Effects of sample survey design on the accuracy of classification tree models in species distribution models

We evaluated the effects of probabilistic (hereafter DESIGN) and non-probabilistic (PURPOSIVE) sample surveys on resultant classification tree models for predicting the presence of four lichen species in the Pacific Northwest, USA. Models derived from both survey forms were assessed using an independent data set (EVALUATION). Measures of accuracy as gauged by resubstitution rates were similar for each lichen species irrespective of the underlying sample survey form. Cross-validation estimates of prediction accuracies were lower than resubstitution accuracies for all species and both design types, and in all cases were closer to the true prediction accuracies based on the EVALUATION data set. We argue that greater emphasis should be placed on calculating and reporting cross-validation accuracy rates rather than simple resubstitution accuracy rates. Evaluation of the DESIGN and PURPOSIVE tree models on the EVALUATION data set shows significantly lower prediction accuracy for the PURPOSIVE tree models relative to the DESIGN models, indicating that non-probabilistic sample surveys may generate models with limited predictive capability. These differences were consistent across all four lichen species, with 11 of the 12 possible species and sample survey type comparisons having significantly lower accuracy rates. Some differences in accuracy were as large as 50%. The classification tree structures also differed considerably both among and within the modelled species, depending on the sample survey form. Overlap in the predictor variables selected by the DESIGN and PURPOSIVE tree models ranged from only 20% to 38%, indicating the classification trees fit the two evaluated survey forms on different sets of predictor variables. The magnitude of these differences in predictor variables throws doubt on ecological interpretation derived from prediction models based on non-probabilistic sample surveys.     

Detecting Insect Pollinator Declines on Regional and Global Scales

Recently there has been considerable concern about declines in bee communities in agricultural and natural habitats. The value of pollination to agriculture, provided primarily by bees, is >$200 billion/year worldwide, and in natural ecosystems it is thought to be even greater. However, no monitoring program exists to accurately detect declines in abundance of insect pollinators; thus, it is difficult to quantify the status of bee communities or estimate the extent of declines. We used data from 11 multiyear studies of bee communities to devise a program to monitor pollinators at regional, national, or international scales. In these studies, 7 different methods for sampling bees were used and bees were sampled on 3 different continents. We estimated that a monitoring program with 200–250 sampling locations each sampled twice over 5 years would provide sufficient power to detect small (2–5%) annual declines in the number of species and in total abundance and would cost U.S.$2,000,000. To detect declines as small as 1% annually over the same period would require >300 sampling locations. Given the role of pollinators in food security and ecosystem function, we recommend establishment of integrated regional and international monitoring programs to detect changes in pollinator communities. Detección de Declinaciones de Insectos Polinizadores a Escalas Regional y Global     

Conservation of Insect Diversity: a Habitat Approach

Abstract: Neither time nor resources exist to design conservation plans for every species, particularly for little-studied, noncharismatic, but ecologically important taxa that make up most of biodiversity. To explore the feasibility of basing conservation action on community-level biogeography, we sampled a montane insect community. We addressed three issues: (1) the appropriate scale for sampling insect communities; (2) the association of habitat specialization—perhaps a measure of extinction vulnerability—with other ecological or physical traits; and (3) the correlation of diversity across major insect groups. Using malaise traps in Gunnison County, Colorado, we captured 8847 Diptera (identified to family and morphospecies), 1822 Hymenoptera (identified to morphospecies), and 2107 other insects (identified to order). We sampled in three habitat types—meadow, aspen, and conifer—defined on the basis of the dominant vegetation at the scale of hundreds of meters. Dipteran communities were clearly differentiated by habitat type rather than geographic proximity. This result also holds true for hymenopteran communities. Body size and feeding habits were associated with habitat specialization at the family level. In particular, habitat generalists at the family level—taxa perhaps more likely to survive anthropogenic habitat alteration—tended to be trophic generalists. Dipteran species richness was marginally correlated with hymenopteran species richness and was significantly correlated with the total number of insect orders sampled by site. Because these correlations result from differences in richness among habitat types, insect taxa may be reasonable surrogates for one another when sampling is done across habitat types. In sum, community-wide studies appear to offer a practical way to gather information about the diversity and distribution of little-known taxa.     

