Machine learned daily life history classification using low frequency tracking data and automated modelling pipelines: application to North American waterfowl.

BACKGROUND: Identifying animal behaviors, life history states, and movement patterns is a prerequisite for many animal behavior analyses and effective management of wildlife and habitats. Most approaches classify short-term movement patterns with high frequency location or accelerometry data. However, patterns reflecting life history across longer time scales can have greater relevance to species biology or management needs, especially when available in near real-time. Given limitations in collecting and using such data to accurately classify complex behaviors in the long-term, we used hourly GPS data from 5 waterfowl species to produce daily activity classifications with machine-learned models using "automated modelling pipelines". METHODS: Automated pipelines are computer-generated code that complete many tasks including feature engineering, multi-framework model development, training, validation, and hyperparameter tuning to produce daily classifications from eight activity patterns reflecting waterfowl life history or movement states. We developed several input features for modeling grouped into three broad categories, hereafter "feature sets": GPS locations, habitat information, and movement history. Each feature set used different data sources or data collected across different time intervals to develop the "features" (independent variables) used in models. RESULTS: Automated modelling pipelines rapidly developed easily reproducible data preprocessing and analysis steps, identification and optimization of the best performing model and provided outputs for interpreting feature importance. Unequal expression of life history states caused unbalanced classes, so we evaluated feature set importance using a weighted F1-score to balance model recall and precision among individual classes. Although the best model using the least restrictive feature set (only 24 hourly relocations in a day) produced effective classifications (weighted F1 = 0.887), models using all feature sets performed substantially better (weighted F1 = 0.95), particularly for rarer but demographically more impactful life history states (i.e., nesting). CONCLUSIONS: Automated pipelines generated models producing highly accurate classifications of complex daily activity patterns using relatively low frequency GPS and incorporating more classes than previous GPS studies. Near real-time classification is possible which is ideal for time-sensitive needs such as identifying reproduction. Including habitat and longer sequences of spatial information produced more accurate classifications but incurred slight delays in processing.

Mercury cycling in agricultural and managed wetlands of California, USA: seasonal influences of vegetation on mercury methylation, storage, and transport.

Plants are a dominant biologic and physical component of many wetland capable of influencing the internal pools and fluxes of methylmercury (MeHg). To investigate their role with respect to the latter, we examined the changing seasonal roles of vegetation biomass and Hg, C and N composition from May 2007-February 2008 in 3 types of agricultural wetlands (domesticated or white rice, wild rice, and fallow fields), and in adjacent managed natural wetlands dominated by cattail and bulrush (tule). We also determined the impact of vegetation on seasonal microbial Hg methylation rates, and Hg and MeHg export via seasonal storage in vegetation, and biotic consumption of rice seed. Despite a compressed growing season of ~3months, annual net primary productivity (NPP) was greatest in white rice fields and carbon more labile (leaf median C:N ratio=27). Decay of senescent litter (residue) was correlated with microbial MeHg production in winter among all wetlands. As agricultural biomass accumulated from July to August, THg concentrations declined in leaves but MeHg concentrations remained consistent, such that MeHg pools generally increased with growth. Vegetation provided a small, temporary, but significant storage term for MeHg in agricultural fields when compared with hydrologic export. White rice and wild rice seeds reached mean MeHg concentrations of 4.1 and 6.2ng gdw(-1), respectively. In white rice and wild rice fields, seed MeHg concentrations were correlated with root MeHg concentrations (r=0.90, p<0.001), suggesting transport of MeHg to seeds from belowground tissues. Given the proportionally elevated concentrations of MeHg in rice seeds, white and wild rice crops may act as a conduit of MeHg into biota, especially waterfowl which forage heavily on rice seeds within the Central Valley of California, USA. Thus, while plant tissues and rhizosphere soils provide temporary storage for MeHg during the growing season, export of MeHg is enhanced post-harvest through increased hydrologic and biotic export.

The impact of item order on ratings of cancer risk perception.

Although perceived risk is central to most theories of health behavior, there is little consensus on its measurement with regard to item wording, response set, or the number of items to include. In a methodological assessment of perceived risk, we assessed the impact of changing the order of three commonly used perceived risk items: quantitative personal risk, quantitative population risk, and comparative risk. Participants were 432 men and women enrolled in an ancillary study of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Three groups of consecutively enrolled participants responded to the three items in one of three question orders. Results indicated that item order was related to the perceived risk ratings of both ovarian (P < 0.05) and colorectal (P < 0.05) cancers. Perceptions of risk were significantly lower when the comparative rating was made first. The findings suggest that compelling participants to consider their own risk relative to the risk of others results in lower ratings of perceived risk. Although the use of multiple items may provide more information than when only a single method is used, different conclusions may be reached depending on the context in which an item is assessed.