Extended statistical approaches to modelling spatial pattern in biodiversity in northeast New South Wales. II. Community-level modelling

Regional conservation planning can often make more effective use of sparse biological data by linking these data to remotely mapped environmental variables through statistical modelling. While modelling distributions of individual species is the best known and most widely used approach to such modelling, there are many situations in which more information can be extracted from available data by supplementing, or replacing, species-level modelling with modelling of communities or assemblages. This paper provides an overview of approaches to community-level modelling employed in a series of major land-use planning processes in the northeast New South Wales region of Australia, and evaluates how well communities and assemblages derived using these techniques function as surrogates in regional conservation planning. We also outline three new directions that may enhance the effectiveness of community-level modelling by: (1) more closely integrating modelling with traditional ecological mapping (e.g. vegetation mapping); (2) more tightly linking numerical classification and spatial modelling through application of canonical classification techniques; and (3) enhancing the applicability of modelling to data-poor regions through employment of a new technique for modelling spatial pattern in compositional dissimilarity.     

Extended statistical approaches to modelling spatial pattern in biodiversity in northeast New South Wales. I. Species-level modelling

Statistical modelling of biological survey data in relation to remotely mapped environmental variables is a powerful technique for making more effective use of sparse data in regional conservation planning. Application of such modelling to planning in the northeast New South Wales (NSW) region of Australia represents one of the most extensive and longest running case studies of this approach anywhere in the world. Since the early 1980s, statistical modelling has been used to extrapolate distributions of over 2300 species of plants and animals, and a wide variety of higher-level communities and assemblages. These modelled distributions have played a pivotal role in a series of major land-use planning processes, culminating in extensive additions to the region's protected area system. This paper provides an overview of the analytical methodology used to model distributions of individual species in northeast NSW, including approaches to: (1) developing a basic integrated statistical and geographical information system (GIS) framework to facilitate automated fitting and extrapolation of species models; (2) extending this basic approach to incorporate consideration of spatial autocorrelation, land-cover mapping and expert knowledge; and (3) evaluating the performance of species modelling, both in terms of predictive accuracy and in terms of the effectiveness with which such models function as general surrogates for biodiversity.     

Regional application of an ecosystem production model for studies of biogeochemistry in Brazilian Amazonia

The degree to which primary production, soil carbon, and trace gas fluxes in tropical forests of the Amazon are limited by moisture availability and other environmental factors was examined using an ecosystem modelling application for the country of Brazil. A regional geographical information system (GIS) serves as the data source of climate drivers, satellite images, land cover, and soil properties for input to the NASA Ames-CASA (Carnegie-Ames-Stanford Approach) model over a 8-km grid resolution. Simulation results lead us to hypothesize that net primary production (NPP) is limited by cloud interception of solar radiation over the humid north-western portion of the region. Peak annual rates for NPP of nearly 1.4 kg C m^–2 year^–1 are localized in the seasonally dry eastern Amazon in areas that we assume are primarily deep-rooted evergreen forest cover. Regional effects of forest conversion on NPP and soil carbon content are indicated in the model results, especially in seasonally dry areas. Comparison of model flux predictions along selected eco-climatic transects reveal moisture, soil, and land use controls on gradients of ecosystem production and soil trace gas emissions (CO₂, N₂O, and NO). These results are used to formulate a series of research hypotheses for testing in the next phase of regional modelling, which includes recalibration of the light-use efficiency term in NASA-CASA using field measurements of NPP, and refinements of vegetation index and soil property (texture and potential rooting depth) maps for the region.     

Using generalized autoregressive error models to understand fire–vegetation–soil feedbacks in a mulga–spinifex landscape mosaic