Reconnecting to the Biosphere

Humanity has emerged as a major force in the operation of the biosphere, with a significant imprint on the Earth System, challenging social-ecological resilience. This new situation calls for a fundamental shift in perspectives, world views, and institutions. Human development and progress must be reconnected to the capacity of the biosphere and essential ecosystem services to be sustained. Governance challenges include a highly interconnected and faster world, cascading social-ecological interactions and planetary boundaries that create vulnerabilities but also opportunities for social-ecological change and transformation. Tipping points and thresholds highlight the importance of understanding and managing resilience. New modes of flexible governance are emerging. A central challenge is to reconnect these efforts to the changing preconditions for societal development as active stewards of the Earth System. We suggest that the Millennium Development Goals need to be reframed in such a planetary stewardship context combined with a call for a new social contract on global sustainability. The ongoing mind shift in human relations with Earth and its boundaries provides exciting opportunities for societal development in collaboration with the biosphere--a global sustainability agenda for humanity.     

Evaluating Kriging as a Tool to Improve Moderate Resolution Maps of Forest Biomass

The USDA Forest Service, Forest Inventory and Analysis program (FIA) recently produced a nationwide map of forest biomass by modeling biomass collected on forest inventory plots as nonparametric functions of moderate resolution satellite data and other environmental variables using Cubist software. Efforts are underway to develop methods to enhance this initial map. We explored the possibility of modeling spatial structure to make such improvements. Spatial structure in the field biomass data as well as in residuals from the map was investigated across 18 ecological zones in the Interior Western U.S. Exploratory tools included directional graphs of summary statistics, three dimensional maps, Moran’s I correlograms, and variograms. Where spatial pattern was present, field and residual biomass were kriged, and predictions made for an independent test set were evaluated for improvement over predictions in the initial biomass map. While kriging has some potential benefit when analyzing the field data and exploring spatial structure, kriging residuals resulted in little or no improvement in the initial biomass map developed using Cubist software. Stationarity assumptions, variogram behavior, and appropriate model fitting strategies are discussed.     

Ambient ozone injury to forest plants in Northeast and North Central USA: 16?years of biomonitoring

The US Forest Service administers a long-term, nationwide ozone biomonitoring program in partnership with other state and federal agencies to address national concerns about ozone impacts on forest health. Biomonitoring surveys begun in 1994 in the East and 1998 in the West provide important regional information on ozone air quality and a field-based record of ozone injury unavailable from any other data source. Surveys in the Northeast and North Central subregions cover 450 field sites in 24 states where ozone-sensitive plants are evaluated for ozone-induced foliar injury every year. Sites are typically large, undisturbed openings (>3 acres in size) close to forested areas where >3 bioindicator species are available for evaluation. Over the 16-year sampling period, injury indices have fluctuated annually in response to seasonal ozone concentrations and site moisture conditions. Sites with and without injury occur at all ozone exposures but when ambient concentrations are relatively low, the percentage of uninjured sites is much greater than the percentage of injured sites; and regardless of ozone exposure, when drought conditions prevail, the percentage of uninjured sites is much greater than the percentage of injured sites. Results indicate a declining trend in foliar injury especially after 2002 when peak ozone concentrations declined across the entire region.     

Functional outcomes of motor vehicle crash thoracic injuries in pediatric and adult occupants

Objective: Characterization of the severity of injury should account for both mortality and disability. The objective of this study was to develop a disability metric for thoracic injuries in motor vehicle crashes (MVCs) and compare the functional outcomes between the pediatric and adult populations.

Methods: Disability risk (DR) was quantified using Functional Independence Measure (FIM) scores within the National Trauma Data Bank for the most frequently occurring Abbreviated Injury Scale (AIS) 2–5 thoracic injuries. Occupants with thoracic injury were classified as disabled or not disabled based on the FIM scale, and comparisons were made between the following age groups: pediatric, adult, middle-aged, and older occupants (ages 7–18, 19–45, 46–65, and 66+, respectively). For each age group, DR was calculated by dividing the number of patients who were disabled and sustained a given injury by the number of patients who sustained a given injury. To account for the effect of higher severity co-injuries, a maximum AIS adjusted DR (DR_MAIS) was also calculated for each injury. DR and DR_MAIS could range from 0 to 100% disability risk.

