Biodiversity Informatics, 19, 2025, pp. 86-108 86 REMOTE SENSING ENABLES ACCURATE ASSESSMENT OF FUNCTIONAL DIVERSITY RATHER THAN SPECIES DIVERSITY IN SANDY GRASSLANDS Wen Li,, Yu Peng*, and Xiaoyue Zhang College of Life & Environmental Sciences, Minzu University of China, Beijing, China 100081 Abstract. The prediction of grassland plant diversity using satellite imagery has been the subject of intensive research. However, the accuracy of functional diversity (FD) predictions remains unclear. To address this, high-spatial-resolution WorldView-3 (WV-3) multispectral data were used to predict species diversity and FD at the pixel scale (1.2 × 1.2 m) in the central Hunshandak Sandland, Inner Mongolia, northern China. Data col- lected from 120 field plots (6 × 6 m) were employed to train and validate several statistical learning methods, with the primary objective of establishing links between 156 satellite-derived spectral and texture indices and 6 plant diversity indices. Among the various diversity indices tested, functional trait diversity—specifically Func- tional Attribute Diversity (FAD1) and Modified Functional Attribute Diversity (MFAD)—were predicted most effectively (with coefficients of determination of approximately 0.29 and 0.14, respectively; n=48) using texture indices. In contrast, species diversity (richness, H, E, or D) and other FD metrics were not well predicted by WV-3 data. Overall, WV data did not significantly improve the accuracy of plant diversity predictions in sandy grasslands. Additionally, high plot-level vegetation coverage was found to enhance the performance of spectral indices in predicting H, E, D, and FD. These results underscore the importance of accounting for variability across field conditions and demonstrate the potential of high-spatial-and-spectral-resolution satellite imagery for monitoring plant functional diversity in sandy grasslands. Key words: Plant diversity, Species diversity, Functional diversity, Texture, Remote sensing, Sandy grassland * Corresponding author: Yu Peng, yuupeng@163.com. Introduction The survival of both humans and animals depends on plant diversity (Radhamoni et al., 2023). Plant diversity can be measured at three different spatial scales: with- in-habitat diversity (α-diversity), between-habitat diversi- ty (β-diversity), and regional diversity (γ-diversity) (So- colar et al., 2016). Species richness (R), Shannon-Wiener index (H), and Simpson index (D) are used to measure α diversity. Recently, leveraging the spectral characteristics and variations of different plant species, plant diversity has been assessed on a large scale using remote sensing techniques—exhibiting distinct advantages over tradi- tional field measurement methods. Different plant species demonstrate different spectral traits. The spectral species approach assumes that there are unique, definable spectral types (“spectral species”) that can be distinguished in image processing, thus, facilitating biodiversity estimation (Rocchini & 2004; Féret & Asner, 2014). The linkage between species and “spectral species” can be validated using relatively simple to sophisticated spectral heterogeneity measures, which include measures of spectral entropy (Gillespie et al., 2008), statistical dis- persion (Gould, 2000; Palmer et al., 2002), mean Euclid- ean distances between spectral clusters derived from prin- cipal components analysis (PCA) (Oldeland et al. 2010; Rocchini, 2007), and the use of first- and second-order im- age texture analysis (Culbert et al., 2012; Viedma, et al., 2012; Wood et al., 2013). The red (RED; 630-690nm) and near infrared (NIR; 760-900 nm) bands in the multi-spec- tral images are usually selected to assess species diversity (Schowengerdt, 2007; White et al., 2010; Peng et al 2019). Various spectral vegetation indices generated from mul- tiple bands, such as Variation in Normalized Difference Vegetation Index (NDVI) (Gould, 2000; Bawa et al., 2002; Xu, 2004; Chawla et al., 2010, Kiran and Mudaliar, 2012), enhanced vegetation index (EVI) (Cabacinha, 2009; Gao X, 2000; Gallardo-Cruz et al., 2012), infrared index (IRI), middle infrared index (MIRI), atmospheric resistance veg- etation index (ARVI), and soil adjusted vegetation index (SAVI) can predict plant diversity with considerable high accuracy (Nagendra, 2001; Bawa, et al., 2002; Schoweng- erdt, 2007; Cabacinha, 2009). mailto:yuupeng@163.com Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 87 Multispectral data from satellite platforms such as Landsat satellites (Sun et al., 2023; Mapfumo et al., 2016), the Sentinel series (Mpakairi et al., 2022; Xin et al 2024), QuickBird (Rocchini, 2007), WorldView series (Cho et al., 2012; Rocchini, 2007), SPOT series (Fauvel et al., 2020), and China’s Gaofen-3 and Haisi-1 satellites (Gu et al., 2024) have been widely used for vegetation mon- itoring and plant diversity estimation. The WV-3 satellite has eight visible near infrared (VNIR) bands (400–1040 nm) at 1.24 m resolution, has advantage in assessing plant diversity (Ferreira et al., 2016). The emergence of near- earth hyperspectral spectroscopy has addressed the scale mismatch between ground-based species information and earlier satellite-based multispectral monitoring (Schnei- der et al., 2017). Hyperspectral sensors offer the techni- cal advantage of integrating fine spectral resolution with hundreds of spectral bands, improves the detection and identification of subtle differences at the plant species, functional, and genetic levels (Asner, 1998; Zhang et al., 2023). Although species diversity has been extensively ex- plored through remote sensing data, functional diversity (FD), i.e., diversity in plant species adaptation to survive in different environments and various strategies in repro- duction, pollination processes, seed-dispersal methods and life forms (Ewers & Didham, 2006; Lindborg et al., 2012), has been rarely assessed using satellite-based high spatial and spectral resolution remote sensing imageries. FD is closely related to ecosystem service and ecologi- cal process, can also be regarded as an important indicator for biodiversity conservation (Diaz and Cabido, 2001). A plant’s life form is one predictor for indicating the sur- vival ability of plants in severely environment (Evju et al., 2015). Plants with suitable life forms will gradually replace unsuitable plants in a poor environment. For ex- ample, in plant communities in the arid environments such as deserts, shrubs and semi-shrubs increasingly substitute annual plants. Species with low offspring and colonization rates are easily influenced (Higgins et al., 2003; Henle et al., 2004) and therefore, dominant species in a communi- ty are usually highly adapted to suit their environments. Therefore, this study aimed to assess the ability of WV-3 data in assessing FD in the sandy grasslands of Hunshan- dak Sandland, by comparing the performances with those Fig. 1 Location of the study area, land use types, distribution of sampled plots and subplots in the field, and cor- responding pixels in the WV-3 image of the study area. Each plot consists of 4 subplots (black circles) and corre- sponds to a 5×5 grid of WV-3 pixels. Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 88 of species diversity across various plot-level vegetation coverages. Methods Study area This study was conducted in the temperate sandy grasslands of the Hunshandak Sandland (41°46′–43°69′ N, 114°55′–116°38′ E), located in Inner Mongolia, north- ern China (Fig. 1). The region has a temperate semi-arid climate, with an annual mean temperature of 1.7°C. The monthly diurnal minimum and maximum temperatures are -18.3°C and 18.7°C, respectively, and the annual precipita- tion ranges from 250 to 350 mm, 80–90% of which occurs between May and September (Wang, 2016a). The Hun- shandak Sandland features a unique landscape comprising fixed sandy dunes, semi-fixed sandy dunes, mobile dunes, and lowlands, all of which support relatively rich plant di- versity. The study area also includes other land cover types such as water ponds and constructed land. These diverse landscape elements, combined with the region’s relatively uniform elevation, make it an ideal site for testing the ca- pacity of WV-3 data to assess both plant species diversity and functional diversity (FD) in sandy grasslands under complex background conditions. Field sampling Field surveys were conducted across 120 plots in the study area during July–August 2016, which corresponds to the peak growing season. Each plot (6 × 6 m, equiva- lent to a 5 × 5 grid of WV-3 pixels) was divided into four subplots, each with a diameter of 0.8 m (Fig. 1). In total, 480 subplots (corresponding to 480 WV-3 pixels), dis- tributed across the 120 plots, were surveyed in this study. Global Positioning System (GPS) data were differentially corrected to achieve high-precision positioning within a geographic information system (GIS) environment. Vascular plant species composition was recorded at the plot scale, with macroplot-level species lists compiled as the aggregated set of species identified across the four subplots within each plot. This nested sampling design for plot configuration has also been adopted in studies by Duccio Rocchini (2007) and Fauvel et al. (2020). All plants were identified to the species level, and the abun- dance of each species was recorded for each subplot. De- tails of the plot characteristics are provided in Table S1 in the Appendix. Plant species diversity