrastructure mapped by the RDM model showed high positional agreement with GNSS-based field measurements, as shown in Table 4. However, vector features digitized from LiDAR data displayed some spatial displacement relative to the 5 and 10 m GNSS buffer zones. These results are consistent with previous studies. For example, Ellis et al. [87] found a 59% overlap of skid trails within a 10 m GNSS buffer in a study conducted in Indonesia, demonstrating similar accuracy levels in LiDAR mapping. Similarly, Locks and Matricardi [86] reported a 65% overlap of skid trails with a 10 m GNSS buffer in an Amazonian study, showing that LiDAR mapping tends to achieve high spatial agreement with GNSS data at broader buffer thresholds. Pinagé et al. [21] also noted that LiDAR mapping is highly accurate within these error margins, particularly in dense forest understory conditions where GNSS positional errors are common. These findings indicate that LiDAR mapping is generally reliable within the operational constraints of GNSS error margins, enhancing its utility in selective logging impact assessment.
However, when considering a positional error of up to 20 m (>70%) for navigation GNSS, LiDAR mapping showed closer alignment to field data, providing greater consistency between the two methods. This degree of positional error is typical in field surveys conducted under dense forest canopy, where signal interference from overhead vegetation frequently impacts GNSS accuracy [94,95]. GNSS navigation equipment in optimal conditions can achieve an accuracy of approximately 5 m; however, under a closed canopy, this accuracy is significantly reduced as the canopy obstructs satellite signal paths, increasing positional errors [96]. This interference is the primary factor contributing to observed discrepancies in positional accuracy between LiDAR and GNSS data within smaller buffer ranges of 5 and 10 m.
The positioning technique used by GNSS receivers in the navigation category provides an accuracy range of 10 to 30 m, as it operates with a single GNSS receiver. This method, which relies on instantaneous positioning, is limited in improving data accuracy even when stationary and collecting data over extended periods, due to inherent systematic errors [94]. Despite the signal degradation caused by dense forest canopy, this model remains effective for positioning and navigation within heavily vegetated areas [97]. Although not the most precise option, particularly for research purposes, the Garmin 76MAP CSX is widely utilized in fieldwork because of its reliability and ease of use in challenging environments [87].
The use of dual-frequency geodetic GNSS equipment as a field truth reference would enhance the accuracy of this research, as geodetic-grade GNSS methods allow for higher precision results compared to single-frequency receivers [94]. Skid trail density, as determined by the RDM within areas logged 0–24 months prior to LiDAR data collection, showed the highest trail densities. This finding aligns with the observed pattern of understory damage decreasing over time following logging, which reflects the progressive recovery of forest vegetation [21].

4.2. Total Understory Impact

In addition to the previously planned infrastructure, such as log landings and roads, logging operations caused collateral damage to their borders, which are more challenging to control, including the formation of drag trails and canopy gaps from felled trees. The extent of these impacts is directly related to the intensity of selective logging and the effectiveness of the logging techniques employed [98].
Beyond providing an estimate of infrastructure dimensions, LiDAR was able to capture the overall impact of logging on the forest understory (Table 5). The results obtained using RDM with LiDAR data were similar to those of Asner et al. [58], who conducted intensive field mapping of roads, trails, log landings, and felled tree gaps in an eastern Amazon logging area. The total understory impact observed in this study is consistent with the percentages of 15.4% and 23.7% reported by d’Oliveira et al. [15] and Andersen et al. [99], respectively, in studies using similar methods in RIL areas of the western Amazon.
Felled tree gaps accounted for 48.7% of the total impact on the understory in our study. This result aligns with Carvalho et al. [41], who conducted a LiDAR-based survey in an RIL forest management area within the Antimary State Forest in Acre, Brazil, and also found that felled tree gaps constituted the highest percentage of understory impact. Similar patterns have been observed in other tropical forest studies where RIL practices have been implemented; Arevalo et al. [88] and Pereira et al. [90] reported that felled tree gaps represented approximately 50% of the total affected area. Although these studies employed different methodologies, their results provide valuable, comparable estimates of understory impact.
The strong correlation observed between canopy gaps created from felled trees and understory impact suggests that as the density of gaps increases, there is a proportional increase in the extent of understory disturbance. In addition to the loss of biomass from the felled trees, the physical impact of tree falls causes significant damage to surrounding vegetation, which directly disrupts the ecological balance and natural regeneration processes of the forest. This disturbance is especially critical given that various forms of life within the forest ecosystem depend on light intensity, which is altered by the creation of canopy gaps [100].
LiDAR technology offers substantial potential for conducting detailed, accurate analyses of structural changes in forest ecosystems, allowing for a comprehensive assessment of the extent and ecological consequences of canopy gaps created by logging activities [57]. This capacity makes LiDAR a valuable tool for evaluating alterations in the forest canopy and vertical structure, assessing impacts on forest regeneration, and monitoring changes in biomass and carbon dynamics. Such capabilities complement the findings of this study and suggest LiDAR as a recommended approach for future research into these dynamics.
However, monitoring logging impacts over extended periods may be challenging, as evidence of selective logging impacts diminishes over time, leading to a partial loss of detectable impacts in subsequent LiDAR surveys [101]. This trend is further supported by the negative correlation observed between the time interval since logging and the total impact on the understory, indicating that LiDAR detects greater impacts when surveys are conducted closer to the logging date. This temporal effect explains the higher mean impact value found in areas surveyed within 12 months post-logging, compared to areas with longer intervals since logging activities, as shown in Table 6.
The results obtained in this study are consistent with those reported by Pinagé et al. [21], who examined logging impacts on a tropical forest in eastern Amazonia using LiDAR. Their study also observed that understory impact decreased as natural regeneration compensated for the damage over time. Degradation was most evident in recently logged areas, with the signs of disturbance diminishing in older logging sites. After six years, no significant difference was detected between understory samples from logged areas and intact forest areas (p > 0.01), indicating effective regeneration. Similarly, Schulze and Zweede [102] found that skid trails and gaps created by felled trees in the understory reached dense regeneration heights of 3 to 8 m within five years after logging, further supporting the notion of substantial regrowth in selectively logged forests.
Monitoring in logging areas is closely linked to forest regeneration processes, which vary depending on the type of disturbance [21]. In this study, it was observed that impacts caused by felled tree gaps and primary roads persisted over the evaluation period (Table 6). This trend is also reflected in the correlation analysis with time intervals, which showed that neither the damage from felled tree gaps nor from primary roads was significantly influenced by the time elapsed since logging. These findings suggest that LiDAR can effectively detect damage from these specific disturbances in the understory for up to three years post-logging, according to the evaluation period of this study.
These results can be explained by the fact that gaps created by tree falls require a longer time for regeneration [21], while primary roads remain heavily compacted due to the continuous passage of heavy trucks. This compaction hinders the natural regrowth of vegetation, and primary roads often continue to be used as access routes even after logging activities conclude, preventing the regeneration of these areas. Additionally, these roads frequently serve as entry points for illegal activities, such as deforestation, forest fires, and biological invasions [103].
In contrast, skid trails showed a decreasing impact over time, as these are temporary structures [30,104]. Due to the natural process of forest regeneration, skid trails become less visible in remote sensing data approximately two years after logging, even when using LiDAR data [41,87,105]. This pattern of diminishing visibility over time was also observed for log landings and secondary roads, confirming that the understory impact from these structures is most prominent in areas of recent selective logging and gradually fades as regeneration occurs. Consequently, for an accurate assessment of understory damage in managed forests, LiDAR data collection should ideally take place within two years of the initiation of logging activities.
The resilience of vegetation observed in skid trails indicates a high capacity for forest regeneration in areas disturbed by this type of infrastructure. This regenerative ability is characteristic of Dense Ombrophilous Forests with portions of Open Ombrophilous Forest, where recovery tends to be particularly rapid [36]. Such forests are often populated by spaced canopy trees and an understory rich in seedlings and young trees from species found in the upper canopy. This finding is consistent with previous studies in natural forests, which emphasize the importance of increased light availability in forest gaps, as it promotes the growth of fast-growing species that capitalize on the enhanced light conditions [40,41,42,43].
However, this can pose a risk, as invasive plants tend to be more efficient than native species at using available nutrients and light [39]. The evidence suggests that forest fragmentation and human activities have a detrimental impact on species diversity, functional diversity, and carbon stocks in forests [45]. Furthermore, invasive grasses modify the fire regime, thereby increasing the probability of fires and creating conditions conducive to their perpetuation within the ecosystem [39]. These facts are corroborated by the research of Barni et al. [46], which demonstrated that selective logging in southern Roraima significantly increased damage to forest biomass, carbon emissions, and the spread of intense fires.
In terms of impact across concession areas, FMU I exhibited the highest impact, followed by FMU III, with FMU II showing the lowest impact (Table 6). These results demonstrate a strong positive correlation between logged volume and total understory impact. This relationship is likely due to the logging intensity implemented by each concessionaire. FMU I recorded the highest logging intensity at 16.04 m3.ha−1, followed by FMU III at 11.89 m3.ha−1 and FMU II at 9.88 m3.ha−1. These findings suggest that higher logging intensity results in greater understory impact, even when RIL practices are applied.
The SFB has established a maximum allowable impact of 8% for damage resulting from logging infrastructure (excluding primary roads) and 10% for damage from tree felling within each APU [23]. Although the mean impact observed in this study fell within these limits, 12 of the 25 areas assessed exceeded the SFB’s defined thresholds (Table 5). These areas had a mean logging intensity lower than the maximum allowable volume of 25.8 m3.ha−1 for a 30-year harvest cycle [86]. This suggests that, even under regulated conditions, the impacts may exceed the established limits unless there is a review and improvement in RIL techniques.
The impact was primarily attributable to skid trails, which could only be accurately identified through LiDAR mapping; these trails would have been challenging to detect with other methods, even using high-resolution imagery. Skid trails are generally undetectable with passive remote sensing techniques [106,107], as these methods have been used in prior studies [64,77,108] but lack the capacity to reveal changes in understory vegetation. As a result, passive methods tend to underestimate the impact, as the upper canopy obscures the true extent of disturbance corridors [87].
The apparent lack of attention from the logging concessionaire to the management plan may have influenced the number of skid trails created, leading to increased damage in these areas. It is likely that environmental impacts could have been minimized if skidder routes had been optimized and operators had exercised appropriate caution during operations [25]. Reducing the disturbances and impacts associated with skid trails is a critical consideration, as high impact rates on these trails often reflect poor management practices. This variable is directly tied to effective strategic planning and the implementation of management practices in the field [109]. The integration of geospatial planning models with technical data and computational techniques, as proposed by Morais et al. [110], can support the optimization of logging infrastructure placement, ultimately reducing understory disturbances.

