Combining geostatistical models and remotely sensed data to improve tropical tree richness mapping

Information on the spatial distribution and composition of biological communities is essential in designing effective strategies for biodiversity conservation and management. Reliable maps of species richness across the landscape can be useful tools for these purposes. Acquiring such information through traditional survey techniques is costly and logistically difficult. The kriging interpolation method has been widely used as an alternative to predict spatial distributions of species richness, as long as the data are spatially dependent. However, even when this requirement is met, researchers often have few sampled sites in relation to the area to be mapped. Remote sensing provides an inexpensive means to derive complete spatial coverage for large areas and can be extremely useful for estimating biodiversity. The aim of this study was to combine remotely sensed data with kriging estimates (hybrid procedures) to evaluate the possibility of improving the accuracy of tree species richness maps. We did this through the comparison of the predictive performance of three hybrid geostatistical procedures, based on tree species density recorded in 141 sampling quadrats: co-kriging (COK), kriging with external drift (KED), and regression kriging (RK). Reflectance values of spectral bands, computed NDVI and texture measurements of Landsat 7 TM imagery were used as ancillary variables in all methods. The R² values of the models increased from 0.35 for ordinary kriging to 0.41 for COK, and from 0.39 for simple regression estimates to 0.52 and 0.53 when using simple KED and RK, respectively. The R² values of the models also increased from 0.60 for multiple regression estimates to 0.62 and 0.66 when using multiple KED and RK, respectively. Overall, our results demonstrate that these procedures are capable of greatly improving estimation accuracy, with multivariate RK being clearly superior, because it produces the most accurate predictions, and because of its flexibility in modeling multivariate relationships between tree richness and remotely sensed data. We conclude that this is a valuable tool for guiding future efforts aimed at conservation and management of highly diverse tropical forests.     

Remotely sensed vegetation phenology for describing and predicting the biomes of South Africa

Questions: What are the patterns of remotely sensed vegetation phenology, including their inter‐annual variability, across South Africa? What are the phenological attributes that contribute most to distinguishing the different biomes? How well can the distribution of the recently redefined biomes be predicted based on remotely sensed, phenology and productivity metrics? Location: South Africa. Method: Ten‐day, 1 km, NDVI AVHRR were analysed for the period 1985 to 2000. Phenological metrics such as start, end and length of the growing season and estimates of productivity, based on small and large integral (SI, LI) of NDVI curve, were extracted and long‐term means calculated. A random forest regression tree was run using the metrics as the input variables and the biomes as the dependent variable. A map of the predicted biomes was reproduced and the differentiating importance of each metric assessed. Results: The phenology metrics (e.g. start of growing season) showed a clear relationship with the seasonality of rainfall, i.e. winter and summer growing seasons. The distribution of the productivity metrics, LI and SI were significantly correlated with mean annual precipitation. The regression tree initially split the biomes based on vegetation production and then by the seasonality of growth. A regression tree was used to produce a predicted biome map with a high level of accuracy (73%). Main conclusion: Regression tree analysis based on remotely sensed metrics performed as good as, or better than, previous climate‐based predictors of biome distribution. The results confirm that the remotely sensed metrics capture sufficient functional diversity to classify and map biome level vegetation patterns and function.     

Mapping species density of trees,shrubs and vines in a tropical forest,using field measurements,satellite multiespectral imagery and spatial interpolation

We estimated the number of species in a tropical forest landscape in Quintana Roo, Mexico, based on the relationship between reflectance values of satellite imagery and field measurements of plant species density (mean number of species per plot). Total species density as well as that of tree, shrub and vine species were identified from 141 sampling quadrats (16543 individuals sampled). Spatial prediction of plant diversity was performed using universal kriging. This approach considered the linear relationship between plant species density and reflectance values of Thematic Mapper™, as well as the spatial dependence of the observations. We explored the linear relationships between spectral properties of TM bands and the species density of trees, shrubs and vines, using regression analysis. We employed Akaike Information Criterion (AIC) to select a set of candidate models. Based on Akaike weights, we calculated model-averaged parameters. Linear regression between number of species and reflectance values of TM bands yielded regression residuals. We used variogram analysis to analyze the spatial structure of these residuals. Results show that accounting for spatial autocorrelation in the residual variation improved model R² from 0.57 to 0.66 for number of all species, from 0.58 to 0.65 for number of tree species, from 0.26 to 0.41 for number of shrub species and from 0.13 to 0.17 for species density of vines. The empirical models we developed can be used to predict landscape-level species density in the Yucatan Peninsula, helping to guide and evaluate management and conservation strategies.     