In July and August 2016, the abundance, cover, and height of each plant species, as well as habitat categories (fixed sandy dunes, semi-fixed sandy dunes, mobile sandy dunes, lowlands, water bodies, and constructed land), were recorded. For each subplot, the number of individu- als was counted for species whose stems were either fully or partially within the subplot. For clonal species, individ- uals were considered separate if their stems or culms were more than 20 cm apart from others of the same species. Canopy cover of all species within the subplot was visual- ly estimated, and consistency in these visual estimates was ensured by having the same observer (Y. Peng) conduct all assessments across plots. Based on the collected data on plant species abun- dance, four biodiversity indices were calculated: Richness (the number of plant species in a subplot), the Shannon– Wiener index (H), Simpson’s species evenness index (D), and the Pielou index (E) (Magurran, 2004), using formu- lae (1)–(3). H = – N n N n is i i ln 1 ∑ = ln N n N n is i i ln 1 ∑ = (1), where ni is the number of individuals of the ith species, N is the total number of individuals of all the species, and ln is the natural logarithm. The value of H ranges from 0, meaning only one species is present, to 4.6, signifying high species richness and also signifying that different species in the quadrat or a community are equally abun- dant (Magurran, 2004). D = 1 – ∑ = s i i N n 1 ）（ 2 (2) The values of the Simpson species evenness index range from 0 (completely uneven) to 1 (different species occur in equal numbers). E = H/lnS (3), where S is the total number of species recorded (γ diversi- ty) and H is the Shannon–Wiener index. The plant diversity and dominant species of study plots are listed in Table S1 in the Appendix. Functional trait diversity Functional traits recorded for each species included life form (annual, biennial, perennial grass, shrub, or woody plant), seed dispersal method (gravitational, regular, wind, or animal-mediated), pollination mode (self-pollination, wind-pollination, or insect-pollination), flowering period (in months), flower longevity (in days), photosynthetic pathway (C3, C4, or CAM), and nitrogen-fixing ability (N-fixing or non-N-fixing). Based on these multiple func- tional traits, seven plant functional diversity (FD) indices were calculated using the FDiversity package (Casanoves et al., 2011; Spasojevic et al., 2014): Functional Attribute Diversity (FAD1), Modified Functional Attribute Diversi- Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 89 image window or kernel) (Anderson et al., 2009; Duro et al., 2014; Levin et al., 2007; Lucas & Carter, 2008). Here, we used the CV to link spectral heterogeneity within 5×5 image pixels to field-measured species diversity and functional diversity (FD) at each plot. Before conducting the calculations, an NDVI thresh- old of ≥0.2 was applied to distinguish vegetation from desert areas, in accordance with the method proposed by Xin et al. (2024). First, 35 spectral indices (Table S2 in the Appendix; details are provided in the ENVI manual) were calculated at the pixel level using ENVI software. Second, for each spectral index, the CV (cv), mean value (mean), majority (maj), and variance (var) were comput- ed using all pixels within a plot, yielding 140 spectral diversity values at the plot level. This approach allowed us to quantify the spectral diversity of each plot. Finally, spectral diversity values derived from all pixels within the plots were compared with field-measured species diversity and FD data. Image texture metrics, derived from multi-scale spectral values, are effective predictors of plant species richness. In this study, we also computed spectral texture values for each plot. In image texture analysis, the value of a central pixel within a moving window is determined by the spectral variability of its neighboring pixels (Hall-Beyer, 2017). We calculated one first-order texture metric and six second-order (_sec) texture metrics: angular second moment (mom), contrast (cont), dissimilarity (dis), homogeneity (hom), entropy (ent), and correlation (corr, _cor). These metrics were derived from panchromatic images and principal compo- nents (PCs) of multispectral images, where the PCs were obtained via principal component analysis (PCA) of the eight spectral bands. Second-order textures are based on the gray-level co-occurrence matrix, thus accounting for the spatial arrangement and relationships among neigh- boring pixels (Farwell et al., 2020). Detailed descriptions of these image texture metrics are available in the ENVI software manual. A single moving window size (5×5 pixels, corresponding to 6×6 m²) was selected for the analysis, as texture metrics across different window sizes are highly correlated and exhibit similar relationships with plant species richness (St-Louis et al., 2006; Culbert et al., 2012). Additionally, the CV (cv), mean (mean), majority (maj), and variance (var) of the PCs were calculated for each plot. In total, 140 spectral indices, 12 texture indices, and 4 PC indices were computed for each plot. Statistical analysis Of the 120 plots, five were excluded as their pix- els primarily covered shifting sandy land. We employed ty (MFAD), rRao, functional evenness (FEve), functional divergence (FDiv), functional dispersion (FDis), and func- tional specialization (FSpe). FAD1 represents the number of distinct attribute combinations present in the commu- nity, with values always less than or equal to species rich- ness. MFAD is a modified index calculated as the sum of standardized distances between all pairs of species in trait space. rRao is derived from ultrametric trait distances and the abundance distribution of species within the commu- nity. FEve measures the regularity of spacing between species in trait space. FDiv quantifies the spread of trait values across the range of the trait space. FDis is a multidimensional index based on multi-trait dispersion. FSpe, associated with threat categories, quan- tifies the average distinctiveness of all threatened species. The calculation formulas for each FD index are detailed in Casanoves et al. (2011). Satellite image acquisition A cloud-free WorldView-3 (WV-3, DigitalGlobe, Inc.) image covering the study area was acquired on 12 September 2015 (Fig. 1). At this time, most grasses remained leafy, making them easily distinguishable in the imagery due to their strong contrast with the surrounding sandy terrain. The WV-3 dataset included a panchromatic image with a spatial resolution of 0.30 m, accompanied by a multispectral image with a 1.20 m spatial resolution, spanning 8 spectral bands: coastal blue (427 nm), blue (482 nm), green (547 nm), yellow (604 nm), red (660 nm), red-edge (723 nm), near-infrared 1 (824 nm), and near-infrared 2 (914 nm). The WV-3 data were delivered at Level 2A. WV-3 data processing followed the meth- ods described by Lelong et al. (2020) and Cerrejón et al. (2023). First, radiometric calibration was performed to convert digital numbers to absolute radiance using the gain and offset values for each spectral band. Next, abso- lute radiance was converted to top-of-atmosphere (TOA) reflectance, and the data were orthorectified to correct geometric distortions while minimizing topographical effects. Finally, the multispectral image was fused with the panchromatic image to generate a pansharpened mul- tispectral image with a resolution of 0.30 m. Spectral and textural analysis This study assessed plant species diversity and functional trait diversity based on the spectral diversity theory, which posits that higher spectral diversity corre- sponds to greater plant diversity. Numerous studies have shown that dispersion metrics, such as the coefficient of variation (CV), serve as simple yet effective indicators of spectral heterogeneity within a sampling unit (e.g., an Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 90 a two-step approach to explore the potential of spectral and texture indices for assessing plant diversity. The first step involved selecting indices with significant Pearson’s correlation coefficients using 67 plots. The second step validated these selected indices across different vegeta- tion coverages using the remaining 48 plots. Pearson’s correlation coefficients were used to evaluate relationships between plant species diversity, functional trait diversity indices, and spectral/texture indices derived from satellite imagery. This provided an assessment of the potential of these spectral indices and texture metrics for estimating plant diversity or functional diversity. Spectral indices or texture metrics were considered optimal if they showed a statistically significant association (P < 0.05) with plant diversity indices based on Pearson’s correlation analysis. Potential indices that exhibited significant relation- ships with plant diversity were designated as final opti- mal indices only if they passed the validation test. For this test, the remaining 48 randomly selected plots were used for model validation. The potential spectral indices were required to show high consistency in predicting plant di- versity across different plant communities. We used veg- etation coverage to examine its influence on the consis- tency of the selected indices in estimating plant diversity. Plot-level vegetation coverage was categorized into three classes based on values: 0–15%, 16–26%, and 27–100%, representing