4.3. Perspectives of Airborne LiDAR Monitoring in RIL Forests

The Brazilian legal framework establishes that monitoring forest dynamics and regeneration as well as assessing impacts on forest cover are essential indicators for ensuring sustainable forest management [10,13,14]. These indicators should be measured reliably by the SFB [26]. In light of the challenges associated with field monitoring [57,58,60,111,112] and the limitations of aerial and satellite imagery [64,77,106,107,108], the findings of this study suggest that LiDAR data represents an effective alternative for monitoring the forest understory. LiDAR data provided reliable estimates, with logging infrastructure and canopy gaps created by tree felling clearly visible in the RDM analysis and aligned with GNSS field measurements. This approach not only allows for the accurate quantification of the impacts of RIL but also facilitates the straightforward detection of conventional logging, where impacts tend to be more pronounced [21]. Despite the relatively high initial cost and logistical challenges of ALSs [113], this technology has become affordable (Table 3). In 2024, the SFB contracted airborne LiDAR surveys at a cost of USD 2.75 per hectare [114], significantly lower than the cost of traditional field inventory methods, which can reach up to USD 181.60 per hectare [115], adjusted to current values [116,117]. Given that concession contracts mandate concessionaires to finance continuous inventories within permanent plots [23], extending this funding to include LiDAR surveys is a viable and logical approach. This method could improve efficiency, as ALSs are not confined to small plots and allow monitoring across entire APUs, representing a significative impact for large-scale monitoring programs.
However, LiDAR data processing and analysis require specialized software and technical expertise, which may not be readily accessible in all regions. Another limitation is the interval between logging activities and the aerial survey period. This is due to the fact that it is not appropriate to carry out monitoring two years after an impact has occurred. Additionally, there could be a need to focus on other cost-effective solutions. This includes the use of SAR data and artificial intelligence techniques for the automatic identification of impact zones, which could also streamline the analysis workflow and reduce the reliance on manual interpretation. This approach would be beneficial as this type of data is free of charge for the users and has the necessary temporal resolution to carry out this task.

5. Conclusions

This study demonstrates that LiDAR-derived RDMs offer a reliable tool for monitoring selective logging impacts in Amazonian forest understories. The RDM approach provided spatially accurate measurements of logging infrastructure—including primary and secondary roads, skid trails, and log landings—aligning with GNSS-based field data. Unlike conventional remote sensing methods, which are limited in detecting changes beneath dense canopies, LiDAR data enabled the mapping of understory disturbances, making it particularly suitable for tracking RIL compliance and detecting areas where conventional logging practices may have occurred. The analysis demonstrated that skid trails had the most significant impact among logging infrastructure, highlighting the need for effective management practices and optimized planning to minimize disturbances. Furthermore, the strong correlation between the time since logging and observed understory impact highlights the value of conducting LiDAR surveys within two years post-logging. This timeframe ensures the most accurate representation of disturbances before natural regeneration processes obscure them.
This methodology has been demonstrated to improve regulatory compliance and forest preservation by reducing the reliance on field measurements or ineffective methods. The employment of LiDAR technology ensures better monitoring precision, covering entire areas rather than being restricted to sampling regions. Additionally, it can lower current costs for concessionaires. To establish this approach as an effective monitoring tool, future research should use LiDAR data to analyze forest canopy changes, vertical profiles, and impacts on regeneration, biomass, and carbon dynamics. Further studies are recommended to validate the results across a range of ecological and management conditions, including regional climate, forest composition, and logging practices. It is also recommended that regular surveys should be conducted on the same areas to enable time series monitoring. Long-term studies are crucial for understanding understory regeneration dynamics and the effects of different logging intensities, ultimately aiding the refinement of sustainable forest management practices.

Author Contributions

Conceptualization, L.F. and E.d.S.B.; methodology, L.F. and Q.S.B.; software, L.F. and Q.S.B.; validation, L.F.; formal analysis, L.F. and Q.S.B.; investigation, L.F.; resources, E.d.S.B. and E.A.T.M.; data curation, L.F.; writing—original draft preparation, L.F. and E.d.S.B.; writing—review and editing, Q.S.B., E.A.T.M., L.P. and J.J.S.; visualization, E.d.S.B., L.P. and J.J.S.; supervision, E.d.S.B., L.P. and J.J.S.; project administration, L.F. and E.d.S.B.; funding acquisition, E.d.S.B., E.A.T.M., L.P. and J.J.S. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgments

The authors would like to thank the Coordination for the Improvement of Higher Education Personnel—Brazil (CAPES)—Financing Code 001; the Brazilian Forest Service (SFB), particularly José Humberto Chaves, for providing the field and LiDAR point cloud data; and Ricardo Seixas Brites, in memoriam. The authors would also like to acknowledge the Portuguese Foundation for Science and Technology (FCT) for financial support through national funds to projects UIDB/04033/2020 (https://doi.org/10.54499/UIDB/04033/2020) and LA/P/0126/2020 (https://doi.org/10.54499/LA/P/0126/2020).