Aim To develop a new modelling approach for spatially autocorrelated non‐normal data, and apply it to a case study of the role that fire–vegetation–soil feedbacks play in maintaining boundaries between fire‐sensitive and fire‐promoted plant communities. Location A mulga (Acacia aneura) shrubland–spinifex (Triodia spp.) grassland mosaic, central Australia. Methods Autoregressive error models were extended to non‐normal data by incorporating neighbourhood values of the response and predictor variables into generalized nonlinear models. These models were used to examine the environmental correlates of three response variables: mulga cover; fire frequency in areas free of mulga; and the presence of mulga banding. Mulga cover and mulga banding were assessed visually by overlaying 4477 × 1 km² grid cells on both Landsat 7 ETM+ and very high resolution imagery. Fire frequency was estimated from an existing fire history for central Australia, based on remotely sensed fire scars. Results The autoregressive error models explained 27%, 47% and 57% of the null deviance of mulga cover, fire frequency and mulga banding, respectively, with 12%, 15% and 24% of the null deviance being explained by environmental variables alone. These models accounted for virtually all residual spatial autocorrelation. While there was a clear negative relationship between mulga cover and fire frequency, there was little evidence that mulga was being restricted to parts of the landscape with inherently low fire frequencies. Mulga was most abundant at very low slope angles and on red earths, both of which are likely to reflect high site productivity, while fire frequency was not clearly affected by slope angle and was also relatively high on red earths. Main conclusions The modelling approach we have developed provides a much needed way of analysing spatially autocorrelated non‐normal data and can be easily incorporated into an information‐theoretic modelling framework. Using this approach, we provide evidence that mulga and spinifex have a highly antagonistic relationship. In more productive parts of the landscape, mulga suppresses spinifex and fire, while in less productive parts of the landscape, fire and spinifex suppress mulga, leading to the remarkable abruptness of mulga–spinifex boundaries that are maintained via fire–vegetation–soil feedbacks.     

Mapping riparian condition indicators in a sub-tropical savanna environment from discrete return LiDAR data using object-based image analysis

Mapping, monitoring and managing the environmental condition of riparian zones are major focus areas for local and state governments in Australia. New remotely sensed data techniques that can provide the required mapping accuracies, complete spatial coverage and processing and mapping transferability are currently being developed for use over large spatial extents. The research objective was to develop and apply an approach for mapping riparian condition indicators using object-based image analysis of airborne Light Detection and Ranging (LiDAR) data. The indicators assessed were: streambed width; riparian zone width; plant projective cover (PPC); longitudinal continuity; coverage of large trees; vegetation overhang; and stream bank stability. LiDAR data were captured on 15 July 2007 for two 5 km stretches along Mimosa Creek in Central Queensland, Australia. Field measurements of riparian vegetation structural and landform parameters were obtained between 28 May and 5 June 2007. Object-based approaches were developed for mapping each riparian condition indicator from the LiDAR data. The validation and empirical modelling results showed that the object-based approach could be used to accurately map the riparian condition indicators (R² = 0.99 for streambed width, R² = 0.82 for riparian zone width, R² = 0.89 for PPC, R² = 0.40 for bank stability). These research findings will be used in a 26,000 km mapping project assessing riparian vegetation and physical form indicators from LiDAR data in Victoria, Australia.     

Vegetation maps based on remote sensing are informative predictors of habitat selection of grassland birds across a wetness gradient

Vegetation is a major environmental factor influencing habitat selection in bird species. High resolution mapping of vegetation cover is essential to model the distribution of populations and improve the management of breeding habitats. However, the task is challenging for grassland birds because microhabitat variations relevant at the territory scale cannot be measured continuously over large areas to delineate areas of higher suitability. Remote sensing may help to circumvent this problem. We addressed this issue by using SPOT 5 imagery and phytosociological data. We mapped grassland vegetation in a floodplain using two methods. We (i) mapped the continuous Ellenberg index of moisture and (ii) identified 5 vegetation classes distributed across the wetness gradient. These two methods produced consistent output maps, but they also provided complementary results. Ellenberg index is a valuable proxy for soil moisture while the class approach provided more information about vegetation structure, and possibly trophic resources. In spite of the apparent uniformity of meadows, our data show that birds do not settle randomly along the moisture and vegetation gradients. Overall birds tend to avoid the driest vegetation classes, i.e. the highest grounds. Thus, vegetation maps based on remote sensing could be valuable tools to study habitat selection and niche partition in grassland bird communities. It is also a valuable tool for conservation and habitat management.     