low, moderate, and high community cover- age, respectively. The performance of the selected indices in the vali- dation test was evaluated using two metrics derived from the validation dataset: the correlation coefficient (r) and the significance level (p) between predicted and observed richness values. Indices with the highest r and P < 0.01 were deemed the best predictors. Additionally, cluster analysis was used to identify groups of indices with similar performance, aiming to ex- plore the underlying mechanisms of the best-performing indices in assessing plant diversity. These groups were not predefined prior to the analysis, and no a priori assump- tions were made regarding the distribution of variables (indices). The indices were z-transformed for the analysis, and results are presented as a dendrogram, with groupings based on a squared Euclidean distance matrix. Results Cluster tree The top 51 spectral and texture indices, which ex- hibited the highest number of significant correlations (P < 0.05) with plant diversity were clustered into five dis- tinct groups (Fig. 2). Indices showing significant positive correlations with the species diversity indices (D, H) and the functional diversity index (FAD1) included correlation (corr), principal component correlation (pca-cor), variance (var), principal component variance (pca-var), entropy (ent), principal component entropy (pca-ent), dissimilar- ity (dis), contrast (cont), principal component dissimilari- ty (pca-dis), and principal component contrast (pca-con). These indices all belong to image texture measures. No- tably, spectral indices based on spatial variability (e.g., coefficient of variation [CV], such as evi-cv) showed no significant relationships with species or functional diver- sity indices. Indices with the most significant relationships were identified as potential optimal indices and selected for model construction and validation. Based on this criterion, corr, pca-cor, var, and pca-var were chosen as potential indices for further analysis. Model development Six texture and spectral indices that passed the afore- mentioned tests were retained for model development. The resulting models exhibited high coefficients of de- termination (R2) and significant relationships between the indices and plant diversity indices (P < 0.05), indicating their potential as optimal indices (Table 1). Among these models, Simpson’s index (D) was significantly predicted by variance (var), correlation (Corr), principal compo- nent variance (PCA-var), principal component correlation (PCA-cor), and var-mean. All functional diversity (FD) indices—Functional Attribute Diversity (FAD1), Modified Functional Attribute Diversity (MFAD), and functional specialization (FSpe)—were significantly predicted by variance (var), correlation (Corr), principal component variance (PCA-var), and principal component correlation (PCA-cor). Model validation Using the six identified optimal spectral and texture indices listed in Table 1, plant diversity indices were cal- culated for the 48 model validation plots using the spectral and texture dataset. At the plot level, linear correlations between diversity estimates derived from spectral and tex- ture indices and field-surveyed diversity were analyzed (Fig. 3). The six selected spectral and texture indices were further compared in terms of the consistency of their rela- tionships with plant species or functional diversity across different vegetation coverage classes (0–15%, 16–26%, and 27–60%). Indices that exhibited significant correla- tions across all vegetation coverage classes were designat- ed as the best-performing indices. All models showed non-significant correlations (P > 0.05) between recorded and predicted plant diversi- Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 91 Fig. 2 Cluster dendrogram showing Pearson’s correlation coefficients between spectral/texture indices and plant diversity indices. Deep red and blue indicate significant positive and negative correlations, respectively (P < 0.05); light colors indicate non-significant correlations (P > 0.05). Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 92 Fig. 3 Linear regressions between the natural log-transformed field-measured values (x-axis) and natural log-transformed predicted values (y-axis) for plant species diversity and functional diversity, using the val- idation dataset from the central Hunshandak Sandland, northern China. Predicted values were derived from the best-performing models listed in Table 1. “low,” “mid,” and “high” indicate plot-level vegetation coverage classes of 0–15%, 16–26%, and 27–60%, respectively. Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 93 ty indices, except for the correlations between Modified Functional Attribute Diversity (MFAD) and four texture indices, and between variance (var) and Simpson’s index (D) (Fig. 3). When comparing correlation coefficient (r) values across vegetation coverage classes, all spectral and texture models exhibited similarly high r values (>0.2) for high-coverage plots (>27%), followed by low-cover- age plots, while models for medium-coverage plots had the lowest r values. Among functional diversity indices, MFAD was most effectively predicted by the selected tex- ture indices. Discussion Among the 156 spectral indices and texture metrics, only four performed relatively well in estimating plant species diversity and functional diversity (FD). These four indices were primarily texture measures: variance (var), principal component variance (PCA-var), var-mean, and correlation (corr). These indices, which reflect canopy structure and functional traits, can effectively indicate FD in sandy grasslands. Notably, leaf traits related to light capture and growth—such as photosynthetic pigments, nutrients, and leaf mass—primarily absorb and scatter light in the 350–700 nm spectral range (Ollinger, 2011). In contrast, secondary metabolites (e.g., lignin, cellulose, phenols, and tannins), which contribute to foliar defense and longevity, actively absorb and scatter near-infrared (NIR) and shortwave infrared (SWIR) radiation (Kokaly et al., 2009). This may partially explain why spectral in- dices show stronger associations with plant FD than with species diversity. Previous studies have observed that visible spectrum (VIS) variability (e.g., captured by var, var-mean, and PCA-var) is significantly higher in sandy landscapes compared to moist or wetland environments (Somers et al., 2015). The visible spectrum, dominated by pigment absorption and canopy structure, exhibits less variability among plant species but greater variability among functional traits (Kattenborn et al., 2017; Pacheco-Labrador et al 2022). Additionally, biomass variation—closely linked to life form, a key functional trait—is primarily explained by vegetation structure (Hernández-Stefanoni et al., 2014) and can be well-captured by texture metrics. In sandy grasslands, environmental conditions strongly drive functional differentiation and adaptation in plants. The NIR region, reflecting leaf cellular structure (which scatters most incident energy) and canopy biochemical traits (e.g., leaf nutrient content), can distinguish plant strategies such as C3, C4, or CAM photosynthesis, or N-fixing capacity (Castillo-Riffart et al., 2017; Pacheco- Labrador et al 2022). Collectively, these factors explain why FD, rather than species diversity, is better estimated by spectral and texture indices. A pixel represents a discrete spatial unit containing multiple objects, and the proportion of mixed objects increases with coarser resolution. A study on species- spectral diversity relationships across spatial grains, using North American floristic data, showed that high spatial and spectral resolution imagery improves plant species diversity estimation accuracy (Rocchini et al., 2014). Coarser resolution data, however, suffer from mixed-pixel issues and are less sensitive to spatial complexity (Rocchini, 2007). Previous research has found that estimation accuracy improves when spectral pixel resolution is finer than the size of the target object (Gholizadeh et al., 2019; Lopatin et al., 2017; Wang et al., 2018). Finer spatial resolution, such as that of WV-3 imagery, enhances the representation of “pure” objects within sampling areas (Rocchini, 2007; Pacheco-Labrador et al 2022) and captures more detailed spectral information. Consequently, WV-3 images exhibited higher spectral variability in 6×6 m plots compared to Aster or Landsat ETM+ imagery, likely due to reduced pixel mixing and larger effective sampling size. WV-3’s high spectral resolution further strengthens its capacity to estimate plant diversity. Table 1 Selected spectral indices, texture indices, models, and parameters for estimating plant species diversity and functional diversity using the training dataset Richness H D FDA1 MFAD FSpe Var Y = 4.761-0.001x Y = 1.249-0.000317x Y = 0.650+0.00014x Y = 4.814-0.001x Y = 1.007+0.000307x Y = 2.366+0.000346x R2=0.048, P = 0.041 R2=0.081, P = 0.011 R2=0.094, P = 0.007 R2=0.061, P = 0.025 R2=0.053, P = 0.034 R2=0.183, P = 0.000 Corr Y = 3.947+0.852x Y = 1.046+0.206x Y = 0.569+0.08x Y = 3.945+0.911x Y = 0.797+0.219x Y = 2.534-0.154x R2=0.074, P = 0.026 R2=0.072, P = 0.016 R2=0.061, P = 0.025 R2=0.076, P = 0.014 R2=0.059, P = 0.027 R2=0.068, P = 0.021 PCA-var Y = 4.755-0.001x Y = 1.247-0.000318x Y = 0.649+0.000141x Y = 4.807-0.001x Y = 1.005+0.000305x Y = 2.369+0.00034x R2=0.047, P = 0.044 R2=0.079, P = 0.012 R2=0.093, P = 0.007 R2=0.059, P = 0.028 R2=0.051, P = 0.038 R2=0.172, P = 0.000 PCA-cor - Y = 1.045+0.207x Y = 0.569+0.080x Y = 3.941+0.915x Y = 0.796+0.219x Y = 2.534-0.155x R2=0.071, P = 0.016 R2=0.06, P = 0.025 R2=0.076, P = 0.015 R2=0.059, P = 0.028 R2=0.068, P = 0.021 Ari-maj - - - - - - Var-mean Y = 3.944+0.855x - Y = 0.534+0.346x - - - R2=0.06, P = 0.026 R2=0.065, P = 0.021 Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 94 In this study, texture metrics were derived from prin- cipal components (PCs) generated via principal compo- nent analysis (PCA) of 12 spectral bands. PCs maximize variance and are non-spatial, and it can be assumed that the first two PCs from 12 bands likely correlate with the most variable top-of-canopy structural and chemical traits (Roth et al., 2016; Dahlin, 2016). Our results align with previous studies using image texture to distinguish veg- etation structural patterns between habitats (Wood et al., 2012), which also reported weak relationships (0.01 < R² < 0.3). Other studies have confirmed that image texture metrics effectively characterize vegetation heterogeneity, including foliage height diversity (Wood et al., 2012), successional stage (Jakubauskas, 1997), and structural complexity (Guo et al., 2004)—consistent with the use of vegetation indices (e.g., NDVI, EVI) to quantify bio- physical aspects of functional traits. Texture metrics from WV-3 can capture spatial variations in vertical structures of mixed grasslands under different grazing regimes (Guo et al., 2004), supporting the utility of medium-resolution image textures for detecting vegetation heterogeneity. We used 5×5 pixel windows to calculate texture metrics, a size well-suited for grassland ecosystems. A study on North American plant species richness found that spectral diver- sity explained little variance, whereas the spatial extent of sampling units (floras) explained much of the variance in plant diversity (Rocchini et al., 2014). In the present study, correlations between observed and predicted values were weak. For plant species rich- ness, the strongest predictor was PCA-var (R² = 0.19), which is lower than the spectral diversity indices (R² = 0.4–0.6) derived from airborne hyperspectral data (Wang et al., 2018). Two key limitations likely contribute to this weakness. First, strong noise from the desert background in the study area weakens the correlation between WV-3- derived spectral indices and plant diversity. Rocchini et al. (2010, 2017) noted that pixels should be at least as large as the sampling unit, particularly when using spectral het- erogeneity to estimate local species diversity. Additional- ly, the trade-off between noise from high resolution and information loss from low resolution must be considered. We speculate that WV-3’s fine resolution may exacerbate this issue: in vegetation-rich areas, numerous pixels may capture small shade patches cast by canopies (Nagendra & Rocchini, 2008), while in sparse vegetation, pixels are dominated by strong desert signals—both reducing the per- ceived strength of vegetation signals. Second, vegetation indices like NDVI struggle to detect plant signals in des- ert-vegetation mosaics, making it difficult to extract veg- etation texture from mixed spectra, especially with fixed thresholds used in preprocessing (Hernández-Stefanoni et al., 2014). In this study, an NDVI threshold of 0.2 was used to distinguish vegetation from desert, but this may have excluded small-stemmed plants. A study conducted in a subalpine grassland of the Italian Alps showed that spectral diversity calculated via the Spectral Angle Map- per (SAM) proved to be a more effective proxy for biodi- versity within the same ecosystem—whereas the broader spectral diversity approach failed to estimate α-diversity (Imran et al., 2024). This finding suggests that introducing additional novel spectral indices may enhance the accu- racy of plant diversity estimation (Cherif et al., 2023). In addition, remote sensing for assessing forest functional diversity has typically demonstrated high estimation ac- curacy (Cimoli et al., 2024; Zeng et al., 2023). For exam- ple, in a subtropical evergreen and deciduous broad-leaved mixed forest, airborne LiDAR-derived parameters were found to correlate well with in situ plot-level morphologi- cal data (R² ≥ 0.67) (Zeng et al., 2023). Currently, improv- ing the estimation accuracy of both species and functional diversity in grasslands remains a key challenge for future research. Conclusion We demonstrate that texture metrics derived from high-spatial-resolution satellite imagery effectively pre- dict plot-level patterns of plant functional traits in sandy grasslands, highlighting their potential to capture envi- ronmental heterogeneity not detected by more conven- tional heterogeneity metrics. While positive correlations were observed between texture metrics and both plant species richness and functional diversity, these relation- ships varied across diversity indices and texture metrics. Notably, for the functional index MFAD, the relationship remained consistently significant and positive. Our results underscore the complexity of the heterogeneity-diversity relationship, emphasizing the need for further investiga- tion into these relationships using different satellite data sources and across diverse ecosystems. We conclude that texture metrics show promise as a tool for modeling func- tional diversity—rather than species diversity—in areas with sparse vegetation. Acknowledgements We thank WN Zhang, FC Li, TT Yang, J Li, and C Ma for their field survey. This research was funded by the General Program of the National Natural Science Founda- tion of China (32271555). Data and Code Availability Data will be made available on request. Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 95 Competing Interests The authors declare no conflict of interest. References Asner, G. P., 1998. Biophysical and biochemical sources of variability in canopy reflectance. Remote Sens. En- viron. 64(3): 23–253. https://doi.org/10.1016/S0034- 4257(98)00014-5 Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., and Liddell, M. J., 2009a. Leaf chemical and spectral diversity in Australian tropical forests. Ecol. Appl. 19: 236–253. https://doi.org/10.1890/08-0541.1 Bawa, K., Rose, J., Ganeshaiah, K. N., Barve, N., Kiran, M. C., and Umashanker, R., 2002. Assessing biodi- versity from space: Example from Western Ghats, In- dia. Conserv. Ecol. 6(1): 7. https://www.ecologyand- society.org/vol6/iss1/art7/ Cabacinha, C. D., and de Castro, S. S., 2009. Relationships between floristic diversity and vegetation indices, forest structure and landscape metrics of fragments in Brazilian Cerrado. For. Ecol. Manag. 257: 2157– 2165. https://doi.org/10.1016/j.foreco.2009.02.024 Carlson, K. M., Asner, G. P., Hughes, R. F., Ostertag, R., and Martin, R. E., 2007. Hyperspectral remote sensing of canopy biodiversity in Hawaiian lowland rainforests. Ecosystems 10: 536–549. https://doi. org/10.1007/s10021-006-9013-5 Casanoves, F., Pla, L., Di Rienzo, J. A., and Díaz, S., 2011. FDiversity: A software package for the inte- grated analysis of functional diversity. Methods Ecol. Evol. 2: 233–237. https://doi.org/10.1111/j.2041- 210X.2010.00088.x Castillo-Riffart, I., Galleguillos, M., Lopatin, J., and Pe- rez-Quezada, A. J. F., 2017. Predicting vascular plant diversity in anthropogenic peatlands: comparison of modeling methods with free satellite data. Re- mote Sens. 9(7): 681–692. https://doi.org/10.3390/ rs9070681 Castro-Esau, K. L., Sánchez-Azofeifa, G. A., and Rivard, B., 2006b. Comparison of spectral indices obtained using multiple spectroradiometers. Remote Sens. Environ. 103: 276–288. https://doi.org/10.1016/j. rse.2006.05.005 Ceccato, P., Gobron, N., Flasse, S., and Asner, G. P., 2002. Designing a spectral index to estimate vegetation wa- ter content from remote sensing data: Part I: Theoret- ical approach. Remote Sens. Environ. 82(2–3): 188– 197. https://doi.org/10.1016/S0034-4257(02)00074-7 Cerrejón, C., Valeria, O., and Fenton, N. J., 2023. Esti- mating lichen α- and β-diversity using satellite data at different spatial resolutions. Ecol. Indic. 149: 110173. https://doi.org/10.1016/j.ecolind.2023.110173 Chawla, A., Kumar, A., Rajkumar, J., Singh, R. D. R., Thukral, A. K., and Ahuja, P. S., 2010. Correlation of multispectral satellite data with plant species diversity vis-à-vis soil characteristics in a landscape of West- ern Himalayan Region, India. Appl. Remote Sens. 1: 013503. https://doi.org/10.1117/1.3486043 Cherif, E., Feilhauer, H., Berger, K., Asner, G. P., Martin, R. E., Ollinger, S. V., Wessman, C. A., and Schmidtle- in, S., 2023. From spectra to plant functional traits: Transferable multi-trait models from heterogeneous and sparse data. Remote Sens. Environ. 292: 113580. https://doi.org/10.1016/j.rse.2023.113580 Cho, M. A., Mathieu, R., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., Liddell, M. J., and Hughes, R. F., 2012. Mapping tree species composition in South African savannas using an integrated airborne spec- tral and LiDAR system. Remote Sens. Environ. 125: 214–226. https://doi.org/10.1016/j.rse.2012.05.015 Cimoli, E., Lucieer, A., Malenovsk, Z., Woodgate, W., Janoutová, R., Turner, D., Asner, G. P., and Martin, R. E., 2024. Mapping functional diversity of canopy physiological traits using UAS imaging spectrosco- py. Remote Sens. Environ. 