Predicting and quantifying the structure of tropical dry forests in South Florida and the Neotropics using spaceborne imagery

Aim This research examines environmental theories and remote sensing methods that have been hypothesized to be associated with tropical dry forest structure. Location Tropical dry forests of South Florida and the Neotropics. Methods Field measurements of stand density, basal area and tree height were collected from 22 stands in South Florida and 30 stands in the Neotropics. In South Florida, field measurements were compared to climatic (temperature, precipitation, hurricane disturbance) and edaphic (rockiness, soil depth) variables, spectral indices (NDVI, IRI, MIRI) from Landsat 7 ETM+, and estimates of tree height from the Shuttle Radar Topography Mission (SRTM) and the National Elevation Dataset (NED). Environmental variables associated with tropical dry forest structure in South Florida were compared to tropical dry forest in other Neotropical sites. Results There were significant correlations among temperature and precipitation, and stand density and tree height in South Florida. There were significant correlations between (i) stand density and mean NDVI and standard deviation of NDVI, (ii) MIRI and stand density, basal area and mean tree height, and (iii) estimates of tree height from SRTM with maximum tree height. In the Neotropics, there were no relationships between temperature or precipitation and tropical dry forest structure, however, Neotropical sites that experience hurricane disturbance had significantly shorter tree heights and higher stand densities. Main conclusions It is possible to predict and quantify the forest structure characteristics of tropical dry forests using climatic data, Landsat 7 ETM+ imagery and SRTM data in South Florida. However, results based on climatic data are region‐specific and not necessarily transferable between tropical dry forests at a continental spatial scale. Spectral indices from Landsat 7 ETM+ can be used to quantify forest structure characteristics, but SRTM data are currently not transferable to other regions. Hurricane disturbance has a significant impact on forest structure in the Neotropics.     

Confounding effects of snow cover on remotely sensed vegetation indices of evergreen and deciduous trees: An experimental study

Located at northern latitudes and subject to large seasonal temperature fluctuations, boreal forests are sensitive to the changing climate, with evidence for both increasing and decreasing productivity, depending upon conditions. Optical remote sensing of vegetation indices based on spectral reflectance offers a means of monitoring vegetation photosynthetic activity and provides a powerful tool for observing how boreal forests respond to changing environmental conditions. Reflectance-based remotely sensed optical signals at northern latitude or high-altitude regions are readily confounded by snow coverage, hampering applications of satellite-based vegetation indices in tracking vegetation productivity at large scales. Unraveling the effects of snow can be challenging from satellite data, particularly when validation data are lacking. In this study, we established an experimental system in Alberta, Canada including six boreal tree species, both evergreen and deciduous, to evaluate the confounding effects of snow on three vegetation indices: the normalized difference vegetation index (NDVI), the photochemical reflectance index (PRI), and the chlorophyll/carotenoid index (CCI), all used in tracking vegetation productivity for boreal forests. Our results revealed substantial impacts of snow on canopy reflectance and vegetation indices, expressed as increased albedo, decreased NDVI values and increased PRI and CCI values. These effects varied among species and functional groups (evergreen and deciduous) and different vegetation indices were affected differently, indicating contradictory, confounding effects of snow on these indices. In addition to snow effects, we evaluated the contribution of deciduous trees to vegetation indices in mixed stands of evergreen and deciduous species, which contribute to the observed relationship between greenness-based indices and ecosystem productivity of many evergreen-dominated forests that contain a deciduous component. Our results demonstrate confounding and interacting effects of snow and vegetation type on vegetation indices and illustrate the importance of explicitly considering snow effects in any global-scale photosynthesis monitoring efforts using remotely sensed vegetation indices.     

