This custom report represents a summary based on user-defined subsets of species and spatial prediction units, and the model results presented at sc-dev.abmi.ca. The model outputs and predictions were summarized with cure4insect R package (see Settings).
Field data collected by the ABMI were used alongside with data collected and maintained by other organizations and individuals.
A set of statistical models were used to relate the relative abundance of each species to land cover and space/climate variables. Analyses were done separately for two overlapping regions: North (boreal, foothills, Canadian shield, parkland), and South (grasslands, parkland, dry mixedwood subregion of the boreal).
Land cover types in the North region included main forest stand types (white spruce, pine, deciduous, mixedwood, black spruce) by broad age classes (0-9 years, 10-19 years, then 20-year classes to 140+ years) and categories of open vegetation (upland grass, upland shrub, treed fens, other treed wetlands, and non-treed wetlands). In the South, land cover types were broad ecosite (soil) types, such as productive, clay, saline, and rapid drained. The South models also included a term for the probability that the site is or was naturally treed. The human footprint classes used both in the North and South included Alienating (cultivation, urban-industrial, and hard (unvegetated) linear features) and Successional disturbances (soft (vegetated) linear features, and forestry which is differentiated by broad stand type and the same age classes as natural forest.
The following climate variables were used in our modeling: Annual heat-moisture index, Frost free period, Mean annual precipitation, Mean annual temperature, Mean cold month (January) temperature, Mean warm month (July) temperature, Potential evapotranspiration. Besides climate variables, we also used public coordinates (latitude, longitude) in modeling.
Spatial and climate variables modify the land cover effects in our models introducing spatial variation into them. We used a grid system for spatial predictions with 1 km2 pixels through the province of Alberta, summarized amount of native vegetation and human footprint types in each 1 km2 pixel to be used in area weighted relative abundance predictions. Centroids of the pixels were intersected with the GIS layers described above to extract spatial information for prediction.
We combined information on the human footprint and vegetation types within 1 km2 units throughout the province with the climate and location information for each 1 km2 unit, and then predicted abundance of each species based on the habitat models and the effects of location and climate. This is the predicted current relative abundance of the species. To map reference relative abundance of each species, we used a map in which all human footprint has been back-filled by the vegetation type that was most likely to have occurred prior to footprint. Back-filling in the North is based on the vegetation types immediately surrounding the human footprint, as well as rules about which footprint types are restricted to certain vegetation types. In the South, the ecosystem variables, and Aspen probability are mapped for all areas, including in human footprint, and therefore we do not need to back-fill human footprint to get reference conditions. We simply ignore the effects of human footprint and predict all areas based only on the ecosite types and Aspen probability of occurrence.
Intactness for a single species is defined as SI = 100 x min(Current, Reference) / min(Current, Reference) and is in the range of 0–100% for each 1 km2 mapping unit. The software uses the same calculation for all species irrespective of their native or non-native status. The 'two-sided' intactness differentiates between the relative abundance increase an decrease and is a trasformation of the 'one-sided' (0–100%) intactness (200 – SI) when Current > Reference, thus takes values in the 0–200% range. Intactness based on multiple species is a simple average of the species in each 1 km2 mapping unit.
Species richness is the sum of occurrence probabilities of multiple species within each 1 km2 mapping unit
Uncertainty in predicted current conditions was determined as the standard error of bootstrap based predictive distribution of current abundance aggregated at the 10 km x 10 km units.
We use the species-habitat models to estimate the effects of individual industrial sectors on each species in a given region. We differentiate the following sectors: Agriculture, Forestry, Energy, Rural/Urban, and Transportation. We summarize current and relative abundance for each industrial sector in that region, besides relative abundance in native (undisturbed) land cover types (where current and reference abundances are equal).
Sector effects on regional population for a given sector are calculated as the difference between sector specific current and reference abundance, standardized by the total reference abundance in the entire region (including all sectors and native land cover types).
Sector effects on population under footprint for a given sector are calculated as the difference between sector specific current and reference abundance, standardized by the reference abundance in that sector.
Per unit area effects are calculated from the regional population effects, but further divided by the proportion of the sector specific footprint in the region. The corresponding graphs show the area of the sector's footprint as the width of the bars, and the per unit area effect as the height of the bars. The area of the bar corresponds to the regional sector effect.