302: 113958. https://doi. org/10.1016/j.rse.2024.113958 Culbert, P. D., Radeloff, V. C., St-Louis, V., Flather, C. H., Rittenhouse, C. D., Albright, T. P., Asner, G. P., and Martin, R. E., 2012. Modeling broad-scale patterns of avian species richness across the midwestern Unit- ed States with measures of satellite image texture. Remote Sens. Environ. 118: 140–150. https://doi. org/10.1016/j.rse.2011.11.013 Diaz, S., and Cabido, M., 2001. Vive la difference: Plant functional diversity matters to ecosystem process- es. Trends Ecol. Evol. 16: 646–655. https://doi. org/10.1016/S0169-5347(01)02224-5 Doğan, H. M., and Doğan, M., 2020. Understanding and modeling plant biodiversity of Nallıhan (A3-An- kara) forest ecosystem by means of geographic in- formation systems and remote sensing. Doctoral Dissertation, Ankara University, Ankara, Turkey. Available from: https://openaccess.ankara.edu.tr/han- dle/20.500.12573/26455 Duro, D. C., Girard, J., King, D. J., Fahrig, L., Mitchell, S., Lindsay, K., and Tischendorf, L., 2014. Predicting species diversity in agricultural environments using Landsat TM imagery. Remote Sens. Environ. 144: 214–225. https://doi.org/10.1016/j.rse.2013.12.013 Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 96 Evju, M., Blumentrath, S., Skarpaas, O., Stabbetorp, O. E., and Sverdrup-Thygeson, A., 2015. Plant species oc- currence in a fragmented grassland landscape, the im- portance of species traits. Biodivers. Conserv. 24(3): 547–561. https://doi.org/10.1007/s10531-014-0847-2 Ewers, R. M., and Didham, R. K., 2006. Confounding fac- tors in the detection of species responses to habitat fragmentation. Biol. Rev. 81: 117–142. https://doi. org/10.1017/S1464793105006950 Farwell, L. S., Elsen, P. R., Razenkova, E., Pidgeon, A. M., and Radeloff, V. C., 2020. Habitat heterogene- ity captured by 30‐m resolution satellite image tex- ture predicts bird richness across the United States. Ecol. Appl. 30(8): e02157. https://doi.org/10.1002/ eap.2157 Fauvel, M., Lopes, M., Dubo, T., Rivers-Moore, J., Frison, P.-L., Gross, N., and Ouin, A., 2020. Prediction of plant diversity in grasslands using Sentinel-1 and -2 satel- lite image time series. Remote Sens. Environ. 237: 111536. https://doi.org/10.1016/j.rse.2019.111536 Feilhauer, H., Faude, U., and Schmidtlein, S., 2011. Com- bining Isomap ordination and imaging spectroscopy to map continuous floristic gradients in a heteroge- neous landscape. Remote Sens. Environ. 115: 2513– 2524. https://doi.org/10.1016/j.rse.2011.05.020 Feilhauer, H., Thonfeld, F., Faude, U., He, K., Rocchi- ni, D., and Schmidtlein, S., 2013. Assessing floristic composition with multispectral sensors: A compari- son based on monotemporal and multiseasonal field spectra. Int. J. Appl. Earth Obs. Geoinf. 21: 218–229. https://doi.org/10.1016/j.jag.2012.11.005 Féret, J.-B., and Asner, G. P., 2014. Mapping tropical forest canopy diversity using high-fidelity imaging spectroscopy. Ecol. Appl. 24: 1289–1296. https://doi. org/10.1890/13-0897.1 Ferreira, M. P., Zortea, M., Zanotta, D. C., Shimabukuro, Y. E., and Filho, C. R. D. S., 2016. Mapping tree species in tropical seasonal semi-deciduous forests with hyperspectral and multispectral data. Remote Sens. Environ. 179: 66–78. https://doi.org/10.1016/j. rse.2016.04.025 Gallardo-Cruz, J. A., Meave, J. A., González, E. J., Leb- rija-Trejos, E. E., Romero-Romero, M. A., Asner, G. P., Martin, R. E., and Hughes, R. F., 2012. Predict- ing tropical dry forest successional attributes from space: Is the key hidden in image texture? PLoS ONE 7(2): e30506. https://doi.org/10.1371/journal. pone.0030506 Gao, X., Huete, A. R., Ni, W. G., and Miura, T., 2000. Op- tical-biophysical relationships of vegetation spectra without background contamination. Remote Sens. Environ. 74: 609–620. https://doi.org/10.1016/S0034- 4257(00)00103-3 Gholizadeh, H., Amani, M., Townsend, P. A., Zygielbaum, A. I., Helzer, C. J., Asner, G. P., Martin, R. E., and Ollinger, S. V., 2019. Detecting prairie biodiversity with airborne remote sensing. Remote Sens. Environ. 221: 38–49. https://doi.org/10.1016/j.rse.2018.10.030 Gillespie, T. W., Foody, G. M., Rocchini, D., Giorgi, A. P., and Saatchi, S., 2008. Measuring and modeling biodi- versity from space. Prog. Phys. Geogr. 32: 203–221. https://doi.org/10.1177/0309133308094513 Gould, W., 2000. Remote sensing of vegetation, plant species richness, and regional biodiversity hotspots. Ecol. Appl. 10(6): 1861–1870. https://doi.org/10.189 0/1051-0761(2000)010[1861:RSOVPS]2.0.CO;2 Gu, C., Liang, J., Liu, X. Y., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., and Liddell, M. J., 2024. Application and prospects of hyperspectral remote sensing in monitoring plant diversity in grassland. Chin. J. Appl. Ecol. 35(5): 1397–1407. https://doi.org /10.13287/j.1001-9332.202405.024 Guisan, A., and Thuiller, W., 2005. Predicting species distribution: Offering more than simple habitat mod- els. Ecol. Lett. 8: 993–1009. https://doi.org/10.1111/ j.1461-0248.2005.00787.x Hall-Beyer, M., 2017. GLCM Texture: A Tutorial v. 3.0, March 2017. University of Calgary, Calgary, Alber- ta, Canada. Available from: https://www.ucalgary.ca/ hallbey/tutorial He, K. S., Rocchini, D., Neteler, M., and Nagendra, H., 2011. Benefits of hyperspectral remote sensing for tracking plant invasions. Divers. Distrib. 17: 381–392. https://doi.org/10.1111/j.1472-4642.2011.00794.x Hector, A., and Bagchi, R., 2007. Biodiversity and ecosys- tem multifunctionality. Nature 448: 188–190. https:// doi.org/10.1038/nature05987 Henle, K., Davies, K. F., Kleyer, M., Margules, C., and Settele, J., 2004. Predictors of species sensitivity to frag- mentation. Biodivers. Conserv. 13: 207–251. https:// doi.org/10.1023/B:BIOC.0000025823.16415.69 Hernández-Stefanoni, J. L., Dupuy, J. M., Johnson, D. K., Birdsey, R., Tun-Dzul, F., Peduzzi, A., Asner, G. P., and Martin, R. E., 2014. Improving species diversi- ty and biomass estimates of tropical dry forests using airborne LiDAR. Remote Sens. 6: 4741–4763. https:// doi.org/10.3390/rs6064741 Higgins, S. I., Lavorel, S., and Revilla, E., 2003. Esti- mating plant migration rates under habitat loss and fragmentation. Oikos 101: 354–366. https://doi. org/10.1034/j.1600-0706.2003.12158.x Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 97 Imran, H. A., Sakowska, K., Gianelle, D., Rocchini, D., Dalponte, M., Scotton, M., and Vescovo, L., 2024. Assessing plant trait diversity as an indicator of spe- cies α- and β-diversity in a subalpine grassland of the Italian Alps. Remote Sens. Ecol. Conserv. 10: 328– 342. https://doi.org/10.1002/rse2.370 John, R., Chen, J., Lu, N., Guo, K., Liang, C., Wei, Y., Asner, G. P., and Martin, R. E., 2008. Predicting plant diversity based on remote sensing products in the semi-arid region of Inner Mongolia. Remote Sens. En- viron. 112(5): 2018–2032. https://doi.org/10.1016/j. rse.2007.12.014 Kattenborn, T., Fassnacht, F., Pierce, S., Lopatin, J., Grime, J. P., and Schmidtlein, S., 2017. Linking plant strategies and plant traits derived by radiative trans- fer modelling. J. Veg. Sci. 28: 42–49. https://doi. org/10.1111/jvs.12533 Kiran, G. S., and Mudaliar, A., 2012. Remote sensing & geo-informatics technology in evaluation of forest tree diversity. Asian J. Plant Sci. Res. 2(3): 237–242. https://doi.org/10.5958/2249-7429.2012.00041.1 Kokaly, R. F., Asner, G. P., Ollinger, S. V., Martin, M. E., and Wessman, C. A., 2009. Characterizing cano- py biochemistry from imaging spectroscopy and its application to ecosystem studies. Remote Sens. En- viron. 113(S1): S78–S91. https://doi.org/10.1016/j. rse.2008.10.017 Lelong, C. C. D., Tshingomba, U. K., and Soti, V., 2020. Assessing Worldview-3 multispectral imaging abili- ties to map the tree diversity in semi-arid parklands. Int. J. Appl. Earth Obs. Geoinf. 93: 102211. https:// doi.org/10.1016/j.jag.2020.102211 Levin, N., Shmida, A., Levanoni, O., Tamari, H., and Kark, S., 2007. Predicting mountain plant richness and rarity from space using satellite-derived vegeta- tion indices. Divers. Distrib. 13: 692–703. https://doi. org/10.1111/j.1472-4642.2007.00366.x Lindborg, R., Helm, A., Bommarco, R., Heikkinen, R. K., Kuhn, I., Asner, G. P., Martin, R. E., and Ford, A. J., 2012. Effect of habitat area and isolation on plant trait distribution in European forests and grasslands. Ecog- raphy 35: 356–363. https://doi.org/10.1111/j.1600- 0587.2011.06962.x Lopatin, J., Fassnacht, F. E., Kattenborn, T., and Schmidtle- in, S., 2017. Mapping plant species in mixed grass- land communities using close range imaging spec- troscopy. Remote Sens. Environ. 201: 12–23. https:// doi.org/10.1016/j.rse.2017.05.034 Lucas, K., and Carter, G., 2008. The use of hyperspec- tral remote sensing to assess vascular plant species richness on Horn Island, Mississippi. Remote Sens. Environ. 112: 3908–3915. https://doi.org/10.1016/j. rse.2008.06.014 Magurran, A. E., 2004. Measuring Biological Diver- sity. Oxford: Blackwell Publishing. https://doi. org/10.1002/9780470996267 Mapfumo, R. B., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., Liddell, M. J., Hughes, R. F., and Os- tertag, R., 2016. The relationship between satellite-de- rived indices and species diversity across African sa- vanna ecosystems. Int. J. Appl. Earth Obs. Geoinf. 