Results: The mean DR_MAIS for MVC thoracic injuries was 20% for pediatric occupants, 22% for adults, 29% for middle-aged adults, and 43% for older adults. Older adults possessed higher DR_MAIS values for diaphragm laceration/rupture, heart laceration, hemo/pneumothorax, lung contusion/laceration, and rib and sternum fracture compared to the other age groups. The pediatric population possessed a higher DR_MAIS value for flail chest compared to the other age groups.

Conclusion: Older adults had significantly greater overall disability than each of the other age groups for thoracic injuries. The developed disability metrics are important in quantifying the significant burden of injuries and loss of quality life years. Such metrics can be used to better characterize severity of injury and further the understanding of age-related differences in injury outcomes, which can influence future age-specific modifications to AIS.     

Bee community shifts with landscape context in a tropical countryside.

The ongoing scientific controversy over a putative "global pollination crisis" underscores the lack of understanding of the response of bees (the most important taxon of pollinators) to ongoing global land-use changes. We studied the effects of distance to forest, tree management, and floral resources on bee communities in pastures (the dominant land-use type) in southern Costa Rica. Over two years, we sampled bees and floral resources in 21 pastures at three distance classes from a large (approximately 230-ha) forest patch and of three common types: open pasture; pasture with remnant trees; and pasture with live fences. We found no consistent differences in bee diversity or abundance with respect to pasture management or floral resources. Bee community composition, however, was strikingly different at forest edges as compared to deforested countryside only a few hundred meters from forest. At forest edges, native social stingless bees (Apidae: Meliponini) comprised approximately 50% of the individuals sampled, while the alien honeybee Apis mellifera made up only approximately 5%. Away from forests, meliponines dropped to approximately 20% of sampled bees, whereas Apis increased to approximately 45%. Meliponine bees were also more speciose at forest edge sites than at a distance from forest, their abundance decreased with continuous distance to the nearest forest patch, and their species richness was correlated with the proportion of forest cover surrounding sample sites at scales from 200 to 1200 m. Meliponines and Apis together comprise the eusocial bee fauna of the study area and are unique in quickly recruiting foragers to high-quality resources. The diverse assemblage of native meliponine bees covers a wide range of body sizes and flower foraging behavior not found in Apis, and populations of many bee species (including Apis), are known to fluctuate considerably from year to year. Thus, the forest-related changes in eusocial bee communities we found may have important implications for: (1) sustaining a diverse bee fauna in tropical countryside; (2) ensuring the effective pollination of a diverse native plant community; and (3) the efficiency and stability of agricultural pollination, particularly for short-time-scale, mass-flowering crops such as coffee.     

Occurrence and persistence of perfluorooctanesulfonate and other perfluorinated surfactants in groundwater at a fire-training area at Wurtsmith Air Force Base,Michigan, USA

Various formulations of fire-extinguishing materials, including aqueous film-forming foams (AFFFs), were used as part of fire-training exercises conducted at Wurtsmith Air Force Base (WAFB) in northeastern Michigan from the 1950s until the base was decommissioned in 1993. As a result of past fire-training exercises, AFFF-laden wastewater containing fuels, solvents, and other materials directly entered groundwater without prior treatment. Perfluorinated surfactants are key components in some AFFF formulations. In this study, groundwater was analyzed for perfluoroalkanesulfonates and perfluorocarboxylates. Perfluoroalkanesulfonates were directly detected using negative-ion electrospray ionization mass spectrometry. Derivatized perfluorocarboxylates were detected using electron impact gas chromatography-mass spectrometry. Groundwater from wells around fire-training area FTA-02 at WAFB contained four perfluorinated surfactants ranging in concentration from 3 to 120 microg L(-1): perfluorooctanesulfonate (PFOS); perfluorohexanesulfonate; perfluorooctanoate; and perfluorohexanoate. This is the first report demonstrating that PFOS, recently shown to be toxic to organisms ranging from zooplankton to primates, is still present in groundwater in measurable quantities five or more years after its last known use.