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Abbreviations

The following abbreviations are used in this manuscript:
RILReduced-impact logging
SFMSustainable forest management
GNSSGlobal navigation satellite system
LiDARLight detection and ranging
RDMRelative density model
REDDReducing emissions from deforestation and forest degradation
SFBBrazilian Forest Service
MMAMinistry of the Environment
SFMPSustainable forest management plan
UAVUnmanned aerial vehicle
ALSAirborne laser scanner
FMUForest management unit
APUAnnual production unit
UTMUniversal Transverse Mercator
EPSGEuropean Petroleum Survey Group
DBHDiameter at breast height
AOPAnnual operational plan
IBAMABrazilian Institute of Environment and Renewable Natural Resources
DTMDigital terrain model
SEStandard error
RMSERoot-mean-square error
R2Coefficient of determination

References

  1. Assis, T.O.; Aguiar, A.P.D.; von Randow, C.; Nobre, C.A. Projections of Future Forest Degradation and CO2 Emissions for the Brazilian Amazon. Sci. Adv. 2022, 8, eabj3309. [Google Scholar] [CrossRef] [PubMed]
  2. Erb, K.-H.; Kastner, T.; Plutzar, C.; Bais, A.L.S.; Carvalhais, N.; Fetzel, T.; Gingrich, S.; Haberl, H.; Lauk, C.; Niedertscheider, M.; et al. Unexpectedly Large Impact of Forest Management and Grazing on Global Vegetation Biomass. Nature 2018, 553, 73–76. [Google Scholar] [CrossRef] [PubMed]
  3. Lapola, D.M.; Pinho, P.; Barlow, J.; Aragão, L.E.O.C.; Berenguer, E.; Carmenta, R.; Liddy, H.M.; Seixas, H.; Silva, C.V.J.; Silva-Junior, C.H.L.; et al. The Drivers and Impacts of Amazon Forest Degradation. Science 2023, 379, eabp8622. [Google Scholar] [CrossRef] [PubMed]
  4. Atmadja, S.S.; Duchelle, A.E.; Sy, V.D.; Selviana, V.; Komalasari, M.; Sills, E.O.; Angelsen, A. How Do REDD+ Projects Contribute to the Goals of the Paris Agreement? Environ. Res. Lett. 2022, 17, 044038. [Google Scholar] [CrossRef]
  5. Guerra, R.; Moutinho, P. Challenges of Sharing REDD+ Benefits in the Amazon Region. Forests 2020, 11, 1012. [Google Scholar] [CrossRef]
  6. Jucker, T.; Asner, G.; Dalponte, M.; Bodrick, P.; Philipson, C.; Vaughn, N.; Brelsford, C.; Burslem, D.; Deere, N.; Ewers, R.; et al. A Regional Model for Estimating the Aboveground Carbon Density of Borneo’s Tropical Forests from Airborne Laser Scanning. arXiv 2017, arXiv:1705.09242. [Google Scholar] [CrossRef]
  7. Sist, P.; Piponiot, C.; Kanashiro, M.; Pena-Claros, M.; Putz, F.E.; Schulze, M.; Verissimo, A.; Vidal, E. Sustainability of Brazilian Forest Concessions. For. Ecol. Manag. 2021, 496, 119440. [Google Scholar] [CrossRef]
  8. Brancalion, P.H.S.; de Almeida, D.R.A.; Vidal, E.; Molin, P.G.; Sontag, V.E.; Souza, S.E.X.F.; Schulze, M.D. Fake Legal Logging in the Brazilian Amazon. Sci. Adv. 2018, 4, eaat1192. [Google Scholar] [CrossRef]
  9. Matricardi, E.A.T.; Skole, D.L.; Costa, O.B.; Pedlowski, M.A.; Samek, J.H.; Miguel, E.P. Long-Term Forest Degradation Surpasses Deforestation in the Brazilian Amazon. Science 2020, 369, 1378–1382. [Google Scholar] [CrossRef]
  10. Brazil Lei No 11.284. Available online: https://www.planalto.gov.br/ccivil_03/_Ato2004-2006/2006/Lei/L11284.htm (accessed on 27 October 2024).
  11. Serviço Florestal Brasileiro Participação Social é Retomada na Gestão das Florestas Públicas. Available online: https://www.gov.br/florestal/pt-br/assuntos/noticias/participacao-social-e-retomada-na-gestao-das-florestas-publicas (accessed on 26 October 2024).
  12. Vidal, E.; West, T.; Lentini, M.; Souza, S.; Klauberg, C.; Waldhoff, P. Sustainable Forest Management (SFM) of Tropical Moist Forests: The Case of the Brazilian Amazon; Burleigh Dodds Science Publishing Limited: Cambridge, UK, 2021; pp. 619–650. ISBN 978-1-78676-248-1. [Google Scholar]
  13. Brazil L12651. Available online: https://www.planalto.gov.br/ccivil_03/_ato2011-2014/2012/lei/l12651.htm (accessed on 29 October 2024).
  14. MMA Instrução Normativa MMA No 5 de 11/12/2006—Federal—LegisWeb. Available online: https://www.legisweb.com.br/legislacao/?id=76720 (accessed on 27 October 2024).
  15. d’Oliveira, M.V.N.; Reutebuch, S.E.; McGaughey, R.J.; Andersen, H.-E. Estimating Forest Biomass and Identifying Low-Intensity Logging Areas Using Airborne Scanning Lidar in Antimary State Forest, Acre State, Western Brazilian Amazon. Remote Sens. Environ. 2012, 124, 479–491. [Google Scholar] [CrossRef]
  16. SFB—Serviço Florestal Brasileiro Concessões e Monitoramento. Available online: https://www.gov.br/florestal/pt-br/assuntos/concessoes-e-monitoramento (accessed on 27 October 2024).
  17. Bicknell, J.E.; Struebig, M.J.; Edwards, D.P.; Davies, Z.G. Improved Timber Harvest Techniques Maintain Biodiversity in Tropical Forests. Curr. Biol. 2014, 24, R1119–R1120. [Google Scholar] [CrossRef] [PubMed]
  18. Neves d’Oliveira, M.V.; Miller, R.P.; Oliveira, L.C.; Braz, E.M.; Thaines, F.; Januário, J.L.; Acuña, M.H.A. Growth Dynamics of an Amazonian Forest: Effects of Reduced Impact Logging and Recurring Atypical Climate Events during a 20-Year Study. For. Ecol. Manag. 2024, 562, 121937. [Google Scholar] [CrossRef]
  19. Okuda, T.; Yamada, T.; Hosaka, T.; Miyasaku, N.; Hashim, M.; Lau, A.M.S.; Saw, L.G. Canopy Height Recovery after Selective Logging in a Lowland Tropical Rain Forest. For. Ecol. Manag. 2019, 442, 117–123. [Google Scholar] [CrossRef]
  20. Putz, F.E.; Sist, P.; Fredericksen, T.; Dykstra, D. Reduced-Impact Logging: Challenges and Opportunities. For. Ecol. Manag. 2008, 256, 1427–1433. [Google Scholar] [CrossRef]
  21. Rangel Pinagé, E.; Keller, M.; Duffy, P.; Longo, M.; dos-Santos, M.N.; Morton, D.C. Long-Term Impacts of Selective Logging on Amazon Forest Dynamics from Multi-Temporal Airborne LiDAR. Remote Sens. 2019, 11, 709. [Google Scholar] [CrossRef]
  22. West, T.A.P.; Vidal, E.; Putz, F.E. Forest Ecology and Management Forest Biomass Recovery after Conventional and Reduced-Impact Logging in Amazonian Brazil. For. Ecol. Manag. 2014, 314, 59–63. [Google Scholar] [CrossRef]
  23. SFB—Serviço Florestal Brasileiro Fichas de Parametrização de Indicadores para Fins de Classificação e Bonificação no Lote de Concessão Florestal. EDITAL No 01/2007—Anexo 12, 26p. Available online: https://www.gov.br/florestal/pt-br/assuntos/concessoes-e-monitoramento/editais-em-licitacao (accessed on 27 October 2024).
  24. Venanzi, R.; Latterini, F.; Civitarese, V.; Picchio, R. Recent Applications of Smart Technologies for Monitoring the Sustainability of Forest Operations. Forests 2023, 14, 1503. [Google Scholar] [CrossRef]