52: 306–317. https://doi.org/10.1016/j.jag.2016.07.013 Mpakairi, K. S., Dube, T., Dondofema, F., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., and Liddell, M. J., 2022. Spatial Characterization of Vegetation Diversity in Groundwater Dependent Ecosystems Us- ing In-Situ and Sentinel-2 MSI Satellite Data. Remote Sens. 14: 2995. https://doi.org/10.3390/rs14132995 Nagendra, H., 2001. Using remote sensing to assess bio- diversity. Int. J. Remote Sens. 22: 2377–2400. https:// doi.org/10.1080/01431160110043459 Nagendra, H., and Rocchini, D., 2008. Satellite imagery applied to biodiversity study in the tropics: The devil is in the detail. Biodivers. Conserv. 17: 3431–3442. https://doi.org/10.1007/s10531-008-9457-0 Oldeland, J., Wesuls, D., Rocchini, D., Schmidt, M., and Jürgens, N., 2010. Does using species abundance data improve estimates of species diversity from remotely sensed spectral heterogeneity? Ecol. Indic. 10: 390– 396. https://doi.org/10.1016/j.ecolind.2009.11.006 Oldeland, J., Dorigo, W., Wesuls, D., and Jürgens, N., 2010. Mapping bush encroaching species by seasonal differences in hyperspectral imagery. Remote Sens. 2: 1416–1438. https://doi.org/10.3390/rs2071416 Ollinger, S. V., 2011. Sources of variability in canopy re- flectance and the convergent properties of plants. New Phytol. 189: 375–394. https://doi.org/10.1111/j.1469- 8137.2010.03527.x Palmer, M. W., Earls, P. G., Hoagland, B. W., White, P. S., and Wohlgemuth, T., 2002. Quantitative tools for perfecting species lists. Environmetrics 13: 121–137. https://doi.org/10.1002/env.507 Pacheco-Labrador, J., Migliavacca, M., Ma, X., Mahe- cha, M., Carvalhais, N., Weber, U., Asner, G. P., and Martin, R. E., 2022. Challenging the link between functional and spectral diversity with radiative trans- fer modeling and data. Remote Sens. Environ. 280: 113170. https://doi.org/10.1016/j.rse.2022.113170 Peng, Y., Fan, M., Bai, L., Sang, W., Feng, J., Zhao, Z., and Tao, Z., 2019. Identification of the best hyper- Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 98 spectral indices in estimating plant species richness in sandy grasslands. Remote Sens. 11: 588. https://doi. org/10.3390/rs11050588 Radhamoni, H. V. N., Queenborough, S. A., Arietta, A. Z. A., Asner, G. P., Martin, R. E., Ford, A. J., Metcal- fe, D. J., and Liddell, M. J., 2023. Local- and land- scape-scale drivers of terrestrial herbaceous plant di- versity along a tropical rainfall gradient in Western Ghats, India. J. Ecol. 111(5): 1021–1036. https://doi. org/10.1111/1365-2745.14010 Rocchini, D., Dadalt, L., Delucchi, L., Neteler, M., and Palmer, M. W., 2014. Disentangling the role of re- motely sensed spectral heterogeneity as a proxy for North American plant species richness. Community Ecol. 15(1): 37–43. https://doi.org/10.1080/1745097 9.2014.881461 Rocchini, D., 2007. Effects of spatial and spectral reso- lution in estimating ecosystem α-diversity by satel- lite imagery. Remote Sens. Environ. 111: 423–434. https://doi.org/10.1016/j.rse.2007.05.014 Rocchini, D., Balkenhol, N., Carter, G. A., Foody, G. M., Gillespie, T. W., He, K., Asner, G. P., and Martin, R. E., 2010. Remotely sensed spectral heterogeneity as a proxy for species diversity: Recent advances and open challenges. Ecol. Inform. 5: 318–329. https:// doi.org/10.1016/j.ecoinf.2010.03.004 Rocchini, D., Chiarucci, A., and Loiselle, S. A., 2004. Testing the spectral variation hypothesis by using sat- ellite multispectral images. Acta Oecol. 26: 117–120. https://doi.org/10.1016/j.actao.2004.03.005 Rocchini, D., Marcantonio, M., and Riccota, C., 2017. Measuring Rao’s Q diversity index from remote sens- ing: An open source solution. Ecol. Indic. 72: 234– 238. https://doi.org/10.1016/j.ecolind.2016.10.021 Rocchini, D., Ricotta, C., and Chiarucci, A., 2007. Us- ing satellite imagery to assess plant species rich- ness: The role of multispectral systems. Appl. Veg. Sci. 10: 325–331. https://doi.org/10.1111/j.1654- 109X.2007.00316.x Roth, K. L., Casas, A., Huesca, M., Ustin, S. L., Alsina, M. M., Mathews, S. A., and Whiting, M. L., 2016. Leaf spectral clusters as potential optical leaf functional types within California. Remote Sens. Environ. 184: 229–246. https://doi.org/10.1016/j.rse.2016.08.015 Schmidtlein, S., Feilhauer, H., and Bruelheide, H., 2012. Mapping plant strategy types using remote sensing. J. Veg. Sci. 23: 395–405. https://doi.org/10.1111/j.1654- 109X.2011.01230.x Schneider, F. D., Morsdorf, F., Schmid, B., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., and Liddell, M. J., 2017. Mapping functional diversity from re- motely sensed morphological and physiological forest traits. Nat. Commun. 8: 1441. https://doi.org/10.1038/ s41467-017-01532-8 Schowengerdt, R., 2007. Remote Sensing: Models and Methods for Image Processing. Oxford: Elsevier. 515 p. https://doi.org/10.1016/B978-012374372-1.00001-1 Socolar, J. B., Gilroy, J. J., Kunin, W. E., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., and Lid- dell, M. J., 2016. How should beta-diversity inform biodiversity conservation? Trends Ecol. Evol. 31(1): 67–80. https://doi.org/10.1016/j.tree.2015.10.004 Somers, B., Asner, G. P., Martin, R. E., Anderson, C. B., Knapp, D. E., Wright, S. J., Ostertag, R., and Hughes, R. F., 2015. Mesoscale assessment of changes in trop- ical tree species richness across a bioclimatic gradi- ent in Panama using airborne imaging spectroscopy. Remote Sens. Environ. 167: 111–120. https://doi. org/10.1016/j.rse.2015.06.032 Somers, B., Verbesselt, J., Ampe, E. M., Sims, N., Ver- straeten, W. W., and Coppin, P., 2010b. Spectral mix- ture analysis to monitor defoliation in mixed aged Eucalyptus globulus Labill plantations in southern Australia using Landsat 5TM and EO-1 Hyperion data. Int. J. Appl. Earth Obs. Geoinf. 12: 270–277. https://doi.org/10.1016/j.jag.2010.01.006 Spasojevic, M. J., Grace, J. B., Harrison, S., and Dam- schen, E. I., 2014. Functional diversity supports the physiological tolerance hypothesis for plant species richness along climatic gradients. J. Ecol. 102: 447– 455. https://doi.org/10.1111/1365-2745.12214 Sun, W., Liu, W., Wang, Y., Asner, G. P., Martin, R. E., Ford, A. J., Metcalfe, D. J., and Liddell, M. J., 2023. Progress and prospects of global hyperspectral re- mote sensing research on wetlands from 2010 to 2022. J. Remote Sens. 27(6): 1281–1299. https://doi. org/10.11834/jrs.20230374 Tuanmu, M. N., and Jetz, M., 2015. A global, remote sens- ing-based characterization of terrestrial habitat het- erogeneity for biodiversity and ecosystem modeling. Global Ecol. Biogeogr. 24: 1329–1339. https://doi. org/10.1111/geb.12316 Vaglio Laurin, G., Chan, J. C.-W., Chen, Q., Lindsell, J. A., Coomes, D. A., Asner, G. P., Martin, R. E., and Ford, A. J., 2014. Biodiversity mapping in a Tropical West African forest with airborne hyperspectral data. PLoS ONE 9(6): e97910. https://doi.org/10.1371/ journal.pone.0097910 Viedma, O., Torres, I., Pérez, B., and Moreno, J. M., 2012. Modeling plant species richness using reflec- tance and texture data derived from QuickBird in a recently burned area of central Spain. Remote Sens. Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 99 Environ. 119: 208–221. https://doi.org/10.1016/j. rse.2011.12.015 Wang, R., Gamon, J. A., Montgomery, R. A., and Townsend, P. A., 2016a. Seasonal variation in the NDVI–species richness relationship in a prairie grass- land experiment (Cedar Creek). Remote Sens. 8: 128– 131. https://doi.org/10.3390/rs8020128 Wang, R., Gamon, J. A., Cavender-Bares, J., Townsend, P. A., and Zygielbaum, A. I., 2018. The spatial sen- sitivity of the spectral diversity-biodiversity relation- ship: An experimental test in a prairie grassland. Ecol. Appl. 28: 541–556. https://doi.org/10.1002/eap.1724 White, J. C., Gómez, C., Wulder, M. A., and Coops, N. C., 2010. Characterizing temperate forest structur- al and spectral diversity with Hyperion EO-1 data. Remote Sens. Environ. 114: 1576–1589. https://doi. org/10.1016/j.rse.2010.02.018 Wood, E. M., Pidgeon, A. M., Radeloff, V. C., and Keul- er, N. S., 2012. Image texture as a remotely sensed measure of vegetation structure. Remote Sens. En- viron. 121: 516–526. https://doi.org/10.1016/j. rse.2012.02.020 Wood, E. M., Pidgeon, A. M., Radeloff, V. C., and Keuler, N. S., 2013. Image texture predicts avian density and species richness. PLoS ONE 8(5): e63211. https://doi. org/10.1371/journal.pone.0063211 Xin, J. X., Li, J. N., Zeng, Q. Q., Peng, Y., Asner, G. P., Martin, R. E., Ford, A. J., and Liddell, M. J., 2024. High-precision estimation of plant alpha diversi- ty in different ecosystems based on Sentinel-2 data. Ecol. Indic. 166: 112527. https://doi.org/10.1016/j. ecolind.2024.112527 Zhang, Y. W., Guo, Y. P., Tang, R., Tang, Z. Y., Asner, G. P., Martin, R. E., Ford, A. J., and Liddell, M. J., 2023. Progress and trends of application of hyperspectral remote sensing in plant diversity research. Natl. Remote Sens. Bull. 27(11): 2467–2483. https://doi. org/10.11834/jrs.20231048 Zheng, Z., Schmid, B., Zeng, Y., Schuman, M. C., Zhao, D., Schaepman, M. E., Asner, G. P., and Martin, R. E., 2023. Remotely sensed functional diversity and its association with productivity in a subtropical for- est. Remote Sens. Environ. 