Appendix A. A description of preliminary sampling conducted in August 2001.
On each island except Chagulak we established 3 transects, with each transect originating at a randomly selected point chosen from those that could be accessed by water around the perimeter of each island. We were unable to establish transects at randomly located coastal points around Chagulak Island because much of the island is extremely steep and inaccessible from the ocean. Therefore, we sampled a series of points along one transect on the only accessible portion of the island.
At distances of 20, 100, 200, and 400 m (or the island center, if the island was less than 800 m across) from the shore along each transect, we established a modified Whittaker plot (30 m × 30 m, 900 m2). Within each plot we established seven 1-m2 subplots arranged equidistant from each other 0 m, 15 m, and 30 m along three sides of the 30 × 30 m plot plus an additional 1-m2 subplot in the center of the plot. Using a soil corer, we took soil samples at 510 cm beneath the soil surface within each of three 1-m2 subplots for nutrient and stable isotope analysis. Soil samples were collected from subplots that were located at two opposite corners and the center of each 30 × 30 m plot (these subplots formed a diagonal line across the plot). Each sample was bagged separately. A small subsample from one of these soil subplot samples was also bagged separately for stable isotope analyses. We also haphazardly collected multiple samples of the widespread grass, Leymus mollis, within each 30 × 30 m plot for stable isotope analysis. The soil aliquot and L. mollis sample were dried at 60 ˚C. The soil aliquot was passed through a 2-mm sieve and ground through a 40-mesh screen in a Wiley Mill. The grass sample was ground and then pulverized to a powder by shaking samples in vials with ball bearings. Processed soil and grass samples were sent to the University of California Davis Stable Isotope Laboratory, where δ13C and δ15N values (as well as %C and %N) were determined. The remaining soil samples were similarly dried, sieved, ground (through a 60-mesh screen in the Wiley Mill) and sent to the U.C. Davis DANR analytical laboratory for determination of total nitrogen (Carlo Erba combustion method) and extractable phosphorus (Bray method).
To quantify plant cover, we photographed the interior 0.8 m × 0.8 m region within all subplots using a digital camera mounted beneath a tripod positioned directly over each subplot. These photos were later analyzed by using PhotoShop (Adobe Systems, San Jose, California, USA) to overlay 56 uniformly distributed points (in a square array) onto each digital plot image. At each of these points we classified the vegetation bisecting each grid point as: (1) L. mollis, (2) graminoid other than L. mollis, (3) umbel, (4) forb, (5) dwarf shrub, (6) equisetum/fern, (7) moss/lichen, (8) rock, (9) bare soil or (10) unknown. Cover of these different classes was then calculated as the percentage of the 56 points that intersected these objects.