Comparing airborne and terrestrial LiDAR with ground-based inventory metrics of vegetation structural complexity in oil palm agroforests

Vegetation structural complexity is an important component of forest ecosystems, influencing biodiversity and functioning. Due to the heterogeneous distribution of vegetation elements, structural complexity underpins ecological dynamics, species composition, microclimate, and habitat diversity. Field measurements and Light Detection and Ranging (LiDAR) data, such as airborne (ALS) and terrestrial (TLS), can assess structural characteristics of forest and agroforestry systems at various spatial scales. This assessment is urgently needed for monitoring ecosystem restoration in degraded lands (e.g., in oil palm landscapes), where it is not well-known how structural measures derived from these different approaches relate to each other. Here, we compared the degree of correlation between individual and multivariate datasets of vegetation structural complexity metrics derived from ALS, TLS, and ground-based inventory approaches. The study was conducted in a 140 ha oil palm monoculture, enriched with 52 plots in the form of tree islands representing agroforestry systems of varying sizes and planted diversity levels in Sumatra, Indonesia. Our datasets comprised 25 ALS, five TLS, and nine ground-based inventory metrics. We studied correlations among metrics related to traditional stand summary, heterogeneity, and vertical and horizontal stand structure. We used principal component analysis for data dimensionality reduction, correlation analysis to quantify the strength of relationships between metrics, and Procrustes analysis to investigate the agreement between datasets. Significant correlations were found between ALS and TLS metrics for canopy density (r = 0.79) and maximum tree height (r = 0.58) and between ALS and ground-based inventory measures of stand heterogeneity and height diversity (r between 0.60 and −0.63). Further, we observed significant agreements between the ordinations of multivariate datasets (r = 0.56 for ALS − TLS; and r = 0.46 for ALS – ground-based inventory). Our findings underline the ability of ALS to capture structural complexity patterns, especially for canopy gap dynamics and vegetation height metrics, as captured by TLS, and for measures of heterogeneity and vertical structure as captured by ground-based inventories. Our study highlights the strength of each approach and underscores the potential of integrating ALS and TLS with ground-based inventories for a comprehensive characterization of vegetation structure in complex agroforestry systems, which can provide guidance for their management and support ecosystem restoration monitoring efforts.