290: 113530. https://doi. org/10.1016/j.rse.2023.113530 Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 100 Appendix Table S1 Coverage, habitat type, plant diversity, and dominant species in study plots Plotid Coverage Altitude Habitat type Richness H D E Dominant species 2 4 1331.0 Semi-fixed 3 0.80 0.45 0.72 Salsola collina 3 14 1330.0 Semi-fixed 3 0.76 0.43 0.70 Ulmus pumila 5 53 1334.0 Semi-fixed 4 1.13 0.63 0.81 Polygonum divaricatum 6 11 1329.0 Semi-fixed 5 1.09 0.53 0.67 Salsola collina 7 26 1328.0 Semi-fixed 3 0.78 0.45 0.71 Polygonum divaricatum 10 6 1329.0 Semi-fixed 4 0.96 0.51 0.69 Salsola collina 12 11 1330.0 Semi-fixed 4 1.28 0.69 0.92 Salsola collina 15 17 1323.0 Semi-fixed 5 1.46 0.74 0.91 Polygonum divaricatum 16 26 1323.4 Semi-fixed 8 1.71 0.78 0.82 Potentilla supina 17 9 1326.0 Semi-fixed 5 1.08 0.56 0.67 Salsola collina 18 20 1329.1 Semi-fixed 5 1.46 0.74 0.91 Lappula myosotis 19 25 1328.0 Semi-fixed 4 1.24 0.69 0.89 Polygonum divaricatum 20 8 1321.0 Semi-fixed 5 1.07 0.53 0.66 Echinops davuricus 21 8 1324.0 Semi-fixed 4 0.77 0.39 0.56 Salsola collina 22 5 1322.0 Semi-fixed 4 1.05 0.58 0.76 Agropyron mongolicum 23 4 1323.7 Semi-fixed 3 1.05 0.64 0.96 Potentilla supina 24 22 1324.0 Semi-fixed 5 1.43 0.74 0.89 Polygonum divaricatum 25 11 1322.0 Semi-fixed 6 1.59 0.77 0.89 Phragmites australis 26 40 1321.0 Semi-fixed 5 1.45 0.74 0.90 Bromus ircutensis 27 37 1321.0 Semi-fixed 4 1.08 0.62 0.78 Potentilla supina 28 12 1326.4 Semi-fixed 5 0.94 0.46 0.59 Leymus chinensis 29 27 1325.0 Semi-fixed 4 1.27 0.69 0.92 Polygonum divaricatum 30 13 1326.0 Semi-shift 3 0.96 0.59 0.87 Artemisia ordosica 31 15 1327.4 Semi-shift 4 1.05 0.60 0.76 Artemisia ordosica 32 22 1330.0 Semi-shift 2 0.68 0.49 0.98 Artemisia ordosica 33 10 1326.8 Semi-shift 3 0.63 0.34 0.57 Leymus secalinus 34 24 1329.0 Semi-shift 3 0.93 0.57 0.84 Artemisia ordosica 35 14 1325.0 Semi-shift 4 1.26 0.69 0.91 Artemisia ordosica 36 13 1323.3 Semi-shift 5 1.26 0.66 0.79 Artemisia ordosica 37 10 1324.0 Semi-shift 4 1.21 0.67 0.87 Agropyron mongolicum 38 11 1323.6 Semi-shift 4 1.24 0.68 0.89 Artemisia ordosica 39 15 1321.3 Semi-shift 3 0.95 0.56 0.86 Polygonum divaricatum 40 32 1324.0 Semi-shift 2 0.64 0.44 0.92 Artemisia ordosica 42 29 1323.0 Semi-shift 4 1.27 0.69 0.92 Polygonum divaricatum 43 9 1322.0 Semi-shift 6 1.37 0.66 0.76 Agropyron mongolicum 44 8 1322.0 Semi-shift 3 0.74 0.45 0.67 Salsola collina 45 14 1319.0 Lowland 5 1.49 0.75 0.93 Agropyron mongolicum 46 25 1322.5 Lowland 4 1.00 0.53 0.72 Leymus secalinus 47 25 1319.2 Lowland 6 1.38 0.69 0.77 Leymus secalinus 48 26 1319.0 Lowland 6 1.31 0.61 0.73 Agropyron mongolicum Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 101 49 27 1317.0 Lowland 3 0.82 0.53 0.74 Carex duriuscula 50 16 1320.0 Lowland 6 1.68 0.80 0.94 Salsola collina 51 45 1319.7 Lowland 5 1.27 0.67 0.79 Carex duriuscula 52 19 1319.0 Lowland 6 1.46 0.71 0.81 Agropyron mongolicum 53 34 1320.6 Lowland 7 1.17 0.54 0.60 Agropyron mongolicum 54 39 1322.0 Lowland 5 1.44 0.73 0.89 Carex duriuscula 55 10 1320.0 Lowland 5 1.38 0.70 0.85 Artemisia ordosica 56 85 1326.0 Lowland 7 1.55 0.73 0.80 Agropyron mongolicum 57 41 1318.7 Lowland 6 1.27 0.65 0.71 Carex duriuscula 58 20 1320.0 Lowland 4 1.06 0.59 0.77 Agropyron mongolicum 59 25 1319.4 Lowland 4 0.86 0.50 0.62 Agropyron mongolicum 60 38 1334.0 Fixed-land 4 1.03 0.61 0.74 Leymus chinensis 61 78 1328.0 Fixed-land 2 0.28 0.15 0.40 Leymus secalinus 62 23 1329.0 Fixed-land 2 0.53 0.35 0.76 Leymus chinensis 63 24 1329.3 Fixed-land 5 1.33 0.70 0.83 Agropyron mongolicum 64 16 1324.5 Fixed-land 4 1.21 0.67 0.87 Corispermum stauntonii 65 18 1326.4 Fixed-land 4 1.15 0.64 0.83 Corispermum stauntonii 66 14 1326.7 Fixed-land 3 1.06 0.64 0.96 Artemisia ordosica 67 31 1327.1 Fixed-land 5 1.32 0.68 0.82 Carex duriuscula 68 23 1326.1 Semi-fixed 4 1.18 0.65 0.85 Artemisia ordosica 69 13 1325.0 Fixed-land 4 1.12 0.61 0.81 Corispermum stauntonii 70 19 1324.4 Fixed-land 4 1.08 0.61 0.78 Artemisia ordosica 71 43 1326.3 Fixed-land 5 1.04 0.50 0.65 Carex duriuscula 72 42 1330.2 Semi-fixed 4 1.14 0.65 0.82 Artemisia ordosica 73 17 1327.6 Semi-fixed 3 0.92 0.57 0.84 Artemisia ordosica 74 32 1327.4 Semi-fixed 7 1.68 0.79 0.86 Agropyron mongolicum 75 28 1328.1 Semi-fixed 6 1.61 0.77 0.90 Artemisia ordosica 76 30 1325.5 Semi-fixed 7 1.90 0.84 0.98 Artemisia ordosica 77 22 1325.9 Semi-fixed 3 0.51 0.26 0.46 Artemisia ordosica 78 23 1324.4 Semi-fixed 4 1.22 0.68 0.88 Artemisia ordosica 79 15 1325.0 Semi-fixed 5 1.20 0.60 0.75 Chenopodium glaucum 80 22 1324.4 Semi-fixed 5 1.33 0.68 0.83 Agropyron mongolicum 81 8 1325.0 Semi-fixed 4 0.99 0.54 0.71 Agropyron mongolicum 82 23 1324.1 Semi-fixed 2 0.48 0.30 0.70 Artemisia ordosica 83 18 1324.0 Semi-fixed 5 1.49 0.75 0.93 Artemisia ordosica 84 35 1319.0 Fixed-land 8 1.48 0.68 0.71 Carex duriuscula 85 22 1322.3 Fixed-land 4 1.18 0.64 0.85 Agropyron mongolicum 86 27 1322.0 Fixed-land 5 1.39 0.72 0.87 Artemisia ordosica 87 32 1324.0 Fixed-land 4 1.07 0.59 0.77 Artemisia ordosica 88 16 1324.0 Fixed-land 5 1.47 0.74 0.91 Artemisia ordosica 89 17 1325.0 Semi-fixed 5 1.39 0.72 0.87 Artemisia ordosica 90 15 1324.7 Semi-fixed 3 1.04 0.63 0.94 Artemisia ordosica 91 6 1326.0 Fixed-land 4 1.39 0.75 1.00 Ulmus pumila 92 33 1322.3 Fixed-land 4 1.23 0.68 0.89 Agropyron mongolicum 93 19 1337.0 Lowland 5 1.30 0.67 0.81 Leymus chinensis 94 45 1335.0 Lowland 8 1.63 0.74 0.79 Cleistogenes caespitosa 95 20 1330.0 Lowland 5 1.49 0.75 0.92 Agropyron cristatum Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 102 96 58 1327.0 Lowland 4 1.07 0.61 0.77 Cleistogenes caespitosa 97 19 1327.7 Lowland 5 1.27 0.65 0.79 Asparagus schoberioides 98 36 1328.0 Lowland 8 1.65 0.75 0.79 Asparagus schoberioides 99 29 1327.0 Lowland 8 1.80 0.80 0.87 Cleistogenes caespitosa 100 37 1327.0 Lowland 6 1.18 0.63 0.66 Cleistogenes caespitosa 101 28 1329.0 Lowland 5 1.28 0.66 0.80 Thymus serpyllum 102 36 1327.0 Lowland 5 1.42 0.72 0.88 Artemisia frigida 103 19 1326.0 Lowland 5 1.32 0.69 0.82 Agropyron mongolicum 104 28 1329.0 Lowland 6 1.64 0.79 0.92 Artemisia ordosica 105 17 1326.0 Lowland 5 1.36 0.70 0.84 Cleistogenes caespitosa 106 34 1326.0 Lowland 4 0.89 0.53 0.64 Artemisia frigida 107 37 1326.0 Lowland 5 1.52 0.76 0.94 Agropyron mongolicum 108 13 1336.0 Fixed-land 5 1.54 0.78 0.96 Leymus chinensis 109 8 1332.4 Fixed-land 4 1.04 0.60 0.75 Leymus chinensis 110 5 1326.2 Fixed-land 4 1.19 0.66 0.86 Leymus chinensis 111 7 1327.0 Fixed-land 5 1.44 0.73 0.89 Leymus chinensis 112 17 1325.6 Fixed-land 5 0.90 0.43 0.56 Carex duriuscula 113 7 1327.4 Fixed-land 4 1.21 0.67 0.87 Leymus chinensis 114 20 1326.7 Fixed-land 4 0.94 0.48 0.68 Lappula myosotis 115 6 1326.7 Fixed-land 4 1.29 0.70 0.93 Leymus chinensis 116 5 1324.0 Fixed-land 4 1.21 0.66 0.88 Leymus chinensis 117 17 1326.0 Fixed-land 4 0.45 0.19 0.32 Carex duriuscula 118 15 1326.0 Fixed-land 3 0.91 0.54 0.83 Cleistogenes caespitosa 119 20 1325.0 Fixed-land 4 1.30 0.71 0.94 Thymus serpyllum 120 11 1325.0 Fixed-land 4 1.03 0.55 0.75 Carex duriuscula 121 12 1324.6 Fixed-land 3 0.90 0.53 0.82 Lappula myosotis 122 29 1326.0 Fixed-land 6 0.92 0.45 0.51 Carex duriuscula Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 103 Table S2 T he nam e, form ula, and ecological m eanings of spectral indices used in the present study O rder N am e Form ula Ecological m eanings R eference 1 A nthocyanin R eflec- tance Index 1 (A R I1) one m easure of stressed vegetation G itelson et al., 2001 2 A nthocyanin R eflec- tance Index 2 (A R I2) is a m odification to the A R I1 that de- tects higher concentrations of anthocy- anins in vegetation. Liu et al., 2022 3 A tm ospherically R esis- tant Vegetation Index (A RV I) an enhancem ent to the N D V I that is relatively resistant to atm ospheric factors K aufm an et al., 1992 4 Enhanced Vegetation Index (EV I) an im provem ent over N D V I by opti- m izing the vegetation signal in areas of high leaf area index H uete et al., 2002 5 G reen A tm ospherically R esistant Index (G A R I) is m ore sensitive to a w ide range of chlorophyll concentrations and less sensitive to atm ospheric effects than N D V I. G itelson et al., 1996 6 G reen C hlorophyll In- dex (G C I) to estim ate leaf chlorophyll content across a w ide range of plant species. G itelson et al., 2003 7 G reen D ifference Vege- tation Index (G D V I) This index w as originally designed w ith color-infrared photography to predict nitrogren requirem ents for corn. Sripada et al., 2006 8 G reen Leaf Index (G LI) to m easure w heat cover, w here the red, green, and blue digital num bers (D N s) range from 0 to 255. Louhaichi et al., 2001 9 G reen N orm alized D if- ference Vegetation Index (G N D V I) This index w as originally designed w ith color-infrared photography to predict nitrogren requirem ents for corn. This index is m ore sensitive to chlorophyll concentration than N D V I. G itelson et al., 1998 Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 104 10 G reen O ptim ized Soil A djusted Vegetation Index (G O SAV I) This index is sim ilar to N D V I except that it m easures the green spectrum from 540 to 570 nm instead of the red spectrum . This index is m ore sensitive to chlorophyll concentration than N D V I. Sripada et al., 2006 11 G reen R atio Vegetation Index (G RV I) sensitive to photosynthetic rates in forest canopies Sripada et al., 2006 12 G reen Soil A djusted Vegetation Index (G SA - V I) This index w as originally designed w ith color-infrared photography to predict nitrogren requirem ents for corn. It is sim ilar to SAV I, but it sub- stitutes the green band for red. Sripada et al., 2006 13 G reen Vegetation Index (G V I) This index m inim izes the effects of background soil w hile em phasizing green 14vegetation K authet al., 1976 14 G lobal Environm en- tal M onitoring Index (G EM I) is sim ilar to N D V I but is less sensitive to atm ospheric effects. Pinty et al., 1992 15 Infrared Percentage Veg- etation Index (IPV I) Like N D V I, is com putationally faster. Values range from 0 to 1. C rippen et al., 1990 16 M odified C hlorophyll A bsorption