The Utility of Spectral Indices from Landsat ETM+ for Measuring the Structure and Composition of Tropical Dry Forests1

There is a growing emphasis on developing methods for quantifying the structure and composition of tropical forests that can be applied over large landscapes, especially for tropical dry forests that are severely fragmented and have a high conservation priority. This study investigates the relationships between various measures of forest structure (annual woody increment, canopy closure, stand density, stand basal area) and composition (tree species diversity, tree community composition) measured in semi‐deciduous tropical dry forests on islands in Lago Guri, Venezuela and three spectral indices derived from Landsat ETM+: Normalized Difference Vegetation Index (NDVI), Infrared Index (IRI), and Mid‐Infrared Index (MIRI). Even though there were significant autocorrelations among spectral indices, there were significant differences in the relationships between spectral indices and forest attributes. IRI was not significantly correlated with any of the structural variables while MIRI was correlated with canopy closure and NDVI values were correlated with canopy closure as well as annual woody increment. NDVI and MIRI were both related to relative tree diversity and all three indices were associated with aspects of tree species composition. Based on the results of this study, it appears that spectral indices, and in particular NDVI, may be useful indicators of forest attributes in tropical dry forest habitats. Further research needs to be undertaken to identify if the results of this study can be applied to other tropical dry forests at a global spatial scale.     

Primary productivity and species richness: relationships among functional guilds, residency groups and vagility classes at multiple spatial scales

One of the major determinants of species richness is the amount of energy available, often measured as primary productivity. Heterogeneity of environmental variables has also been found to influence species richness. Predicting species distributions across landscapes and identifying areas that have high species richness, or vulnerable groups of species, is useful for land management. Remotely sensed data may help identify such areas, with the Normalized Difference Vegetation Index (NDVI) providing an estimate of primary productivity. We examined the relationship between maximum productivity (NDVI), heterogeneity of productivity, and species richness of birds and butterflies at multiple spatial scales. We also explored relationships between productivity, functional guilds and residency groups of birds, and vagility classes of butterflies. Positive linear relationships between maximum NDVI and number of functional guilds of birds were found at two spatial scales. We also found positive linear relationships between maximum NDVI and species richness of neotropical migrant birds at two scales. Heterogeneity of NDVI, by contrast, was negatively associated with number of functional guilds of birds and species richness of resident birds. Maximum NDVI was associated with species richness of all butterflies and of the most vagile butterflies. No association was found between heterogeneity of NDVI and species richness of butterflies. In the Great Basin, where high greenness and availability of water correspond to areas of high species richness and maximum NDVI, our results suggest that NDVI can provide a reliable basis for stratifying surveys of biodiversity, by highlighting areas of potentially high biodiversity across large areas. Measures of heterogeneity of NDVI appear to be less useful in explaining species richness.     

Spatial predictability of juvenile fish species richness and abundance in a coral reef environment

Juvenile reef fish communities represent an essential component of coral reef ecosystems in the current focus of fish population dynamics and coral reef resilience. Juvenile fish survival depends on habitat characteristics and is, following settlement, the first determinant of the number of individuals within adult populations. The goal of this study was to provide methods for mapping juvenile fish species richness and abundance into spatial domains suitable for micro and meso-scale analysis and management decisions. Generalized Linear Models predicting juvenile fish species richness and abundance were developed according to spatial and temporal environmental variables measured from 10 m up to 10 km in the southwest lagoon of New Caledonia. The statistical model was further spatially generalized using a 1.5-m resolution, independently created, remotely sensed, habitat map. This procedure revealed that : (1) spatial factors at 10 to 100-m scale explained up to 71% of variability in juvenile species richness, (2) a small improvement (75%) was gained when a combination of environmental variables at different spatial and temporal scales was used and (3) the coupling of remotely sensed data, geographical information system tools and point-based ecological data showed that the highest species richness and abundance were predicted along a narrow margin overlapping the coral reef flat and adjacent seagrass beds. Spatially explicit models of species distribution may be relevant for the management of reef communities when strong relationships exist between faunistic and environmental variables and when models are built at appropriate scales.     