Richness of species and growth-forms within sclerophyll and mesophyll vegetation in eastern Australia

Abstract The patterns in total species richness and in the richness of the dominant growth-forms of vegetation communities of coastal sclerophyll and mesophyll vegetation in eastern Australia are examined. Plant species richness data were obtained from two 500 m² quadrats from 50 sites within a single geographical region north of Sydney, New South Wales. Concentrically nested subquadrats within each quadrat enabled the determination of species-area relationships for total species richness and its components. Three growth-forms were examined (trees, shrubs and ground cover) and patterns in the richness of these components were compared to those exhibited by total species richness. Total species richness was higher in sclerophyll communities on Hawkesbury Sandstone soils than in adjacent mesophyll communities on Narrabeen shales and sandstones. Significant patterns in total species richness within the two soil types were also found. Shrub and ground cover species richness showed strong correlations with total species richness, with higher richness in the sclerophyll communities. However, tree species richness contributed little to the patterns in total species richness. The results of this study suggest that differential patterns in the components of total species richness must be taken into account for effective modelling of natural areas based on species richness and diversity parameters.     

Remote sensing imagery in vegetation mapping: a review

Aims Mapping vegetation through remotely sensed images involves various considerations, processes and techniques. Increasing availability of remotely sensed images due to the rapid advancement of remote sensing technology expands the horizon of our choices of imagery sources. Various sources of imagery are known for their differences in spectral, spatial, radioactive and temporal characteristics and thus are suitable for different purposes of vegetation mapping. Generally, it needs to develop a vegetation classification at first for classifying and mapping vegetation cover from remote sensed images either at a community level or species level. Then, correlations of the vegetation types (communities or species) within this classification system with discernible spectral characteristics of remote sensed imagery have to be identified. These spectral classes of the imagery are finally translated into the vegetation types in the image interpretation process, which is also called image processing. This paper presents an overview of how to use remote sensing imagery to classify and map vegetation cover.Methods Specifically, this paper focuses on the comparisons of popular remote sensing sensors, commonly adopted image processing methods and prevailing classification accuracy assessments.Important findings The basic concepts, available imagery sources and classification techniques of remote sensing imagery related to vegetation mapping were introduced, analyzed and compared. The advantages and limitations of using remote sensing imagery for vegetation cover mapping were provided to iterate the importance of thorough understanding of the related concepts and careful design of the technical procedures, which can be utilized to study vegetation cover from remote sensed images.     

Remote sensing of upland vegetation: the potential of high spatial resolution satellite sensors

Aim Traditional methodologies of mapping vegetation, as carried out by ecologists, consist primarily of field surveying or mapping from aerial photography. Previous applications of satellite imagery for this task (e.g. Landsat TM and SPOT HRV) have been unsuccessful, as such imagery proved to have insufficient spatial resolution for mapping vegetation. This paper reports on a study to assess the capabilities of the recently launched remote sensing satellite sensor Ikonos, with improved capabilities, for mapping and monitoring upland vegetation using traditional image classification methods. Location The location is Northumberland National Park, UK. Methods Traditional remote sensing classification methodologies were applied to the Ikonos data and the outputs compared to ground data sets. This enabled an assessment of the value of the improved spatial resolution of satellite imagery for mapping upland vegetation. Post‐classification methods were applied to remove noise and misclassified pixels and to create maps that were more in keeping with the information requirements of the NNPA for current management processes. Results The approach adopted herein for quick and inexpensive land cover mapping was found to be capable of higher accuracy than achieved with previous approaches, highlighting the benefits of remote sensing for providing land cover maps. Main conclusions Ikonos imagery proved to be a useful tool for mapping upland vegetation across large areas and at fine spatial resolution, providing accuracies comparable to traditional mapping methods of ground surveys and aerial photography.     

