A condition metric for Eucalyptus woodland derived from expert evaluations.

The evaluation of ecosystem quality is important for land-management and land-use planning. Evaluation is unavoidably subjective, and robust metrics must be based on consensus and the structured use of observations. We devised a transparent and repeatable process for building and testing ecosystem metrics based on expert data. We gathered quantitative evaluation data on the quality of hypothetical grassy woodland sites from experts. We used these data to train a model (an ensemble of 30 bagged regression trees) capable of predicting the perceived quality of similar hypothetical woodlands based on a set of 13 site variables as inputs (e.g., cover of shrubs, richness of native forbs). These variables can be measured at any site and the model implemented in a spreadsheet as a metric of woodland quality. We also investigated the number of experts required to produce an opinion data set sufficient for the construction of a metric. The model produced evaluations similar to those provided by experts, as shown by assessing the model's quality scores of expert-evaluated test sites not used to train the model. We applied the metric to 13 woodland conservation reserves and asked managers of these sites to independently evaluate their quality. To assess metric performance, we compared the model's evaluation of site quality with the managers' evaluations through multidimensional scaling. The metric performed relatively well, plotting close to the center of the space defined by the evaluators. Given the method provides data-driven consensus and repeatability, which no single human evaluator can provide, we suggest it is a valuable tool for evaluating ecosystem quality in real-world contexts. We believe our approach is applicable to any ecosystem.

Historical Maps from Modern Images: Using Remote Sensing to Model and Map Century-Long Vegetation Change in a Fire-Prone Region.

Understanding the age structure of vegetation is important for effective land management, especially in fire-prone landscapes where the effects of fire can persist for decades and centuries. In many parts of the world, such information is limited due to an inability to map disturbance histories before the availability of satellite images (~1972). Here, we describe a method for creating a spatial model of the age structure of canopy species that established pre-1972. We built predictive neural network models based on remotely sensed data and ecological field survey data. These models determined the relationship between sites of known fire age and remotely sensed data. The predictive model was applied across a 104,000 km(2) study region in semi-arid Australia to create a spatial model of vegetation age structure, which is primarily the result of stand-replacing fires which occurred before 1972. An assessment of the predictive capacity of the model using independent validation data showed a significant correlation (rs = 0.64) between predicted and known age at test sites. Application of the model provides valuable insights into the distribution of vegetation age-classes and fire history in the study region. This is a relatively straightforward method which uses widely available data sources that can be applied in other regions to predict age-class distribution beyond the limits imposed by satellite imagery.

Quantifying ecosystem quality by modeling multi-attribute expert opinion.

The evaluation of ecosystem quality is inherently subjective, requiring decisions about which variables to notice or measure, and how these variables are integrated into a coherent evaluation. Despite the central role of human judgment, few evaluation methods address the subjectivity that is inherent in their design. There are, however, advantages to directly using opinion to create an expert system where the metric is constructed around opinion data. These advantages include stakeholder inclusion and the encouragement of a dialogue of data-driven criticism rather than subjective counter-opinion. We create an expert system to express the quality of a grassland ecosystem in Australia. We use an ensemble of bagged regression trees trained on calibrated expert preference data, to model the perceived quality of this grassland using a set of eight site variables as inputs. The model provides useful predictions of grassland quality, producing predictions similar to real expert evaluations of independent synthetic test sites not used to train the model. We apply the model to real grassland sites ranging from pristine to highly degraded, and confirm that our model orders the sites according to their degree of modification. We demonstrate that the use of too few experts produces relatively poor results, and show that for our problem the use of data from over twenty experts is appropriate. The scaling approach we used to calibrate between-expert data is shown to be an appropriate mechanism for aggregating the opinions of multiple experts. The resultant model will be useful in many contexts, and can be used by managers as a tool to evaluate real sites. It can also be integrated into ecological models of change as a means of evaluating predicted changes, for example, as a measure of utility when combined with cost estimates. The basic approach demonstrated here is applicable to any ecosystem, and we discuss the opportunities and limitations of its wider use.

Using presence-only and presence-absence data to estimate the current and potential distributions of established invasive species.

1.Predicting the current and potential distributions of established invasive species is critical for evaluating management options, but methods for differentiating these distributions have received little attention. In particular, there is uncertainty among invasive species managers about the value of information from incidental sightings compared to data from designed field surveys. This study compares the two approaches, and develops a unifying framework, using the case of invasive sambar deer Cervus unicolor in Victoria, Australia.2.We first used 391 incidental sightings of sambar deer and 12 biophysical variables to construct a presence-only habitat suitability model using Maxent. We then used that model to stratify field sampling, with proportionately greater sampling of cells with high predicted habitat suitability. Field sampling, consisting of faecal pellet surveys, sign surveys and camera trapping, was conducted in 80 4-km(2) grid cells. A Bayesian state-space occupancy model was used to predict probability of suitable habitat from the field data.3.The Maxent and occupancy models predicted similar spatial distributions of habitat suitability for sambar deer in Victoria and there was a strong positive correlation between the rankings of cells by the two approaches. The congruence of the two models suggests that any spatial and detection biases in the presence-only data were relatively unimportant in our study.4.We predicted the extent of suitable habitat from the occupancy model using a threshold that gave a false negative error rate of 0·05. The current distribution was the suitable habitat within a kernel that had a 99·5% chance of including the presence locations pooled from incidental sightings and field surveys: the potential distribution was suitable habitat outside that kernel. Several discrete areas of potential distribution were identified as priorities for surveillance monitoring with the aim of detecting and managing incursions of sambar deer.5.Synthesis and applications.Our framework enables managers to robustly estimate the current and potential distributions of established invasive species using either presence-only and/or presence-absence data. Managers can then focus control and/or containment actions within the current distribution and establish surveillance monitoring to detect incursions within the potential distribution.