  25. Otone Aguiar, M.; Fernandes da Silva, G.; Regis Mauri, G.; Ribeiro de Mendonça, A.; Ferreira da Silva, E.; Orfano Figueiredo, E.; Pereira Martins Silva, J.; Alves da Silva, V.; Freitas Silva, R.; Lessa Lavagnoli, G. Integrated Planning of Forest Exploration Infrastructures in an Amazonian Sustainable Forest Management Area. For. Ecol. Manag. 2023, 549, 121265. [Google Scholar] [CrossRef]
  26. Brazil D12046—Regulamenta, Em Âmbito Federal, a Lei No 11.284, de 2 de Março de 2006, Que Dispõe Sobre a Gestão de Florestas Públicas Para a Produção Sustentável, e Dá Outras Providências. Available online: https://www.planalto.gov.br/ccivil_03/_Ato2023-2026/2024/Decreto/D12046.htm#art58 (accessed on 2 November 2024).
  27. Cazzolla Gatti, R.; Castaldi, S.; Lindsell, J.A.; Coomes, D.A.; Marchetti, M.; Maesano, M.; Di Paola, A.; Paparella, F.; Valentini, R. The Impact of Selective Logging and Clearcutting on Forest Structure, Tree Diversity and above-Ground Biomass of African Tropical Forests. Ecol. Res. 2015, 30, 119–132. [Google Scholar] [CrossRef]
  28. Tallei, E.; Rivera, L.; Schaaf, A.; Vivanco, C.; Politi, N. Post-Logging Changes in a Neotropical Dry Forest Composition and Structure Modify the Ecosystem Functioning. For. Ecol. Manag. 2023, 537, 120944. [Google Scholar] [CrossRef]
  29. Bartels, S.F.; Macdonald, S.E. Dynamics and Recovery of Forest Understory Biodiversity over 17 Years Following Varying Levels of Retention Harvesting. J. Appl. Ecol. 2023, 60, 725–736. [Google Scholar] [CrossRef]
  30. Pinagé, E.R.; Matricardi, E.A.T. Detecção da Infraestrutura para Exploração Florestal em Rondônia Utilizando Dados de Sensoriamento Remoto. Floresta Ambient. 2015, 22, 377–390. [Google Scholar] [CrossRef]
  31. Norghauer, J.M.; Newbery, D.M. Herbivores Differentially Limit the Seedling Growth and Sapling Recruitment of Two Dominant Rain Forest Trees. Oecologia 2014, 174, 459–469. [Google Scholar] [CrossRef]
  32. Kapos, V.; Mulkey, S.S.; Chazdon, R.L.; Smith, A.P. Tropical Forest Plant Ecophysiology. J. Appl. Ecol. 1997, 34, 831. [Google Scholar] [CrossRef]
  33. Costa, F.R.C.; Senna, C.; Nakkazono, E.M. Effects of Selective Logging on Populations of Two Tropical Understory Herbs in an Amazonian Forest1. Biotropica 2002, 34, 289–296. [Google Scholar] [CrossRef]
  34. Kozera, C.; Rodrigues, R.R.; Dittrich, V.A.D.O. Composição florística do sub-bosque de uma floresta ombrófila densa montana, morretes, PR, Brasil. FLORESTA 2009, 39, 323–334. [Google Scholar] [CrossRef]
  35. Darrigo, M.R.; Dos Santos, F.A.M.; Venticinque, E.M. The Confounding Effects of Logging on Tree Seedling Growth and Herbivory in Central Amazon. Biotropica 2018, 50, 60–68. [Google Scholar] [CrossRef]
  36. IBGE Manual Técnico da Vegetação Brasileira. Available online: https://loja.ibge.gov.br/manual-tecnico-da-vegetac-o-brasileira.html (accessed on 27 October 2024).
  37. Zhao, K.; He, F. Estimating Light Environment in Forests with a New Thresholding Method for Hemispherical Photography. Can. J. For. Res. 2016, 46, 1103–1110. [Google Scholar] [CrossRef]
  38. Keeling, H.C.; Phillips, O.L. A Calibration Method for the Crown Illumination Index for Assessing Forest Light Environments. For. Ecol. Manag. 2007, 242, 431–437. [Google Scholar] [CrossRef]
  39. Sampaio, A.B.; Schmidt, I.B. Espécies Exóticas Invasoras em Unidades de Conservação Federais do Brasil. Biodivers. Bras. 2013, 3, 32–49. [Google Scholar]
  40. Bauer, L.; Knapp, N.; Fischer, R. Mapping Amazon Forest Productivity by Fusing GEDI Lidar Waveforms with an Individual-Based Forest Model. Remote Sens. 2021, 13, 4540. [Google Scholar] [CrossRef]
  41. de Carvalho, A.L.; d’Oliveira, M.V.N.; Putz, F.E.; de Oliveira, L.C. Natural Regeneration of Trees in Selectively Logged Forest in Western Amazonia. For. Ecol. Manag. 2017, 392, 36–44. [Google Scholar] [CrossRef]
  42. Miller, S.D.; Goulden, M.L.; Hutyra, L.R.; Keller, M.; Saleska, S.R.; Wofsy, S.C.; Figueira, A.M.S.; da Rocha, H.R.; de Camargo, P.B. Reduced Impact Logging Minimally Alters Tropical Rainforest Carbon and Energy Exchange. Proc. Natl. Acad. Sci. USA 2011, 108, 19431–19435. [Google Scholar] [CrossRef] [PubMed]
  43. Winstanley, P.; Dalagnol, R.; Mendiratta, S.; Braga, D.; Galvão, L.S.; Bispo, P.d.C. Post-Logging Canopy Gap Dynamics and Forest Regeneration Assessed Using Airborne LiDAR Time Series in the Brazilian Amazon with Attribution to Gap Types and Origins. Remote Sens. 2024, 16, 2319. [Google Scholar] [CrossRef]
  44. Richards, L.A.; Coley, P.D. Seasonal and Habitat Differences Affect the Impact of Food and Predation on Herbivores: A Comparison between Gaps and Understory of a Tropical Forest. Oikos 2007, 116, 31–40. [Google Scholar] [CrossRef]
  45. Bastos, J.R.; Capellesso, E.S.; Vibrans, A.C.; Marques, M.C.M. Human Impacts, Habitat Quantity and Quality Affect the Dimensions of Diversity and Carbon Stocks in Subtropical Forests: A Landscape-Based Approach. J. Nat. Conserv. 2023, 73, 126383. [Google Scholar] [CrossRef]
  46. Barni, P.E.; Rego, A.C.M.; Silva, F.d.C.F.; Lopes, R.A.S.; Xaud, H.A.M.; Xaud, M.R.; Barbosa, R.I.; Fearnside, P.M. Logging Amazon Forest Increased the Severity and Spread of Fires during the 2015–2016 El Niño. For. Ecol. Manag. 2021, 500, 119652. [Google Scholar] [CrossRef]
  47. Santos, E.G.; Svátek, M.; Nunes, M.H.; Aalto, J.; Senior, R.A.; Matula, R.; Plichta, R.; Maeda, E.E. Structural Changes Caused by Selective Logging Undermine the Thermal Buffering Capacity of Tropical Forests. Agric. For. Meteorol. 2024, 348, 109912. [Google Scholar] [CrossRef]
  48. Verhegghen, A.; Eva, H.; Achard, F. Assessing Forest Degradation from Selective Logging Using Time Series of Fine Spatial Resolution Imagery in Republic of Congo. In Proceedings of the 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Milan, Italy, 26–31 July 2015; pp. 2044–2047. [Google Scholar]
  49. Souza, J.; Siqueira, J.V.; Sales, M.H.; Fonseca, A.V.; Ribeiro, J.G.; Numata, I.; Cochrane, M.A.; Barber, C.P.; Roberts, D.A.; Barlow, J. Ten-Year Landsat Classification of Deforestation and Forest Degradation in the Brazilian Amazon. Remote Sens. 2013, 5, 5493–5513. [Google Scholar] [CrossRef]
  50. Welsink, A.-J.; Reiche, J.; de Sy, V.; Carter, S.; Slagter, B.; Suarez, D.R.; Batros, B.; Peña-Claros, M.; Herold, M. Towards the Use of Satellite-Based Tropical Forest Disturbance Alerts to Assess Selective Logging Intensities. Environ. Res. Lett. 2023, 18, 054023. [Google Scholar] [CrossRef]
  51. Reiche, J.; Mullissa, A.; Slagter, B.; Gou, Y.; Tsendbazar, N.-E.; Odongo-Braun, C.; Vollrath, A.; Weisse, M.J.; Stolle, F.; Pickens, A.; et al. Forest Disturbance Alerts for the Congo Basin Using Sentinel-1. Environ. Res. Lett. 2021, 16, 024005. [Google Scholar] [CrossRef]