Estimating the influence of land management change on weed invasion potential using expert knowledge

Aim To develop and test a general framework for estimating weed invasion potential (suitability and susceptibility) that utilized expert knowledge of dispersal, establishment and persistence and considered the influence of land management. Location The semi‐arid Desert Channels Region of Queensland, Australia (476,000 km²). Methods We developed a general framework that integrated knowledge and empirical data of the environmental and land management variables influencing the dispersal, establishment and persistence of the invasive shrub parkinsonia (Parkinsonia aculeata) using a Bayesian network linked to a Geographic Information System (GIS). We evaluated the influence of different land management scenarios on landscape suitability for parkinsonia. Model performance was assessed by comparing predicted landscape suitability with mapped parkinsonia locations and estimated parkinsonia density. Results Our predictions of moderate to high suitability corresponded reasonably well with mapped parkinsonia locations (71% match) and areas of common to abundant estimated density (92% match). They also suggested that parkinsonia has not reached its potential distribution within the study region. Under current land management conditions, 77,000 km² of land was found to be highly or moderately suitable for parkinsonia. Scenario analysis indicated that maintaining moderate herbaceous ground cover levels, and using sheep to browse juvenile parkinsonia, reduced the predicted moderate to high suitability area to 27,000 km², offering a potential management strategy for limiting parkinsonia invasion. Main conclusions Weed invasion potential can be reasonably estimated using expert knowledge of dispersal, establishment and persistence, integrated using a Bayesian network linked to a GIS. This modelling approach can be an alternative to process‐based and phenomenological modelling, which can be problematic for modelling new and emerging weed invasions, particularly where data are patchy. The modelling approach also allows the influence of land management change on invasion potential to be investigated through scenario analysis.     

Mapping vegetation in an arid,mountainous region of Western Australia

Abstract. Predictive mapping of vegetation using models linking vegetation units to mapped environmental variables has been advocated for remote areas. In this study, three different types of model were employed (within a GIS) to produce vegetation maps of the Hamersley Ranges region of Western Australia. The models were: (1) decision trees; (2) statistical models; and (3) heuristic/conceptual models. Maps were produced for three different levels of a floristic classification, i.e. 16 communities in two community groups with eight subgroups. All models satisfactorily established relationships between the vegetation units and available predictor variables, except where the number of sites of a particular unit was small. The different models often made similar predictions, especially for more widespread vegetation units. Map accuracy (as determined by field testing of maps) improved with increasing level of abstraction, with plant community maps ca. 50 % correct, subgroup maps ca. 60 % correct and group maps 90 % correct. Map inaccuracies were due to several factors, including low sample numbers producing unrepresentative models, poor resolution of and errors in available maps of predictor variables, and available predictor variables not being able to differentiate between certain vegetation units, particularly at the plant community level. Of these factors, poor resolution of maps was seen as the most critical. One type of model could not be recommended over another; however the choice of model will be largely dependent on the nature of the data set and the type of map coverage required.     

Vegetation characteristics influence fine-scale intensity of habitat use by wild turkey and white-tailed deer in a loblolly pine plantation

Habitat quality is often evaluated based on food availability. However, ecological theory suggests cover should be a more important decision rule when food is not a proximate threat to fitness, as cover mediates predation risk as well as other important factors of fitness. In reality, vegetation characteristics related to food availability and cover are rarely coupled with animal use in the same space and time to determine their relative influences on habitat use. Using an array of 81 camera traps in a matrix of forest management strategies used to deliberately cause a wide disparity in vegetation characteristics, we monitored intensity of use by white-tailed deer (Odocoileus virginianus) and wild turkey (Meleagris gallopavo). We measured vegetation characteristics related to food and cover at each camera trap location then used a generalized additive model to determine how vegetation characteristics specific to the location affected intensity of habitat use by animals at the location. Consistent among both species, cover best explained intensity of habitat use. Contrastingly, food did not explain intensity of habitat use well for either species. Some vegetation simultaneously provides cover and food, and our data indicate that areas with vegetation characteristics providing both resources had the greatest influence on intensity of habitat use by both species. Our results suggest deer and turkey may perceive cover as a more important habitat component when food is not a proximate fitness threat.     