Modelling tree diversity in a highly fragmented tropical montane landscape

Aim There is an urgent need for conservation in threatened tropical forest regions. We explain and predict the spatial variation of α (i.e. within plot) and β (i.e. between plot) tree diversity in a tropical montane landscape subjected to a high deforestation rate. A major aim is to demonstrate the potential of a method that combines data from multiple sources (field data, remote sensing imagery and GIS) to evaluate and monitor forest diversity on a broad scale over large unexplored areas. Location The study covered an area of c. 3500 km² in the Highlands of Chiapas, southern Mexico. Methods We identified all of the tree species within 204 field plots (1000 m² each) and measured different environmental, human disturbance‐related, and spatial variables using remote sensing and GIS data. To obtain a predictive model of α tree diversity (Fisher's alpha) based on selected explanatory variables, we used a generalized linear model with a gamma error distribution. Mantel tests of matrix correspondence were used to determine whether similarities in floristic composition were correlated with similarities in the explanatory variables. Finally, we used a method that combines α and β tree diversity to define priority areas for conservation. Results The model for α tree diversity explained 44% of the overall variability, of which most was mainly related to precipitation, temperature, NDVI, and canopy (all relationships were positive, and quadratic for temperature and NDVI). There were no spatially structured regional factors that were ignored. Similarity in tree composition was correlated positively with climate and NDVI. Main conclusions The results were used to: (1) identify and assign conservation priority of unexplored areas that have high tree diversity, and (2) demonstrate the importance of several vegetation formations in the region's biodiversity. The method we present can be particularly useful in assessing regional needs and in developing local conservation strategies in poorly surveyed (and often at risk) tropical areas worldwide, where accessibility is usually limited.     

Regional scale relationships between ecosystem structure and functioning: the case of the Patagonian steppes

Aims 1. To characterize ecosystem functioning by focusing on above‐ground net primary production (ANPP), and 2. to relate the spatial heterogeneity of both functional and structural attributes of vegetation to environmental factors and landscape structure. We discuss the relationship between vegetation structure and functioning found in Patagonia in terms of the capabilities of remote sensing techniques to monitor and assess desertification. Location Western portion of the Patagonian steppes in Argentina (39°30′ S to 45°27′ S). Methods We used remotely‐sensed data from Landsat TM and AVHRR/NOAA sensors to characterize vegetation structure (physiognomic units) and ecosystem functioning (ANPP and its seasonal and interannual variation). We combined the satellite information with floristic relevés and field estimates of ANPP. We built an empirical relationship between the Landsat TM‐derived normalized difference vegetation index (NDVI) and field ANPP. Using stepwise regressions we explored the relationship between ANPP and both environmental variables (precipitation and temperature surrogates) and structural attributes of the landscape (proportion and diversity of different physiognomic classes (PCs)). Results PCs were quite heterogeneous in floristic terms, probably reflecting degradation processes. Regional estimates of ANPP showed differences of one order of magnitude among physiognomic classes. Fifty percent of the spatial variance in ANPP was accounted for by longitude, reflecting the dependency of ANPP on precipitation. The proportion of prairies and semideserts, latitude and, to a lesser extent, the number of PCs within an 8 × 8 km cell accounted for an additional 33% of the ANPP variability. ANPP spatial heterogeneity (calculated from Landsat TM data) within an 8 × 8 km cell was positively associated with the mean AVHRR/NOAA NDVI and with the diversity of physiognomic classes. Main conclusions Our results suggest that the spatial and temporal patterns of ecosystem functioning described from ANPP result not only from water availability and thermal conditions but also from landscape structure (proportion and diversity of different PCs). The structural classification performed using remotely‐sensed data captured the spatial variability in physiognomy. Such capability will allow the use of spectral classifications to monitor desertification.     

Tree biodiversity in protected and logged Bornean tropical rain forests and its measurement by satellite remote sensing

Aim Conservation activities have increasingly focused on issues at the level of the landscape but are constrained by limited data and knowledge relating to biodiversity at this scale. Satellite remote sensing has considerable, but under‐exploited, potential as a source of information on biodiversity at the landscape level. Remote sensing has generally been used to assess biodiversity indirectly, using approaches that often fail to fully exploit the information content of the imagery and typically only with regard to the species richness component of biodiversity. The aim of this paper was to assess the potential of remote sensing as a source of information on the richness, evenness and composition of tree species in a tropical rain forest. Location The test site was a c. 225 km² region centred on the Danum Valley Field Centre, Borneo. This test site contained regions of undisturbed and differentially logged rain forest. Methods Data on tree biodiversity had been acquired for fifty‐two sample plots by standard field survey methods and were used to derive summary indices of biodiversity for seedlings, saplings and mature trees. Differences between logged and unlogged sites were evaluated by comparison of the indices and species accumulation curves. A Landsat Thematic Mapper (TM) image of the site acquired close to the date of the field survey was obtained and rigorously pre‐processed. Feedforward neural networks were used to derive predictions of biodiversity indices from the imagery. A Kohonen self organizing map neural network was used to ordinate the field data to derive classes of forest defined by relative similarity in species composition. The separability of the defined classes in the Landsat TM image was evaluated with a discriminant analysis. Results Analyses of the field data revealed considerable variation in the biodiversity of seedlings, saplings and trees at the site, associated, in part, with differences in logging activities. This variation in biodiversity was manifest in the remotely sensed data. The analyses indicated an ability to (1) predict biodiversity indices, with the highest correlation between predicted and actual index observed for evenness described by Shannon entropy (r = 0.546), but especially to (2) classify nine forest classes defined on the basis of similarity in tree species composition (accuracy 95.8%). Main conclusions Logging activities impacted on biodiversity and the resulting variation in biodiversity was reflected in the remotely sensed imagery. Using methods that exploit more fully the information content of the imagery than those used in other previous studies, a richer representation of biodiversity may be derived. This representation includes estimates of key summary indices of biodiversity, notably richness and evenness, as well as information on species composition. The results indicate that remotely sensed data may be used as a source of information on biodiversity at the landscape scale that may be used to inform conservation science and management.     