  52. Castillo, G.V.B.; de Freitas, L.J.M.; Cordeiro, V.A.; Orellana, J.B.P.; Reategui-Betancourt, J.L.; Nagy, L.; Matricardi, E.A.T. Assessment of Selective Logging Impacts Using UAV, Landsat, and Sentinel Data in the Brazilian Amazon Rainforest. JARS 2022, 16, 014526. [Google Scholar] [CrossRef]
  53. Jackson, C.M.; Adam, E. Remote Sensing of Selective Logging in Tropical Forests: Current State and Future Directions. iForest Biogeosci. For. 2020, 13, 286. [Google Scholar] [CrossRef]
  54. Zhang, H.; Bauters, M.; Boeckx, P.; Van Oost, K. Mapping Canopy Heights in Dense Tropical Forests Using Low-Cost UAV-Derived Photogrammetric Point Clouds and Machine Learning Approaches. Remote Sens. 2021, 13, 3777. [Google Scholar] [CrossRef]
  55. Pádua, L.; Vanko, J.; Hruška, J.; Adão, T.; Sousa, J.J.; Peres, E.; Morais, R. UAS, Sensors, and Data Processing in Agroforestry: A Review towards Practical Applications. Int. J. Remote Sens. 2017, 38, 2349–2391. [Google Scholar] [CrossRef]
  56. d’Oliveira, M.V.N.; Figueiredo, E.O.; de Almeida, D.R.A.; Oliveira, L.C.; Silva, C.A.; Nelson, B.W.; da Cunha, R.M.; de Almeida Papa, D.; Stark, S.C.; Valbuena, R. Impacts of Selective Logging on Amazon Forest Canopy Structure and Biomass with a LiDAR and Photogrammetric Survey Sequence. For. Ecol. Manag. 2021, 500, 119648. [Google Scholar] [CrossRef]
  57. Gomes, L.F.; Brites, R.S.; Locks, C.J.; Anjos, R.R. dos Estimativas das Alterações na Biomassa Florestal Utilizando LiDAR em Área de Manejo Florestal Sustentável na Amazônia Sul-Ocidental. Anuário Inst. Geociências 2020, 43. [Google Scholar] [CrossRef]
  58. Asner, G.P.; Keller, M.; Silva, J.N.M. Spatial and Temporal Dynamics of Forest Canopy Gaps Following Selective Logging in the Eastern Amazon. Glob. Change Biol. 2004, 10, 765–783. [Google Scholar] [CrossRef]
  59. Lentini, M.W.; Zweede, J.C.; Holmes, T.P. Measuring Ecological Impacts from Logging in Natural Forests of the Eastern Amazonia as a Tool to Assess Forest Degradation; Forest Resources Assessment Working Paper 165; 2010; pp. 1–9. Available online: https://research.fs.usda.gov/treesearch/36443 (accessed on 27 October 2024).
  60. Ørka, H.O.; Hansen, E.H.; Dalponte, M.; Gobakken, T.; Næsset, E. Large-Area Inventory of Species Composition Using Airborne Laser Scanning and Hyperspectral Data. Silva Fenn. 2021, 55, 10244. [Google Scholar] [CrossRef]
  61. Gomes, L.F. Alterações Estruturais Ocorridas em Área de Concessão na Floresta Estadual do Antimary—Acre. Master’s Thesis, (Mestrado em Ciências Florestais), Universidade de Brasília, Brasília, Brazil, 2016. [Google Scholar]
  62. Barros, Q.S.; d’ Oliveira, M.V.N.; da Silva, E.F.; Görgens, E.B.; de Mendonça, A.R.; da Silva, G.F.; Reis, C.R.; Gomes, L.F.; de Carvalho, A.L.; de Oliveira, E.K.B.; et al. Indicators for Monitoring Reduced Impact Logging in the Brazilian Amazon Derived from Airborne Laser Scanning Technology. Ecol. Inform. 2024, 82, 102654. [Google Scholar] [CrossRef]
  63. Lima, T.A.; Beuchle, R.; Griess, V.C.; Verhegghen, A.; Vogt, P. Spatial Patterns of Logging-Related Disturbance Events: A Multi-Scale Analysis on Forest Management Units Located in the Brazilian Amazon. Landsc. Ecol. 2020, 35, 2083–2100. [Google Scholar] [CrossRef]
  64. Pearson, T.R.H.; Brown, S.; Casarim, F.M. Carbon Emissions from Tropical Forest Degradation Caused by Logging. Environ. Res. Lett. 2014, 9, 034017. [Google Scholar] [CrossRef]
  65. d’Oliveira, M.V.N.; Broadbent, E.N.; Oliveira, L.C.; Almeida, D.R.A.; Papa, D.A.; Ferreira, M.E.; Zambrano, A.M.A.; Silva, C.A.; Avino, F.S.; Prata, G.A.; et al. Aboveground Biomass Estimation in Amazonian Tropical Forests: A Comparison of Aircraft- and GatorEye UAV-Borne LiDAR Data in the Chico Mendes Extractive Reserve in Acre, Brazil. Remote Sens. 2020, 12, 1754. [Google Scholar] [CrossRef]
  66. Dalagnol, R.; Phillips, O.L.; Gloor, E.; Galvão, L.S.; Wagner, F.H.; Locks, C.J.; Aragão, L.E.O.C. Quantifying Canopy Tree Loss and Gap Recovery in Tropical Forests under Low-Intensity Logging Using VHR Satellite Imagery and Airborne LiDAR. Remote Sens. 2019, 11, 817. [Google Scholar] [CrossRef]
  67. Fisher, A.; Armston, J.; Goodwin, N.; Scarth, P. Modelling Canopy Gap Probability, Foliage Projective Cover and Crown Projective Cover from Airborne Lidar Metrics in Australian Forests and Woodlands. Remote Sens. Environ. 2020, 237, 111520. [Google Scholar] [CrossRef]
  68. Papa, D.d.A.; de Almeida, D.R.A.; Silva, C.A.; Figueiredo, E.O.; Stark, S.C.; Valbuena, R.; Rodriguez, L.C.E.; d’ Oliveira, M.V.N. Evaluating Tropical Forest Classification and Field Sampling Stratification from Lidar to Reduce Effort and Enable Landscape Monitoring. For. Ecol. Manag. 2020, 457, 117634. [Google Scholar] [CrossRef]
  69. Rex, F.E.; Corte, A.P.D.; Silva, C.A.; Machado, S.d.A.; Sanquetta, C.R. Dynamics of Above-Ground Biomass in the Brazilian Amazon Using LiDAR Data. Anuário Inst. Geociências 2020, 43, 228–238. [Google Scholar] [CrossRef]
  70. SFB—Serviço Florestal Brasileiro Floresta Nacional do Jamari (RO). Available online: https://www.gov.br/florestal/pt-br/assuntos/concessoes-e-monitoramento/concessoes-florestais-em-andamento/floresta-nacional-do-jamari-ro-2/floresta-nacional-do-jamari-ro-1 (accessed on 27 October 2024).
  71. McGaughey, B. FUSION/LDV: Software for LIDAR Data Analysis and Visualization; Forest Service: Washington, DC, USA, 2024; p. 226.
  72. Andersen, H.-E.; Clarkin, T.; Winterberger, K.; Strunk, J. An Accuracy Assessment of Positions Obtained Using Survey- and Recreational-Grade Global Positioning System Receivers across a Range of Forest Conditions within the Tanana Valley of Interior Alaska. West. J. Appl. For. 2009, 24, 128–136. [Google Scholar] [CrossRef]
  73. IBAMA Norma de Execução No 1, de 24 de Abril de 2007. Diretrizes Técnicas Para Elaboração Dos Planos de Manejo Florestal Sustentável—PMFS. In Diário Oficial Da União: Seção 1, Brasília, DF; 30 April 2007; p. 405. Available online: https://www.ibama.gov.br/component/legislacao/?view=legislacao&legislacao=113233 (accessed on 27 October 2024).
  74. ESRI ArcMap Software, ArcGIS Release 10.8.2; ESRI: Redlands, CA, USA, 2023.
  75. Han, X.-F.; Jin, J.S.; Wang, M.-J.; Jiang, W.; Gao, L.; Xiao, L. A Review of Algorithms for Filtering the 3D Point Cloud. Signal Process. Image Commun. 2017, 57, 103–112. [Google Scholar] [CrossRef]
  76. Kraus, K.; Pfeifer, N. Advanced DTM Generation From Lidar Data. Int. Arch. Photogramm. Remote Sens. 2001, 3-W4, 23–30. [Google Scholar]
