Book references for the study area: Ryzhanovsky V.N. (1997) Ecology of postbreeding period of subarctice passerine life. Ural University, Russian Academy of Science, Ekaterinburg, Russia. 288 pages.
Golovatin M, Paskhalny (2005). Birds of the Polar Ural. Ural University, Russian Academy of Science, Ekaterinburg, Russia. 560 pages.
BIODIVERSITY GRID For efficiency reasons a systematic sampling approach was chosen. First of all an equally spaced GRID was implemented: 25 points were arranged in five rows and five columns in order to cover a consistent area but also to have a known spatial neighbor relationship among all plots. The distance between plots was 100 m, resulting in a total GRID size of 400 m x 400 m. While the final GRID system ideally covers the globe systematically without intentional placement, for these initial studies the GRIDs were placed in a way that roughly half to two thirds of the plots fell inside a forested area, the remaining plots at the forest edge or inside the cultural landscape. This survey setup enables other studies on the same data set to make realistic and representative statements about fragmentation effects. The only exception is GRID in Barrow in northern Alaska, where naturally only one habitat type, arctic tundra, occurs. Additionally, five points were randomly placed within the GRID to be able to model the influence of random patterns on the results and their spatial relations (Figure 8). The coordinates of each plot were obtained from a regular hand-held GPS receiver and re-visited by using the Go to function. All plots as well as the path between them were marked with decomposing flagging tape to make recognition in the field easier. A simple schematic map was drawn by hand for each field work participant to ensure that plots are found when the GPS does not receive signals, as was often the case in dense forest settings. BUDGET CONSTRAINTS The biodiversity GRID is meant as a method for cost-efficient rapid biodiversity assessment that allows for an analysis of spatial relations as well. All methods involved have to work in relatively short time, with low costs and little demand of technological equipment. There is no objection to include more sophisticated methods in add-on protocols, but they are discouraged for the main protocol to keep the inhibition threshold for decision makers low. Trained taxonomists were not available, as they rarely are for many ecosystems. All notes regarding the observed species were made as precisely as possible, although most of the observers were not trained especially in tropical ornithology or entomology. Data collection followed the motto the more detail the better, but it was not intended to refuse data because of lacking taxonomic details. If the observer did not readily know the correct scientific name of a specimen, a common name or, in lack of knowledge of a common name, a short description was noted. This original field note is referred to as the narrative name of an observation respectively of a species. Such process is common when dealing with large numbers of species and in largely unexplored environments, where huge fractions of the biodiversity remains still unknown, or where appropriate taxonomic guide books are missing. This resulted in good abundance and occupancy estimates, but in less detailed taxonomic data. Such is the characteristic in rapid biodiversity assessments on shoestring budgets, which allow for a first impression and provide detailed information for deeper investigation if desired. This type of rapid assessment additionally serves as a pilot study for further assessments. In the present study the focus lies on spatial global coverage, instead of local detail. ANIMAL SPECIES DATA COLLECTION In the ideal case, the protocol should result not only in information about the presence or absence of species, but also in an estimate of population size. The DISTANCE sampling approach uses the concept of a detection function based on distance of the observed object from the observer to estimate population density. It plays a central role in this study and is used in a number of ways. At each of the 30 plots (25 systematic and 5 random), five minute point transect DISTANCE sampling counts for birds were conducted within 360 degrees. A short settle-in period of one minute was granted prior to counting to allow for the snapshot character of DISTANCE sampling, especially meeting the assumption that presence of the observer does not introduce bias by causing responsive movements of animals. Following common practice the point counts took place only in the morning between 5:30 and 10 am. Birds are known to show higher activity at this time, which generally increases detectability and maximizes inventory accuracy. Each bird seen or heard was noted, including an estimate of the radial distance from the observer. Double counts were avoided by the observers attention and the relatively short counting period. Observers decided to make two adjustments: - in study area on Sakhalin Island, Russia seabird observations were excluded from plot A1; - in study area in Barrow, Alaska the survey time was reduced from five to four minutes. The second method of DISTANCE sampling used was a trapping web. 17 pitfall traps with a diameter of 9 cm each were arranged in a DISTANCE sampling trapping web design to estimate ground-living insects. This sampling method is very labor-intensive and could not be implemented at all 30 plots given the short time period available. Thus, four of the plots were systematically selected to capture the general patterns of species and abundances within the GRID: B2, D2, B4 and D4 (underlined in Figure 8) to gather at least some information about ground-living insects. Trapping webs were usually checked every 24 hours; and records were taken every 48 hours. In between check dates the cups were emptied without recording to avoid correlation in time between trapping events, and obtain spatially independent results. Because of the low number of traps and more available work force it was decided to add a third circle of traps at 3 m from the centre in study areas in Russia, Papua New Guinea and Barrow, Alaska. This increased the total number of pitfall traps in these areas to 25. The third application of DISTANCE sampling was an add-on sampling protocol using DISTANCE sampling line transects, conducted at each of the 30 plots. Transects with a length of 10 m and traversing the plot at its centre were surveyed to estimate numbers of butterflies, amphibians and reptiles. DISTANCE sampling point counts for birds and trapping webs for ground living insects were repeated three times. These repetitive visits further allow for an analysis with the software PRESENCE, which gives an estimate of general occurrence of a species in the area in a point-based sense. PRESENCE generates a detection function based on multiple visits under the assumption that the population is closed, meaning that no animals leave or enter the area of interest between several visits. Repetitions were not realized for the add-on protocol for DISTANCE sampling line transects. VEGETATION & ENVIRONMENT Additionally, basic data about the plot environment was collected. If at all possible, the GPS coordinates were noted. A plot picture and a canopy picture were taken with a digital camera to give a general impression of the area and also allow for an analysis of light conditions in other studies on the same data set, e.g. remote sensing investigations. All pictures are available from the authors. A short description of the ecosystem was noted as well (for example: pasture, forest interior, forest edge). Height and diameter at breast height were recorded for all trees within 5 m of plot centre. Estimates were noted regarding canopy cover percentage, understory cover percentage, shrub cover percentage (at 1.35 m height), bare soil percentage, duff coverage percentage, leaf browsing percentage, and number of flowers visible. The thickness of epiphytes, hemi-epiphytes, mosses and lichen was noted in categories (none, low, medium, high). Presence/absence of identified plant species or plant families was noted, as well as remarkable animal tracks (e.g. land crab holes, large mammal tracks, etc). Those are referred to as Covariates 1 to 32 in all six study areas, but the actual meaning is different in each. Detailed lists and the full protocol are available from the authors. The covariates can have one of four effects: 1. affecting habitat quality (presence/ absence of a species) 2. affecting detectability (detection/ non-detection of a species that is present) 3. affecting both of the above 4. affecting none of the above.