R atio Index (M C A R I) This index indicates the relative abun- dance of chlorophyll.It is designed prim arily to am plify the leaf-level chlorophyll signal, thereby indicating vegetation “physiological health and nutritional status,” rather than canopy structure or biom ass. Zarco-Tejada et al., 2005 https://envi.geoscene.cn/help/Subsystems/envi/Content/Vegetation Analysis/BroadbandGreenness.htm#Soil Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 105 17 M odified C hlorophyll A bsorption R atio Index Im proved (M C A R I2) This index can m ore accurately and consistently characterize the canopy’s “green biom ass” and its “gross prim a- ry productivity potential.” It rem ains sensitive to variations in leaf area index even in dense, high-biom ass stands w here N D V I tends to saturate, and it m inim izes the influence of changing chlorophyll concentrations H aboudane et al., 2004 18 M odified N on-Linear Index (M N LI) This index is an enhancem ent to the N on-Linear Index (N LI) that incor- porates the Soil A djusted Vegetation Index (SAV I) Yang et al., 2021 19 M odified Soil A djust- ed Vegetation Index 2 (M SAV I2) It reduces soil noise and increases the dynam ic range of the vegetation signal. Q i et al., 1994 20 M odified Triangu- lar Vegetation Index (M TV I) suitable for LA I estim ations H aboudane et al., 2004 21 M odified Triangular Vegetation Index - Im - proved (M TV I2) a better predictor of green LA I Eitelet al., 2007 22 N orm alized D ifference M ud Index (N D M I) This index highlights m uddy or shal- low w ater pixels Liu et al., 2022 23 O ptim ized Soil A djust- ed Vegetation Index (O SAV I) is best used in areas w ith relatively sparse vegetation Peng et al., 2018 24 PC A -cv C oeffi cients of variation in principle com ponents 25 R enorm alized D iffer- ence Vegetation Index (R D V I) to highlight healthy vegetation R oujean et al., 1995 26 R ed G reen R atio Index (R G R I) indicates the relative expression of leaf redness caused by anthocyanin to that of chlorophyll. G am on et al., 1999 27 Soil A djusted Vegetation Index (SAV I) This index is best used in areas w ith relatively sparse vegetation H uete et al., 1988 Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 106 28 Sum G reen Index (SG I) SG I is the m ean of reflectance across the 500 nm to 600 nm portion of the spectrum . This index is used for detecting chang- es in vegetation greenness. Lobell et al., 2003 29 Transform ed C hloro- phyll A bsorption R eflec- tance Index (TC A R I) This index indicates the relative abun- dance of chlorophyll.It is highly sen- sitive to chlorophyll concentration; by am plifying the chlorophyll absorption trough, it rapidly and quantitatively reflects vegetation nitrogen status and early physiological stress. H aboudane et al., 2004 30 Transform ed D iffer- ence Vegetation Index (TD V I) for m onitoring vegetation cover in urban environm ents. B annari et al., 2002 31 Triangular G reenness Index (TG I) it is suitable for R G B cam eras. R aym ond et al., 2011 32 V isible A tm ospherically R esistant Index (VA R I) to estim ate the fraction of vegetation in a scene G itelson et al., 2002 33 W ide D ynam ic R ange Vegetation Index (W D R - V I) is m ore sensitive to a w ider range of vegetation fractions and to changes in LA I. G ITELSO N et al., 2004 34 W orldView Im proved Vegetative Index(W IV I) The com m on range for green vegeta- tion. W olf et al., 2010 35 W orldView N on-H om o- geneous Feature D iffer- ence (W N H FD ) to identify features that contrast highly against the background. W olf et al., 2010 Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 107 Reference Bannari, A., Asalhi, H., and Teillet, P. M., 2002. Transformed difference vegetation index (TDVI) for vegetation cover mapping. Proc. IEEE Int. Geosci. Remote Sens. Symp. (IGARSS 2002), Toronto, ON, Canada, 24–28 June 2002, Vol. 5: 3053–3055. Crippen, R. E., 1990. Calculating the vegetation in- dex faster. Remote Sens. Environ. 34(1): 71–73. https://doi.org/10.1016/0034-4257(90)90009-3 Eitel, J. U. H., Long, D. S., Gessler, P. E., and Smith, A. C., 2007. Using in-situ measurements to evaluate the new RapidEye™ satellite series for prediction of wheat nitrogen status. Remote Sens. Environ. 108(2–3): 4183–4190. https://doi. org/10.1016/j.rse.2007.08.013 Gamon, J. A., Surfus, J. S., 1999. Assessing leaf pigment content and activity with a reflectom- eter. New Phytol. 143: 105–117. https://doi. org/10.1046/j.1469-8137.1999.00480.x Gitelson, A. A., Merzlyak, M., 1998. Remote Sens- ing of Chlorophyll Concentration in Higher Plant Leaves. Adv. Space Res. 22(5): 689–692. https:// doi.org/10.1016/S0273-1177(98)00144-8 Gitelson, A. A., Gritz, Y., Merzlyak, M. N., 2003. Re- lationships between leaf chlorophyll content and spectral reflectance and algorithms for non-de- structive chlorophyll assessment in higher plant leaves. Plant Physiol. 160(3): 271–282. https:// doi.org/10.1071/PP02082 Gitelson, A. A., Kaufman, Y. J., Merzlyak, M. N., 1996. Use of a green channel in remote sensing of global vegetation from EOS-MODIS. Re- mote Sens. Environ. 58(3): 289–298. https://doi. org/10.1016/0034-4257(96)00042-7 Gitelson, A. A., Merzlyak, M. N., Chivkunova, O. B., 2001. Optical properties and nondestructive es- timation of anthocyanin content in plant leaves. Photochem. Photobiol. 74(1): 38–45. https://doi. org/10.1562/0031-8655(2001)074[0038:OPA- NEO]2.0.CO;2 Gitelson, A. A., 2004. Wide dynamic range vegeta- tion index for remote quantification of biophysi- cal characteristics of vegetation. J. Plant Physiol. 161(2): 165–173. https://doi.org/10.1078/0176- 1617-01084 Haboudane, D., Miller, J. R., Pattey, E., and Smith, A. C., 2004. Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop canopies: Modeling and validation in the context of precision agriculture. Remote Sens. Environ. 90(3): 337–352. https://doi. org/10.1016/j.rse.2004.01.011 Huete, A. R., Didan, K., Miura, T., and Smith, A. C., 2002. Overview of the radiometric and biophys- ical performance of the MODIS vegetation indi- ces. Remote Sens. Environ. 83: 195–213. https:// doi.org/10.1016/S0034-4257(02)00096-2 Huete, A. R., 1988. A soil-adjusted vegetation index (SAVI). Remote Sens. Environ. 25(3): 295–309. https://doi.org/10.1016/0034-4257(88)90106-X Kaufman, Y. J., Tanré, D., 1992. Atmospherically Re- sistant Vegetation Index (ARVI) for EOS-MO- DIS. IEEE Trans. Geosci. Remote Sens. 30(2): 261–270. https://doi.org/10.1109/36.129241 Kauth, R. J., 1976. Tasselled Cap—A Graphic De- scription of the Spectral-Temporal Development of Agricultural Crops as Seen by Landsat. LARS Symp.: 41–51. https://ageconsearch.umn.edu/re- cord/222424 Liu, L. X., Pang, Y., Sang, G. Q., and Smith, A. C., 2022. Applying high resolution remote sensing to access tree species diversity of monsoon- al broad-leaved evergreen forest in Puer City. Acta Ecol. Sin. 42(20): 8398–8410. https://doi. org/10.5846/stxb202106231710 Lobell, D., Asner, G., 2003. Hyperion Studies of Crop Stress in Mexico. Proc. 12th Annu. JPL Airborne Earth Sci. Workshop, Pasadena, CA. https:// earth.jpl.nasa.gov/files/airs2003_abstracts.pdf Louhaichi, M., Borman, M. M., Johnson, D. E., 2001. Spatially Located Platform and Aerial Photography for Documentation of Grazing Im- pacts on Wheat. Geocarto Int. 16: 65–70. https:// doi.org/10.1080/10106040108544267 Peng, Y., Fan, M., Song, J. Y., and Smith, A. C., 2018. Assessment of plant species diversity based on hyperspectral indices at a fine scale. Sci. Rep. 8: 4776. https://doi.org/10.1038/s41598-018- 23183-5 Pinty, B., Verstraete, M., 1992. GEMI: a Non-Lin- ear Index to Monitor Global Vegetation From Satellites. Vegetatio 101(1): 15–20. https://doi. org/10.1007/BF00044317 Qi, J., Chehbouni, A., Huete, A. R., Kerr, Y. H., and Sorooshian, S., 1994. A modified soil ad- justed vegetation index. Remote Sens. Environ. 48(2): 119–126. https://doi.org/10.1016/0034- 4257(94)90134-1 Raymond Hunt, E., Daughtry, C. S. T., Eitel, J. U. H., and Smith, A. C., 2011. Remote sensing leaf chlorophyll content using a visible band in- dex. Agron. J. 103(4): 1090–1099. https://doi. org/10.2134/agronj2010.0477 Roujean, J.-L., Breon, F.-M., 1995. Estimating PAR absorbed by vegetation from bidirectional re- flectance measurements. Remote Sens. Environ. Wen Li et al. – Remote Sensing to Assess Functional Diversity in Sandy Grasslands 108 51(3): 375–384. https://doi.org/10.1016/0034- 4257(95)00048-4 Sripada, R. P., Heiniger, R. W., White, J. G., and Smith, A. C., 2006. Aerial Color Infrared Photog- raphy for Determining Early In-Season Nitrogen Requirements in Corn. Agron. J. 98(4): 968–977. https://doi.org/10.2134/agronj2005.0146 Wolf, A., 2010. Using WorldView 2 Vis-NIR MSI Imagery to Support Land Mapping and Fea- ture Extraction Using Normalized Difference Index Ratios. Unpublished Report, Longmont, CO: DigitalGlobe. Available from: https://www. maxar.com/sites/default/files/documents/world- view-2-imagery-land-mapping.pdf Yang, Z., Willis, P., and Mueller, R., 2008. Impact of band-ratio enhanced AWiFS image on crop clas- sification accuracy. Proc. 2008 ASPRS Annu. Conf., Portland, OR, USA, 28 April–2 May 2008. https://www.asprs.org/wp-content/uploads/pers/ 2008conference/proceedings/0295.pdf Zarco-Tejada, P. J., Ustin, S., Whiting, M., 2005. Temporal and spatial relationships between with- in-field yield variability in cotton and high-spa- tial hyperspectral remote sensing imagery. Agron. J. 97: 641–653. https://doi.org/10.2134/ agronj2004.0148