Does climate determine broad‐scale patterns of species richness? A test of the causal link by natural experiment

Aim Broad‐scale spatial patterns of species richness are very strongly correlated with climatic variables. If there is a causal link, i.e. if climate directly or indirectly determines patterns of richness, then when the climatic variables change, richness should change in the manner that spatial correlations between richness and climate would predict. The present study tests this prediction using seasonal changes in climatic variables and bird richness. Location We used a grid of equal area quadrats (37 000 km²) covering North and Central America as far south as Nicaragua. Methods Summer and winter bird distribution data were drawn from monographs and field guides. Climatic data came from published sources. We also used remotely sensed NDVI (normalized difference vegetation index — a measure of greenness). Results Bird species richness changes temporally (between summer and winter) in a manner that is close to, but statistically distinguishable from, the change one would predict from models relating the spatial variation in richness at a single time to climatic variables. If one further takes into account the seasonal changes in NDVI and within‐season variability of temperature and precipitation, then winter and summer richness follow congruent, statistically indistinguishable patterns. Main conclusions Our results are consistent with the hypothesis that climatic variables (temperature and precipitation) and vegetation cover directly or indirectly influence patterns of bird species richness.     

Predicting krill swarm characteristics important for marine predators foraging off East Antarctica

Open ocean predator‐prey interactions are often difficult to interpret because of a lack of information on prey fields at scales relevant to predator behaviour. Hence, there is strong interest in identifying the biological and physical factors influencing the distribution and abundance of prey species, which may be of broad predictive use for conservation planning and evaluating effects of environmental change. This study focuses on a key Southern Ocean prey species, Antarctic krill Euphausia superba, using acoustic observations of individual swarms (aggregations) from a large‐scale survey off East Antarctica. We developed two sets of statistical models describing swarm characteristics, one set using underway survey data for the explanatory variables, and the other using their satellite remotely sensed analogues. While survey data are in situ and contemporaneous with the swarm data, remotely sensed data are all that is available for prediction and inference about prey distribution in other areas or at other times. The fitted models showed that the primary biophysical influences on krill swarm characteristics included daylight (solar elevation/radiation) and proximity to the Antarctic continental slope, but there were also complex relationships with current velocities and gradients. Overall model performance was similar regardless of whether underway or remotely sensed predictors were used. We applied the latter models to generate regional‐scale spatial predictions using a 10‐yr remotely‐sensed time series. This retrospective modelling identified areas off east Antarctica where relatively dense krill swarms were consistently predicted during austral mid‐summers, which may underpin key foraging areas for marine predators. Spatiotemporal predictions along Antarctic predator satellite tracks, from independent studies, illustrate the potential for uptake into further quantitative modelling of predator movements and foraging. The approach is widely applicable to other krill‐dependent ecosystems, and our findings are relevant to similar efforts examining biophysical linkages elsewhere in the Southern Ocean and beyond.     