Measuring, mapping and modelling: an integrated approach to the management of mangrove and saltmarsh in the Minnamurra River estuary, southeast Australia

Mangrove and saltmarsh ecosystems appear particularly vulnerable to the impacts of climate change, and their effective management will require forecasts of how these wetland habitats are likely to respond to sea-level rise through the twenty-first century. We describe a preliminary study of a small stand of mangrove and saltmarsh that involves measuring of elevation change and accretion, mapping of wetland communities, and modelling of their potential response to sea-level rise. The wetland occurs on the banks of the Minnamurra River estuary in southern New South Wales and has been the focus of several studies over recent decades. The research includes empirical measurements of sedimentation at sites in both mangrove and saltmarsh vegetation using the surface elevation table-marker horizon technique. This is a site at which mapping has been undertaken to delineate the extent of each vegetation community from a time-series of aerial photographs using geographical information systems; the gradual incursion of the mangrove, Avicennia marina, into more landward saltmarsh communities, observed over past decades for many systems in southeastern Australia, has continued into the twenty-first century. The observed patterns of change are compared with simulations of how this wetland system might respond to future sea-level rise, adopting several different approaches and the upper and lower bounds of Intergovernmental Panel on Climate Change sea-level rise projections. The model results show considerable variability in response depending on the parameters adopted. We advocate the need for the integration of these three approaches, measuring, mapping and modelling, as a basis for future management and adaptation. Our study demonstrates the considerable opportunities to refine the data input and model outputs as part of adaptive management, as more sophisticated technologies and data become available.     

Traits reveal how habitat‐quality gradients structure small mammal communities in a fragmented tropical landscape

Understanding the relationships among community structure, vegetation structure and availability of food resources are a key to unravelling the ecological processes that structure biological communities. In this study, we tested (i) whether the composition of small mammal communities changed across gradients in habitat quality in tropical forest fragments, and (ii) whether any observed change could be explained by the functional traits of species. We sampled 24 trapping grids in fragments of semi‐deciduous forest, in each of two 6‐month periods. We considered each trapping grid as a sampling unit, for which we collected three datasets: an environmental matrix (vegetation structure and food resource availability), the abundance of small mammal species (community structure) and a matrix of functional traits (ecological and morphological traits which express tolerance to habitat disturbance and trophic guild). We used an RLQ approach to evaluate the association between traits and environmental gradients. Forest‐specialist and scansorial–arboreal species were associated with more complex habitat that had greater litter and canopy cover and more fallen logs. In relation to trophic guilds, granivore (fruit seeds), insectivorous and omnivorous species were also associated with higher complexity habitat, while frugivores were associated with shrub cover and availability of fruits. We conclude that functional traits (habitat use, use of vertical strata and diet) provide valuable insights into the distribution of small mammals along gradients of habitat quality in tropical forest fragments. We highlight that communities studies in fragmented landscapes should investigate the different components of biodiversity not only in landscape‐scale but also in habitat scale. Abstract in Portuguese is available with online material.     

Forecasting insect species richness scores in poorly surveyed territories: the case of the Portuguese dung beetles (Col. Scarabaeinae)

Large-scale biodiversity assessment of faunal distribution is needed in poorly sampled areas. In this paper, Scarabaeinae dung beetle species richness in Portugal is forecasted from a model built with a data set from areas identified as well sampled. Generalized linear models are used to relate the number of Scarabaeinae species in each Portuguese UTM 50 × 50 grid square with a set of 25 predictor variables (geographic, topographic, climatic and land cover) extracted from a geographic information system (GIS). Between-squares sampling effort unevenness, spatial autocorrelation of environmental data, non-linear relationships between variables and an assessment of the models' predictive power, the main shortcomings in geographic species richness modelling, are addressed. This methodological approach has proved to be reliable and accurate enough in estimating species richness distribution, thus providing a tool to identify areas as potential targets for conservation policies in poorly inventoried countries.     