  77. Azizi, Z.; Najafi, A.; Sadeghian, S. Forest Road Detection Using LiDAR Data. J. For. Res. 2014, 25, 975–980. [Google Scholar] [CrossRef]
  78. Clode, S.; Rottensteiner, F.; Kootsookos, P.; Zelniker, E. Detection and Vectorization of Roads from Lidar Data. Photogramm. Eng. Remote Sens. 2007, 73, 517–535. [Google Scholar] [CrossRef]
  79. White, R.A.; Dietterick, B.C.; Mastin, T.; Strohman, R. Forest Roads Mapped Using LiDAR in Steep Forested Terrain. Remote Sens. 2010, 2, 1120–1141. [Google Scholar] [CrossRef]
  80. Quackenbush, L.J. A Review of Techniques for Extracting Linear Features from Imagery. Photogramm. Eng. Remote Sens. 2004, 70, 1383–1392. [Google Scholar] [CrossRef]
  81. Goodchild, M.F.; Hunter, G.J. A Simple Positional Accuracy Measure for Linear Features. Int. J. Geogr. Inf. Sci. 1997, 11, 299–306. [Google Scholar] [CrossRef]
  82. d’Oliveira, M.V.N.; Figueiredo, E.O.; Papa, D.A. Uso Do LiDAR Como Ferramenta Para o Manejo de Precisão Em FLorestas Tropicais; Embrapa: Brasília, DF, Brazil; 130p, Available online: https://www.embrapa.br/busca-de-publicacoes/-/publicacao/1029435/uso-do-lidar-como-ferramenta-para-o-manejo-de-precisao-em-florestas-tropicais (accessed on 27 October 2024).
  83. R Core Team. R: The R Project for Statistical Computing. Available online: https://www.r-project.org/ (accessed on 27 October 2024).
  84. Box, G.E.; Hunter, J.S.; Hunter, W.G. Statistics for Experimenters: Design, Innovation, and Discovery, 2nd ed.; Wiley-Interscience: Hoboken, NJ, USA, 2005; p. 672. ISBN 978-0-471-71813-0. [Google Scholar]
  85. Mukaka, M. Statistics Corner: A Guide to Appropriate Use of Correlation Coefficient in Medical Research. Malawi Med. J. 2012, 24, 69–71. [Google Scholar]
  86. Locks, C.J.; Matricardi, E.A.T. Estimativa de Impactos Da Extração Seletiva de Madeiras Na Amazônia Utilizando Dados LIDAR. Ciência Florest. 2019, 29, 481. [Google Scholar] [CrossRef]
  87. Ellis, P.; Griscom, B.; Walker, W.; Gonçalves, F.; Cormier, T. Mapping Selective Logging Impacts in Borneo with GPS and Airborne Lidar. For. Ecol. Manag. 2016, 365, 184–196. [Google Scholar] [CrossRef]
  88. Arevalo, B.; Valladarez, J.; Muschamp, S.; Kay, E.; Finkral, A.; Roopsind, A.; Putz, F.E. Effects of Reduced-Impact Selective Logging on Palm Regeneration in Belize. For. Ecol. Manag. 2016, 369, 155–160. [Google Scholar] [CrossRef]
  89. Johns, J.S.; Barreto, P.; Uhl, C. Logging Damage during Planned and Unplanned Logging Operations in the Eastern Amazon. For. Ecol. Manag. 1996, 89, 59–77. [Google Scholar] [CrossRef]
  90. Pereira, R.; Zweede, J.; Asner, G.P.; Keller, M. Forest Canopy Damage and Recovery in Reduced-Impact and Conventional Selective Logging in Eastern Para, Brazil. For. Ecol. Manag. 2002, 168, 77–89. [Google Scholar] [CrossRef]
  91. Jackson, S.M.; Fredericksen, T.S.; Malcolm, J.R. Area Disturbed and Residual Stand Damage Following Logging in a Bolivian Tropical Forest. For. Ecol. Manag. 2002, 166, 271–283. [Google Scholar] [CrossRef]
  92. Braz, E.M.; Oliveira, M.V.N. D’ Planejamento da Extração Madeireira Dentro de Critérios Econômicos e Ambientais; Embrapa Acre: Rio Branco, AC, Brazil, 2001; p. 17. ISSN 0100-9915. [Google Scholar]
  93. CONAMA National Environmental Council 2009. Resolução no 406/2009. Available online: https://www.gov.br/mma/pt-br/pagina-inicial (accessed on 27 October 2024).
  94. Garrastazu, M.C.; Rosot, M.A.D. Manual de Orientação e uso do GPS de Navegação (Garmin 76MAP CSX); Embrapa Florestas: Colombo, PR, Brazil, 2011; N. 229; p. 53. ISSN 1980-3958. [Google Scholar]
  95. Lee, T.; Bettinger, P.; Merry, K.; Cieszewski, C. The Effects of Nearby Trees on the Positional Accuracy of GNSS Receivers in a Forest Environment. PLoS ONE 2023, 18, e0283090. [Google Scholar] [CrossRef] [PubMed]
  96. Sigrist, P.; Coppin, P.; Hermy, M. Impact of Forest Canopy on Quality and Accuracy of GPS Measurements. Int. J. Remote Sens. 1999, 20, 3595–3610. [Google Scholar] [CrossRef]
  97. Figueiredo, E.; Braz, E.; d’Oliveira, M. Manejo de Precisão Em Florestas Tropicais: Modelo Digital de Exploração Florestal; Embrapa Acre: Rio Branco, AC, Brazil, 2007; p. 184. ISBN 978-85-99190-04-3. [Google Scholar]
  98. Sist, P.; Rutishauser, E.; Peña-Claros, M.; Shenkin, A.; Hérault, B.; Blanc, L.; Baraloto, C.; Baya, F.; Benedet, F.; da Silva, K.E.; et al. The Tropical Managed Forests Observatory: A Research Network Addressing the Future of Tropical Logged Forests. Appl. Veg. Sci. 2015, 18, 171–174. [Google Scholar] [CrossRef]
  99. Andersen, H.; Reutebuch, S.E.; Mcgaughey, R.J.; Marcus, V.N.; Keller, M. Remote Sensing of Environment Monitoring Selective Logging in Western Amazonia with Repeat Lidar Fl Ights. Remote Sens. Environ. 2013, 151, 157–165. [Google Scholar] [CrossRef]
  100. Mendes, F.d.S.; Jardim, F.C.d.S.; de Carvalho, J.O.P.; Souza, D.V.; Araújo, C.B.; de Oliveira, M.G.; Leal, E.d.S. Dinâmica da estrutura da vegetação do sub-bosque sob influência da exploração em uma floresta de terra firme no município de Moju—PA. Ciência Florest. 2013, 23, 377–389. [Google Scholar] [CrossRef]
  101. da Costa, V.A.M.; Oliveira, A.d.F.d.; dos Santos, J.G.; Bovo, A.A.A.; de Almeida, D.R.A.; Gorgens, E.B. Assessing the Utility of Airborne Laser Scanning Derived Indicators for Tropical Forest Management. South. For. J. For. Sci. 2020, 82, 352–358. [Google Scholar] [CrossRef]
  102. Schulze, M.; Zweede, J. Canopy Dynamics in Unlogged and Logged Forest Stands in the Eastern Amazon. For. Ecol. Manag. 2006, 236, 56–64. [Google Scholar] [CrossRef]
  103. Kleinschroth, F.; Healey, J.R. Impacts of Logging Roads on Tropical Forests. Biotropica 2017, 49, 620–635. [Google Scholar] [CrossRef]
  104. Rocha, N.C.V.; Adami, M.; Galbraith, D.; Freitas, L.J.M. de Signature of Logging in the Brazilian Amazon Still Detected after 17 Years. For. Ecol. Manag. 2024, 561, 121850. [Google Scholar] [CrossRef]
  105. Matricardi, E.A.T.; Skole, D.L.; Pedlowski, M.A.; Chomentowski, W.; Fernandes, L.C. Assessment of Tropical Forest Degradation by Selective Logging and Fire Using Landsat Imagery. Remote Sens. Environ. 2010, 114, 1117–1129. [Google Scholar] [CrossRef]
  106. Lima, T.A.; Beuchle, R.; Langner, A.; Grecchi, R.C.; Griess, V.C.; Achard, F. Comparing Sentinel-2 MSI and Landsat 8 OLI Imagery for Monitoring Selective Logging in the Brazilian Amazon. Remote Sens. 2019, 11, 961. [Google Scholar] [CrossRef]
  107. Sloan, S.; Talkhani, R.R.; Huang, T.; Engert, J.; Laurance, W.F. Mapping Remote Roads Using Artificial Intelligence and Satellite Imagery. Remote Sens. 2024, 16, 839. [Google Scholar] [CrossRef]
