Search

Integration of species traits in direct gradient analysis generates insights into how communities are assembled and may respond to environmental changes. We investigated phytoplankton trait‐environment relationships for over 300 taxa during the open‐water season across 75 north‐temperate lakes and reservoirs in Alberta, Canada. An innovative, data‐driven approach was applied using iterative model selection for RLQ optimization to reveal key monthly associations. Fourth‐corner analysis then tested the significance of relationships using spatially and phylogenetically constrained null models derived by Moran spectral randomization. Both local‐ and regional‐scale drivers of succession were found with evidence of deterministic filtering by traits increasing in mid‐summer. Trait associations to land‐use and water quality highlighted potential anthropogenic cross‐scale interactions, such as logging and pasture development affecting phytoplankton via allochthonous nutrient inputs. Biogeographic factors (e.g., elevation and habitat size) and associated temperature and chemical gradients (e.g., pH and bicarbonate) were also linked to multiple morphological, physiological, and behavioral traits, including potential toxin production. Several correlated traits emphasized importance of trait syndromes corresponding to distinct taxonomic groups (e.g., cyanobacteria and green algae). However, rather than clustering species with shared ecological preferences or roles, our study builds on past trait‐based approaches to phytoplankton by testing for explicit trait associations with a range of environmental factors. Thus, we provide a novel, quantitative means of revealing environmental constraints on communities and translating their compositional changes under climate and other human influences into functional impacts on freshwater ecosystems. [ABSTRACT FROM AUTHOR]

Estimating number of prey killed by carnivores such as wolves (Canis lupus) per unit time, or kill rate, is important for the conservation and management of carnivores and their prey. We reviewed published methods to estimate wolf kill rates and found them inconsistent and lacking a basis in statistical sampling theory. We developed a general statistical estimator for kill rate and variance using ratio-variable sampling theory. We illustrate our ratio estimator by estimating wolf kill rates in a multiple-prey system in Banff National Park, Alberta, for winters from 1986 to 2000. We used snow tracking and radiotelemetry to locate 429 kills during 195 sampling intervals covering 1,294 days. Mean kill rate by wolf packs (expressed as the number of kills/day/pack, k/d/p) was 0.33 k/d/p, most of which, 0.23 k/d/p, were elk (Cervus elaphus), the most abundant ungulate. Kill-rate estimates were variable despite intensive sampling effort. The mean kill rate of 0.33 k/d/p had a pooled 95% confidence interval of 0.29 to 0.37. An intensive sampling effort of ≥6-8 individual sampling intervals covering approximately 25% of the winter was required to minimize sampling variation. We compared our method to 3 other published methods for estimating kill rates and examined the bias and precision of kill-rate methods using simulations. Our ratio estimator approach was the least biased and most precise when compared to other approaches. Ratio estimators provide a standardized method to estimate kill rates in other predator-prey systems and will facilitate comparison across studies and examination of patterns of kill-rate variation. [ABSTRACT FROM AUTHOR]