A global model of the response of tropical and sub-tropical forest biodiversity to anthropogenic pressures

Habitat loss and degradation, driven largely by agricultural expansion and intensification, present the greatest immediate threat to biodiversity. Tropical forests harbour among the highest levels of terrestrial species diversity and are likely to experience rapid land-use change in the coming decades. Synthetic analyses of observed responses of species are useful for quantifying how land use affects biodiversity and for predicting outcomes under land-use scenarios. Previous applications of this approach have typically focused on individual taxonomic groups, analysing the average response of the whole community to changes in land use. Here, we incorporate quantitative remotely sensed data about habitats in, to our knowledge, the first worldwide synthetic analysis of how individual species in four major taxonomic groups—invertebrates, ‘herptiles’ (reptiles and amphibians), mammals and birds—respond to multiple human pressures in tropical and sub-tropical forests. We show significant independent impacts of land use, human vegetation offtake, forest cover and human population density on both occurrence and abundance of species, highlighting the value of analysing multiple explanatory variables simultaneously. Responses differ among the four groups considered, and—within birds and mammals—between habitat specialists and habitat generalists and between narrow-ranged and wide-ranged species.     

Tree‐mycorrhizal associations detected remotely from canopy spectral properties

A central challenge in global ecology is the identification of key functional processes in ecosystems that scale, but do not require, data for individual species across landscapes. Given that nearly all tree species form symbiotic relationships with one of two types of mycorrhizal fungi – arbuscular mycorrhizal (AM) and ectomycorrhizal (ECM) fungi – and that AM‐ and ECM‐dominated forests often have distinct nutrient economies, the detection and mapping of mycorrhizae over large areas could provide valuable insights about fundamental ecosystem processes such as nutrient cycling, species interactions, and overall forest productivity. We explored remotely sensed tree canopy spectral properties to detect underlying mycorrhizal association across a gradient of AM‐ and ECM‐dominated forest plots. Statistical mining of reflectance and reflectance derivatives across moderate/high‐resolution Landsat data revealed distinctly unique phenological signals that differentiated AM and ECM associations. This approach was trained and validated against measurements of tree species and mycorrhizal association across ~130 000 trees throughout the temperate United States. We were able to predict 77% of the variation in mycorrhizal association distribution within the forest plots (P < 0.001). The implications for this work move us toward mapping mycorrhizal association globally and advancing our understanding of biogeochemical cycling and other ecosystem processes.     

NDVI is not reliable as a surrogate of forage abundance for a large herbivore in tropical forest habitat

Remotely sensed vegetation indices are increasingly being used in wildlife studies but field‐based support for their utility as a measure of forage availability comes largely from open‐canopy habitats. We assessed whether normalized difference vegetation index (NDVI) represents forage availability for Asian elephants in a southern Indian tropical forest. We found that the number of food species was a small percentage of all plant species. NDVI was not a good measure of food abundance in any vegetation category partly because of (a) small to moderate proportional abundances of food species relative to the total abundance of all species in that category (herbs and shrubs), (b) abundant overstory vegetation resulting in low correlations between NDVI and food abundance, despite a high proportional abundance of food species and a concordance between total abundance and food species abundance (graminoids), and (c) the relevant variables measured and important as food at the ground level (count and GBH) not being related to primary productivity (trees and recruits). NDVI had a negative relationship with the total abundance of graminoids, which represent a bulk of elephant and other herbivore diet, because of negative interaction with other vegetation and canopy cover that positively explained NDVI. Spatially interpolated total graminoid abundance modeled from field data outperformed NDVI in predicting total graminoid abundance, although interpolation models of food graminoid abundance were not satisfactory. Our results reject the utility of NDVI in mapping elephant forage abundance in tropical forests, a finding that has implications for studies of other herbivores also. Abstract in Kannada is available with online material.     

Energy density and its variation in space limit species richness of boreal forest birds

Aim An area’s ability to support species may be dependent not only on the total amount of available energy it contains but also on energy density (i.e. available energy per unit area). Acknowledging these two aspects of energy availability may increase mechanistic understanding of how increased energy availability results in increased species richness. We studied the relationship between energy density, its variation in space and boreal forest bird species richness and investigated two possible mechanisms: (1) metabolic constraints of organisms, and (2) increased resource availability for specialists. Location Protected areas in Finland’s boreal forest. Methods We tested whether bird species richness was best determined by total energy availability in an area or by energy density and its variation within the area, before and after including bird abundance in the models. We evaluated two main explanatory variables: tree growth reflecting the rate of energy production and tree volume as a measure of biomass. In addition, we modelled individual species’ responses to energy density and its variation, and evaluated the prediction of the metabolic constraints hypothesis that small species are limited by energy density whereas large species are limited by total energy availability in the area. Results Energy density and its variation were good predictors of species richness: together with abundance they explained 84% of variation in species richness (compared with 74% for abundance alone). Pure metabolic constraints were unlikely to explain this relationship. Instead, the mechanism probably involved increased habitat heterogeneity benefiting specialist species. Total energy availability was also an important factor determining species richness but its effect was indirect via abundance. Main conclusions Our results corroborate the importance of energy availability as a driver of species richness in forest bird communities, and they indicate that energy density and its variation in the landscape strongly influence species richness even after accounting for abundance.     