  108. Sui, H.; Zhou, N.; Zhou, M.; Ge, L. Vector Road Map Updating from High-Resolution Remote-Sensing Images with the Guidance of Road Intersection Change Detection and Directed Road Tracing. Remote Sens. 2023, 15, 1840. [Google Scholar] [CrossRef]
  109. Kazama, V.S.; Corte, A.P.D.; Robert, R.C.G.; Sanquetta, C.R.; Arce, J.E.; Oliveira-Nascimento, K.A.; DeArmond, D. Global Review on Forest Road Optimization Planning: Support for Sustainable Forest Management in Amazonia. For. Ecol. Manag. 2021, 492, 119159. [Google Scholar] [CrossRef]
  110. Morais, P.d.P.; Arima, E.Y.; de Souza, Á.N.; Pereira, R.S.; Emmert, F.; Cardoso, R.M.; Miguel, E.P.; Matricardi, E.A.T. Assessment of Forest Road Models in Concession Areas in the Brazilian Amazon. Forests 2023, 14, 1388. [Google Scholar] [CrossRef]
  111. Veríssimo, A.; Pereira, D. Produção na Amazônia Florestal: Características, desafios e oportunidades. Parceiras Estratégicas 2014, 18, 13–44. [Google Scholar]
  112. Garrido Filha, I. Manejo florestal: Questões econômico-financeiras e ambientais. Estud. Av. 2002, 16, 91–106. [Google Scholar] [CrossRef]
  113. Barros, Q.S.; Rodrigues, N.M.M.; Pinheiro, R.d.M.; Ferreira, E.J.L. Potencial da tecnologia LiDAR terrestre na área florestal. Sci. Nat. 2023, 5. [Google Scholar] [CrossRef]
  114. SFB—Serviço Florestal Brasileiro Extrato De Contrato No 7/2024—Uasg 440075—Ministério do Meio Ambiente e Mudança do Clima|Serviço Florestal Brasileiro. Available online: https://diariolink.com.br/resultado/7989208 (accessed on 2 November 2024).
  115. de Freitas, L.J.M.; de Souza, A.L.; Leite, H.G.; da Silva, M.L. Análise técnica e estimativas de custos de inventário de prospecção em uma floresta estacional semidecidual submontana. Rev. Árvore 2005, 29, 65–75. [Google Scholar] [CrossRef]
  116. Cotações e Boletins. Available online: https://www.bcb.gov.br/estabilidadefinanceira/historicocotacoes (accessed on 2 November 2024).
  117. Cotações do Dólar, Euro e Risco País—Brasil, de 1994 a 2020. Available online: https://portalbrasil.net/indices_dolar.htm (accessed on 3 November 2024).
Figure 1. Location of the state of Rondônia (a), location of the Jamari National Forest within the state (b) and within the municipalities (c) along with the forest management units, and the selectively logged areas where LiDAR data were collected (d).
Figure 1. Location of the state of Rondônia (a), location of the Jamari National Forest within the state (b) and within the municipalities (c) along with the forest management units, and the selectively logged areas where LiDAR data were collected (d).
Forests 16 00130 g001
Figure 2. A schematic representation of the relative density model (RDM) calculation showing the method for determining relative vegetation density across the understory layer (adapted from D’Oliveira et al. [15]).
Figure 2. A schematic representation of the relative density model (RDM) calculation showing the method for determining relative vegetation density across the understory layer (adapted from D’Oliveira et al. [15]).
Forests 16 00130 g002
Figure 3. LiDAR data used to map the logging impact on the forest understory: (a) relative density model (RDM), highlighting the changed understory in black; (b) each created logging infrastructure digitized by RDM and the impact zone (c) determined the impacted area (density 0–20). Area 12 RDM with a 1 m resolution, based on LIDAR data collected 4 months after harvest. Logging intensity of 12.8 m3.ha−1 with a total impact on the understory of 14.3%.
Figure 3. LiDAR data used to map the logging impact on the forest understory: (a) relative density model (RDM), highlighting the changed understory in black; (b) each created logging infrastructure digitized by RDM and the impact zone (c) determined the impacted area (density 0–20). Area 12 RDM with a 1 m resolution, based on LIDAR data collected 4 months after harvest. Logging intensity of 12.8 m3.ha−1 with a total impact on the understory of 14.3%.
Forests 16 00130 g003
Figure 4. Digitization validation, based on LiDAR data of impact on the remaining forest obtained via mapping using GNSS measurements: (a) central points over the gaps of logged trees with a buffer of 5, 10, and 20 m based on field data; (b) skid trails superimposed on buffers of 5, 10, and 20 m based on field data; (c) primary and secondary roads superimposed on buffers of 5, 10, and 20 m based on field data; and (d) a polygon of the log landings superimposed on buffers of 5, 10, and 20 m from the central coordinate of the landing obtained in the field. The data correspond to Area 19 of the Jamari National Forest.
Figure 4. Digitization validation, based on LiDAR data of impact on the remaining forest obtained via mapping using GNSS measurements: (a) central points over the gaps of logged trees with a buffer of 5, 10, and 20 m based on field data; (b) skid trails superimposed on buffers of 5, 10, and 20 m based on field data; (c) primary and secondary roads superimposed on buffers of 5, 10, and 20 m based on field data; and (d) a polygon of the log landings superimposed on buffers of 5, 10, and 20 m from the central coordinate of the landing obtained in the field. The data correspond to Area 19 of the Jamari National Forest.
Forests 16 00130 g004
Figure 5. Scatter plot matrix between logging-related parameters obtained with LiDAR data. The correlation coefficients in the upper triangular panel indicate the degree of correlation, with diagonal frequency histograms to visualize the data distribution in each variable. The lower triangular panel shows the scatter plots for each pair of variables in the correlation matrix, with a straight line indicating the correlation direction.
Figure 5. Scatter plot matrix between logging-related parameters obtained with LiDAR data. The correlation coefficients in the upper triangular panel indicate the degree of correlation, with diagonal frequency histograms to visualize the data distribution in each variable. The lower triangular panel shows the scatter plots for each pair of variables in the correlation matrix, with a straight line indicating the correlation direction.
Forests 16 00130 g005
Table 1. Technical specification and costs of LiDAR data collection for Jamari National Forest, across multiple survey years. Source: [70].
Table 1. Technical specification and costs of LiDAR data collection for Jamari National Forest, across multiple survey years. Source: [70].
Period20112013201420152017201820192020
Laser
scanner
Optech 3100Optech OrionTrimble Harrier 68iOptech 3100Optech ALTM