Step‐less models for regional environmental variation in Norway

Aim To assess the scale of variation for major environmental gradients in Norway. To obtain a step‐less model for this variation and to use this model to evaluate the extent to which the consensus expert classification of Norway into vegetation regions can be predicted from environmental variables. To discuss the potential of step‐less models for understanding natural variation at regional scales, for stratification and for predictive modelling of species distributions and land‐cover types. Location The mainland of Norway. Methods Fifty‐four climatic, topographical, hydrological and geological variables were recorded for grid cells with spatial resolution (grain size) of 1 × 1, 5 × 5 and 10 × 10 km, spanning the entire mainland of Norway. Principal components analyses (PCA) were used to summarize variation in three primary data matrices and three random subsets of these. Results The first four principal components explained between 75% and 85% of the variation in the data sets. All PCAs revealed four consistent environmental gradients, in order of decreasing importance: (1) regional variation (gradient) from coast to inland and from oceanic/humid to continental areas; (2) regional variation from north to south and from high to low altitudes; (3) regional variation from north to south and from inland to coast, related to solar radiation; and (4) topographic (terrain relief) variation on finer scales than (1–3). The first two PCA axes corresponded to the two bioclimatic gradients used in expert classifications of Norway into biogeographical regions: vegetation sections (from highly oceanic to slightly continental) and vegetation zones (from nemoral to alpine). Main conclusions Our PCA analyses substantiate the current view of bioclimatic regional vegetation variation in Norway, provide an explicit characterization of this variation in terms of climatic variables, and show that environmental variability can be reproduced as GIS layers in step‐less models. These models have the potential to become important tools for future predictive modelling within resource management, conservation planning and biogeographical (and other ecological) research.     

When the method for mapping species matters: defining priority areas for conservation of African freshwater turtles

Aim Defining priority areas for conservation is essential to minimize biodiversity loss, but the adoption of different methods for describing species distributions influences the outcomes. In order to provide a robust basis for the conservation of freshwater turtles in Africa, we compared the effect that different species‐mapping approaches had on derived patterns of species richness, species vulnerability and protected‐area representativeness. Location Africa. Methods We adopted three different approaches with increasing complexity for generating species distribution maps. The first approach was based on the geographic intersection of species records and grid squares; the second on the union of local convex polygons; and the third on inductive distribution modelling techniques. We used distribution maps, generated using these three approaches, to determine conservation priorities based on geographic patterns of species richness and vulnerability, as well as for conducting gap and irreplaceability analyses. Results We obtained markedly different distribution maps using the three methods, which in turn caused differences in conservation priorities. The grid‐square approach underestimated range sizes and species richness, while the polygon approach overestimated these attributes. The distribution modelling approach provided the most realistic outcome in terms of diversity patterns, by minimizing both commission and omission errors. An integrated map of conservation priority – derived by combining individual measures of priority based on the distribution modelling approach – identified the Gulf of Guinea coast and the Albertine Rift as major priority areas. Main conclusions Each species‐mapping approach has both advantages and disadvantages. The choice of the most appropriate approach in any given situation depends on the availability of locality records and on the relative importance of mitigating omission and commission errors. Our findings suggest that in most circumstances, the use of distribution modelling has many advantages relative to the other approaches. The priority areas identified in this study should be considered for targeting efforts to conserve Africa freshwater turtles in the coming years.     

Soil and vegetation responses to hydrological manipulation in a partially drained polje fen in New Zealand

Anthropogenic drainage causes loss of natural character in herbaceous wetlands due to increased soil oxygen penetration. We related vegetation gradients in a New Zealand polje fen to long-term effects of drains by using hydrological, edaphic and vegetation data, and a before-after-control-impact (BACI) design to test responses to experimental drain closure. Soil profiles and continuous water level records revealed a site subject to frequent disturbance by intense but brief floods, followed by long drying periods during which areas close to drains experienced lower water tables and more variable water levels. Classification of vegetation data identified 12 groups along a moisture gradient, from dry areas dominated by pastoral alien species, to wet communities dominated by native wetland sedges. Lower total species diversity and native representation in pastoral communities were related to the high proportion of alien competitor and competitor-disturbance species, compared with the stress tolerator-dominated flora of other groups. Species–environment relationships revealed highly significant correlations with soil water content and aeration as measured by redox potential (E_H) and steel rod oxidation depth, as well as soil nutrient content and bulk density. Comparison of root anatomy confirmed greater development of flood-tolerant traits in native species than in pastoral aliens, and vegetation N:P ratios indicated that most communities were probably nitrogen-limited. Flooding rapidly re-established wetland hydrology in dried sites in the impact area, and lowered E_H and soil oxidation depth, but had no effect on N and P availability. Presence and cover of pastoral alien species decreased in these areas. This study supports the use of hydrological manipulation as a tool for reducing soil oxidation and thus the impact of alien plant species at restoration sites with minimal intervention, but suggests the need for detailed information on species flooding tolerances and hydrological preferences to underpin this approach.