Leaf trait variations associated with habitat affinity of tropical karst tree species

Karst hills, that is, jagged topography created by dissolution of limestone and other soluble rocks, are distributed extensively in tropical forest regions, including southern parts of China. They are characterized by a sharp mosaic of water and nutrient availability, from exposed hilltops with poor soil development to valleys with occasional flooding, to which trees show species‐specific distributions. Here we report the relationship of leaf functional traits to habitat preference of tropical karst trees. We described leaf traits of 19 tropical tree species in a seasonal karst rainforest in Guangxi Province, China, 12 species in situ and 13 ex situ in a non‐karst arboretum, which served as a common garden, with six species sampled in both. We examined how the measured leaf traits differed in relation to species’ habitat affinity and evaluated trait consistency between natural habitats vs. the arboretum. Leaf mass per area (LMA) and optical traits (light absorption and reflectance characteristics between 400 and 1,050 nm) showed significant associations with each other and habitats, with hilltop species showing high values of LMA and low values of photochemical reflectance index (PRI). For the six species sampled in both the karst forest and the arboretum, LMA, leaf dry matter content, stomatal density, and vein length per area showed inconsistent within‐species variations, whereas some traits (stomatal pore index and lamina thickness) were similar between the two sites. In conclusion, trees specialized in exposed karst hilltops with little soils are characterized by thick leaves with high tissue density indicative of conservative resources use, and this trait syndrome could potentially be sensed remotely with PRI.     

Factors affecting species richness of tree regeneration in mixed‐wood stands of central Maine

Question: Two questions about within‐stand spatial variability are addressed in this paper. How does species richness of tree regeneration respond to small‐scale ecological gradients, and what effect does natural Abies balsamea abundance have on the species richness of other tree regeneration? Location: A long‐term, gap‐silviculture experiment, Acadian mixed‐wood forest, Maine, USA. Methods: Eight stands treated with and without gap harvesting were sampled to capture sub‐stand heterogeneity of understorey tree regeneration concurrently with patterning of local stand conditions. Spatial and non‐spatial models were developed to test the relationships between two response variables [species richness of small (height ≥0.1 m, but <0.75 m) and large (height ≥0.75 m, but <1.4 m) regeneration] and five explanatory variables (depth to water table, percentage canopy transmittance, A. balsamea regeneration density, and overstorey basal area and species richness). Results: Despite high unexplained variance for all models, consistent associations among variables were found. Negative associations were found between: (1) the species richness of small regeneration and A. balsamea regeneration density and (2) the species richness of large regeneration and overstorey basal area. Positive associations were found between: (1) the species richness of small regeneration and both overstorey basal area and species richness and (2) the species richness of small and large regeneration and canopy transmittance. Conclusions: Promoting tree species diversity in Acadian mixed‐wood stands may not be achievable through the use of gap‐harvesting alone if the density of understorey Abies balsamea is not reduced either naturally or through silvicultural intervention.     

Mapping the biomass of Bornean tropical rain forest from remotely sensed data

The biomass and biomass dynamics of forests are major uncertainties in our understanding of tropical environments. Remote sensing is often the only practical means of acquiring information on forest biomass but has not always been used successfully. Here the conventional approaches to the estimation of forest biomass from remotely sensed data were evaluated relative to techniques based on the application of artificial neural networks. Together these approaches were used to estimate and map the biomass of tropical forests in north‐eastern Borneo from Landsat TM data. The neural networks were found to be particularly suited to the application. A basic multi‐layer perceptron network, for example, provided estimates of biomass that were strongly correlated with those measured in the field (r = 0.80). Moreover, these estimates were more strongly correlated with biomass than those derived from 230 conventional vegetation indices, including the widely used normalized difference vegetation index (NDVI).