Gemini
Optech ALTM
Gemini
Optech ALTM
Gemini
Optech ALTM
Gemini
Flight
altitude (m)
850853500750700700700700
Scanning frequency59.8 kHz67.5 kHz360 kHz55 kHz100 kHz100 kHz100 kHz100 kHz
Scanning angle11.1°11.1°15°15°15°15°15°15°
Side overlap65%65%65%70%65%70%70%70%
Mean return density
(per m2)
25.832.949.659.230.7230.1828.550.0
Price
(USD ha−1)
9.478.011.09.03.83.3
Table 2. Characteristics of sample areas and corresponding LiDAR coverage in Jamari National Forest. FMU: forest management unit; APU: annual production unit; SE: standard error.
Table 2. Characteristics of sample areas and corresponding LiDAR coverage in Jamari National Forest. FMU: forest management unit; APU: annual production unit; SE: standard error.
Area CodeFMUAPUInterval (Months) *LiDAR Coverage (ha)Scan (m3.ha−1)
1I19102.815.6
2I233133.015.4
3I326118.218.3
4I42550.412.6
5I55124.119.4
6I67132.810.4
7I73209.616.2
8I817249.817.6
9I93171.617.6
10I1015124.517.3
11I114124.619.2
12I164432.212.8
13II18305.99.9
14III142205.813.9
15III221228.614.4
16III314115.510.8
17III314102.99.9
18III44189.011.5
19III53187.012.6
20III536134.08.1
21III610224.410.7
22III119241.014.5
23III113199.713.4
24III1216624.811.1
25III1411448.512.9
Mean (±SE)227.2 (±28.3)13.8 (±0.64)
* Mean estimated time between start and end of exploration and airborne LiDAR survey in each area, based on Brazilian Forest Service information.
Table 3. Comparison between GNSS data and LiDAR-derived relative density model (RDM) measurements of logging infrastructure densities in the Jamari National Forest. RMSE: root-mean-square error; R2: coefficient of determination; r: Pearson’s correlation coefficient.
Table 3. Comparison between GNSS data and LiDAR-derived relative density model (RDM) measurements of logging infrastructure densities in the Jamari National Forest. RMSE: root-mean-square error; R2: coefficient of determination; r: Pearson’s correlation coefficient.
InfrastructureDensityStatistics
GNSSRDMDifference (%)p-ValueRMSER2r
Primary roads (m.ha−1)5.745.62−2.10.930.530.980.99 (p < 0.01)
Secondary roads (m.ha−1)21.1121.01−0.50.952.330.750.87 (p < 0.01)
Skid trails (m.ha−1)77.4198.2827.00.2726.80.680.83 (p = 0.08)
Log landings (n.ha−1)0.060.056−0.70.840.0030.920.96 (p < 0.01)
Table 4. Positional accuracy of the relative density model (RDM) logging infrastructure mapping, indicating the overlap with GNSS data within 5, 10, and 20 m buffer zones for different infrastructure types.
Table 4. Positional accuracy of the relative density model (RDM) logging infrastructure mapping, indicating the overlap with GNSS data within 5, 10, and 20 m buffer zones for different infrastructure types.
InfrastructureOverlap (%)
5 m10 m20 m
Primary roads58 ± 4.279 ± 3.698 ± 1.2
Secondary roads55 ± 2.982 ± 2.496 ± 1.0
Skid trails44 ± 2.660 ± 2.274 ± 2.3
Log landings55 ± 0.665 ± 2.292 ± 4.3
Felled tree gaps35 ± 3.147 ± 3.081 ± 2.5
Table 5. Estimated understory impact of logging infrastructure and felled tree gaps, based on LiDAR-derived relative density model measurements in the Jamari National Forest. SE: standard error.
Table 5. Estimated understory impact of logging infrastructure and felled tree gaps, based on LiDAR-derived relative density model measurements in the Jamari National Forest. SE: standard error.
Area CodeLog
Landings
TrailsSec. RoadsPrim. RoadsGapsTotal
Impact
10.56%7.00%1.51%1.81%16.93%27.81%
20.37%2.56%1.33%1.74%11.29%17.28%
30.30%3.44%1.93%0.00%12.61%18.28%
40.58%7.01%1.94%0.44%8.74%18.70%
50.39%10.47%2.06%0.00%14.04%26.95%
60.36%4.93%1.41%0.73%5.35%12.78%
70.42%5.49%1.65%0.72%5.49%13.76%
80.31%4.75%1.35%0.17%7.04%13.62%
90.59%12.24%2.36%0.55%14.92%30.66%
100.54%10.25%2.04%2.38%14.71%29.91%
110.64%10.61%2.38%1.87%16.31%31.81%
120.50%4.05%2.75%1.41%5.59%14.30%
130.34%5.46%1.13%0.87%6.06%13.86%
140.09%1.37%0.70%0.32%2.58%5.06%
150.25%5.93%1.69%0.81%5.55%14.24%
160.32%5.89%1.43%0.00%3.74%11.38%
170.27%4.85%1.65%1.28%4.91%12.96%
180.42%6.83%2.51%0.24%8.10%18.10%
190.38%6.90%2.32%1.16%11.29%22.05%
200.26%5.15%1.41%0.69%7.98%15.49%
210.30%4.67%1.40%0.50%3.74%10.61%
220.34%4.91%2.17%0.69%5.97%14.09%
230.34%8.51%1.62%0.60%10.19%21.27%
240.22%2.97%1.36%0.75%4.56%9.87%
250.28%8.00%1.52%1.18%9.30%20.28%
Mean ± (SE)0.37% ± (0.03%) 6.17% ± (0.54%)1.74% ± (0.10%)0.84% ± (0.13%)8.68% ± (0.85%)17.80% ± (1.41%)
Table 6. Mean understory impact by time interval since logging, logging intensity, and forest management unit (FMU) in Jamari National Forest.
Table 6. Mean understory impact by time interval since logging, logging intensity, and forest management unit (FMU) in Jamari National Forest.
MetricsLog Landings (%)Trails (%)Secondary Roads (%)Primary Roads (%)Gaps (%)Total
Impact (%)
Time
Interval (months)
1 to 120.43 ± 0.037.14 ± 0.631.91 ± 0.130.85 ± 0.149.47 ± 1.1219.80 ± 1.8
13 to 240.32 ± 0.055.77 ± 1.001.59 ± 0.110.90 ± 0.356.75 ± 1.6515.33 ± 2.9
>240.25 ± 0.063.13 ± 0.801.34 ± 0.250.68 ± 0.388.61 ± 2.2414.03 ± 3.0
Volume Logged
(m3.ha−1)
8 to 120.31 ± 0.025.09 ± 0.391.54 ± 0.150.63 ± 0.145.55 ± 0.6113.13 ± 0.9
12.1 to 160.37 ± 0.055.62 ± 0.751.76 ± 0.181.02 ± 0.178.74 ± 1.2817.51 ± 1.9
>160.45 ± 0.058.18 ± 1.321.97 ± 0.140.81 ± 0.3612.16 ± 1.5923.57 ± 3.1
Concession AreasFMU I0.46 ± 0.036.90 ± 0.941.89 ± 0.130.98 ± 0.2411.08 ± 1.2821.32 ± 2.16
FMU II0.345.461.130.876.0613.86
FMU III0.29 ± 0.025.50 ± 0.581.65 ± 0.140.69 ± 0.116.49 ± 0.8114.62 ± 1.48
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ferreira, L.; Bias, E.d.S.; Barros, Q.S.; Pádua, L.; Matricardi, E.A.T.; Sousa, J.J. Assessing the Impacts of Selective Logging on the Forest Understory in the Amazon Using Airborne LiDAR. Forests 2025, 16, 130. https://doi.org/10.3390/f16010130

AMA Style

Ferreira L, Bias EdS, Barros QS, Pádua L, Matricardi EAT, Sousa JJ. Assessing the Impacts of Selective Logging on the Forest Understory in the Amazon Using Airborne LiDAR. Forests. 2025; 16(1):130. https://doi.org/10.3390/f16010130

Chicago/Turabian Style

Ferreira, Leilson, Edilson de Souza Bias, Quétila Souza Barros, Luís Pádua, Eraldo Aparecido Trondoli Matricardi, and Joaquim J. Sousa. 2025. "Assessing the Impacts of Selective Logging on the Forest Understory in the Amazon Using Airborne LiDAR" Forests 16, no. 1: 130. https://doi.org/10.3390/f16010130

APA Style

Ferreira, L., Bias, E. d. S., Barros, Q. S., Pádua, L., Matricardi, E. A. T., & Sousa, J. J. (2025). Assessing the Impacts of Selective Logging on the Forest Understory in the Amazon Using Airborne LiDAR. Forests, 16(1), 130. https://doi.org/10.3390/f16010130

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop