Combining animal interactions and habitat selection into models of space use: a case study with white‐tailed deer

Wildlife Biology, Год журнала: 2024, Номер 2024(3)

Опубликована: Фев. 8, 2024

Animals determine their daily movement trajectories in response to a network of ecological processes, including interactions with other organisms, memories previous events, and the changing environment. These combine cause emergent space use patterns observed over longer periods time, such as whole season. Understanding which processes these emerge, how, requires process‐based modelling approach. Individual‐based decisions can be described system partial‐differential equations (PDEs) produce dynamic description built from underlying process. Here we PDE‐based models step‐selection analysis investigate combined effects three established that partially shape use: 1) heterogeneous environment; 2) environmental markings moving conspecifics; 3) memory direct conspecifics. We apply this framework large GPS‐based dataset white‐tailed deer Odocoileus virginianus southeastern US. fit at population level provide predictive models, then tailor individual deer. specifically incorporate relationships between each possible pair define animal's responses unique local environments using separate integrated analyses. show how movements yield animal distributions, full generalised so it may applied any species simultaneously responding multiple potentially interacting stimuli (e.g. sociality, morphology, etc.). found bucks had highly varied preferences for vegetation, but were shaping conspecific interactions, dependent on two advocate increased consideration individual‐based rules determinants realized use, particularly affect distributions entire species.

Язык: Английский

---

Trophic interactions will expand geographically but be less intense as oceans warm

Global Change Biology, Год журнала: 2020, Номер 26(12), С. 6805 - 6812

Опубликована: Окт. 5, 2020

Abstract Interactions among species are likely to change geographically due climate‐driven range shifts and in intensity physiological responses increasing temperatures. Marine ectotherms experience temperatures closer their upper thermal limits the paucity of temporary refugia compared those available terrestrial organisms. Thermal marine also vary trophic levels, making interactions more prone changes as oceans warm. We assessed how temperature affects reef fish Western Atlantic modeled projections occurrence, biomass, feeding across latitudes climate change. Under ocean warming, tropical reefs will diminished interactions, particularly herbivory invertivory, potentially reinforcing algal dominance this region. Tropicalization events occur northern hemisphere, where by herbivores is predicted expand from Caribbean extratropical reefs. Conversely, omnivores decrease area with minor increases southern Brazil. Feeding invertivores declines all future predictions, jeopardizing a critical link. Most 2050 can significantly affect ecosystem functioning, causing rise novel ecosystems.

Язык: Английский

---

Can We Use Machine Learning for Agricultural Land Suitability Assessment?

Agronomy, Год журнала: 2021, Номер 11(4), С. 703 - 703

Опубликована: Апрель 7, 2021

It is vital for farmers to know if their land suitable the crops that they plan grow. An increasing number of studies have used machine learning models based on use data as an efficient means mapping suitability. This approach relies assumption grow in best-suited areas, but no systematically tested this assumption. We aimed test specialty Denmark. First, we mapped suitability 41 using learning. Then, compared predicted suitabilities with mechanistic model ECOCROP (Ecological Crop Requirements). The results showed there was little agreement between and ECOCROP. Therefore, argue methods represent different phenomena, which label socioeconomic ecological suitability, respectively. In most cases, predicts ambiguity term can lead misinterpretation. highlight need awareness distinction a way forward agricultural assessment.

Язык: Английский

---

Characterising Essential Fish Habitat using spatio‐temporal analysis of fishery data: A case study of the European seabass spawning areas

Fisheries Oceanography, Год журнала: 2021, Номер 30(4), С. 413 - 428

Опубликована: Янв. 8, 2021

Abstract Fish habitats sustain essential functions for fish to complete their life cycle, such as feeding, growing and spawning. Conservation is crucial maintain populations exploitation. Since 2013, the spawning stock biomass of northern European seabass ( Dicentrarchus labrax ) has been in a worrying state. A series low recruitments with persistently high level fishing blamed, raising concerns about processes involved reproduction settlement nurseries. Here, we characterise areas along French Atlantic coast using vessel monitoring system (VMS) data. non‐linear geostatistical approach was applied, from 2008 2014, detect locations where aggregate Occurrence maps distribution were combined into probability quantify seasonal inter‐annual variability highlight recurrent, occasional unfavourable areas. We identified three main areas: Rochebonne Plateau Bay Biscay, Western English Channel North Cotentin peninsula Eastern Channel. The correlative link between this geographical environmental factors investigated Bayesian spatio‐temporal model. structure accounted vast majority model predictive skills, whereas covariates had negligible effect. Our revealed persistence spatial intra‐ variability. Offshore appear be seabass, should considered management strategies.

Язык: Английский

---

The New Dominator of the World: Modeling the Global Distribution of the Japanese Beetle under Land Use and Climate Change Scenarios

Land, Год журнала: 2022, Номер 11(4), С. 567 - 567

Опубликована: Апрель 12, 2022

The spread of invasive species is a threat to global biodiversity. Japanese beetle native Japan, but alien populations this insect occur in North America, and recently, also southern Europe. This was recently included on the list priority European concern, as it highly agricultural pest. Thus, study, we aimed at (i) assessing its current distribution range, identifying areas potential invasion, (ii) predicting using future climatic land-use change scenarios for 2050. We collected occurrences available citizen science platform iNaturalist, combined data with predictors Bayesian framework, specifically integrated nested Laplace approximation, stochastic partial differential equation. found that mainly, positively, driven by percentage croplands, annual range temperature, habitat diversity, human settlements, population density; negatively related distance airports, elevation, mean temperature diurnal wetlands, waters. As result, based conditions, likely 47,970,200 km2, while will from between 53,418,200 59,126,825 according 2050 scenarios. concluded high-risk species, able find suitable conditions colonization several regions around globe, especially light ongoing change. strongly recommend strict biosecurity checks quarantines, well regular pest management surveys, order reduce spread.

Язык: Английский

---

Identifying optimal variables for machine-learning-based fish distribution modeling

Canadian Journal of Fisheries and Aquatic Sciences, Год журнала: 2024, Номер 81(6), С. 687 - 698

Опубликована: Март 5, 2024

Machine learning occupies a central position in the modeling of fish distribution patterns. The augmentation explanatory variables habitat through many kinds observational methodologies necessitates discernment an optimal combination these for modeling. We proposed feature selection technique, recursive elimination with cross-validation (RFECV), to determine combinations yellowfin tuna Pacific Ocean. Four tree-based models, random forest, eXtreme Gradient Boosting, Light Boosting Machine, and categorical boosting driven by RFECV, were developed using comprehensive fisheries biotic/abiotic data. Habitat including sea temperature, dissolved oxygen concentration, chlorophyll-a salinity, surface height identified as significant features all models. models trained corresponding selected variables, employed predict spatiotemporal from 1995 2019. results obtained could inform useful knowledge sustainable exploitation Ocean furnish benchmark machine-learning-based other pelagic species.

Язык: Английский

---

Dealing with area‐to‐point spatial misalignment in species distribution models

Ecography, Год журнала: 2024, Номер 2024(5)

Опубликована: Март 22, 2024

Species distribution models (SDMs) are extensively used to estimate species–environment relationships (SERs) and predict species across space time. For this purpose, it is key choose relevant spatial grains for predictor response variables at the onset of modelling process. However, environmental often derived from large‐scale climate a grain that can be coarser than one variable. Such area‐to‐point misalignment bias estimates SER jeopardise robustness predictions. We virtual approach, running simulations different levels seek statistical solutions problem. specifically compared accuracy predictive performances, assessed degrees heterogeneity in conditions, three SDMs: GLM, GLM Berkson error model (BEM) accounts fine‐grain within coarse‐grain cells. Only BEM accurately relatively data (up 50 times grain), while two GLMs provide flattened SER. all perform poorly when predicting data, particularly environments more heterogeneous training conditions. Conversely, decreasing relative dataset reduces biases. Because predictions made covariate‐grain displays lower performance GLMs. Thus, standard selection methods would fail select best SERs (here, BEM), which could lead false interpretations about drivers distributions. Overall, we conclude BEM, because robustly grain, holds great promise overcome misalignment.

Язык: Английский

---

RangeShiftR: an R package for individual‐based simulation of spatial eco‐evolutionary dynamics and species' responses to environmental changes

Ecography, Год журнала: 2021, Номер 44(10), С. 1443 - 1452

Опубликована: Авг. 29, 2021

Reliably modelling the demographic and distributional responses of a species to environmental changes can be crucial for successful conservation management planning. Process‐based models have potential achieve this goal, but so far they remain underused predictions species' distributions. Individual‐based offer additional capability model inter‐individual variation evolutionary dynamics thus capture adaptive change. We present RangeShiftR, an R implementation flexible individual‐based platform which simulates eco‐evolutionary in spatially explicit way. The package provides fast simulations by making software RangeShifter available widely used statistical programming R. features auxiliary functions support specification analysis results. provide outline package's functionality, describe underlying structure with its main components short example. RangeShiftR offers substantial complexity, especially dispersal processes. It comes elaborate tutorials comprehensive documentation facilitate learning help at all levels. As core code is implemented C++, computations are fast. complete source published under public licence, adaptations contributions feasible. facilitates application mechanistic questions operating powerful simulation from allows effortless interoperation existing packages create streamlined workflows that include data preparation, integrated results analysis. Moreover, strengthens coupling other models.

Язык: Английский

---

The shadow model: how and why small choices in spatially explicit species distribution models affect predictions

PeerJ, Год журнала: 2022, Номер 10, С. e12783 - e12783

Опубликована: Фев. 14, 2022

The use of species distribution models (SDMs) has rapidly increased over the last decade, driven largely by increasing observational evidence distributional shifts terrestrial and aquatic populations. These permit, for example, quantification range shifts, estimation co-occurrence, association habitat to abundance. complexity contemporary SDMs presents new challenges—as choices among modeling options increase, it is essential understand how these affect model outcomes. Using a combination original analysis literature review, we synthesize effects three common in semi-parametric predictive process modeling: structure, spatial extent data, scale predictions. To illustrate choices, develop case study centered around sablefish ( Anoplopoma fimbria ) on west coast USA. represent decisions necessary virtually all ecological applications methods, are important because consequences impact derived quantities interest e.g ., estimates population size their management implications). Truncating data near observed edge, or using that misspecified terms covariates spatiotemporal fields, led bias biomass trends mean compared from full dataset appropriate structure. In some cases, suboptimal may be unavoidable, but understanding tradeoffs impacts predictions critical. We seemingly small often made out necessity simplicity, can scientific advice informing decisions—potentially leading erroneous conclusions about changes abundance precision such estimates. For show incorrect could cause overestimation abundance, which result resulting overfishing. Based findings gaps, outline frontiers SDM development.

Язык: Английский

---

Seasonal occurrence and abundance of dabbling ducks across the continental United States: Joint spatio‐temporal modelling for the Genus Anas

Diversity and Distributions, Год журнала: 2019, Номер 25(9), С. 1497 - 1508

Опубликована: Июнь 18, 2019

Abstract Aim Estimating the distribution and abundance of wildlife is an essential task in species conservation, management habitat prioritization. Although a host methods tools have been proposed to accomplish this undertaking, several challenges remain accurately forecasting occurrence for highly mobile species. Exhibiting extensive geographic ranges with seasonally varying local occupancy, migratory ducks are exemplar foci waterfowl conservation globally. With goal informing management, our aim was leverage citizen science data estimate relative ten dabbling duck across continental United States. Location Conterminous Methods We applied spatially temporally explicit Bayesian hierarchical modelling jointly season‐specific Genus Anas . Our conditionally dependent model design enabled estimates be informed by probability while accounting cumulative spatial temporal errors shared components. Results Outcomes suggest that although distributions show little inter‐annual variability at scale, abundances may differ year year. Commensurate being species, indicate considerable intra‐annual variation probability, preferences differing season Main conclusions approach offers powerful flexible framework quantifying intra‐/inter‐annual spatial, collection biases. believe produced maps can inform throughout

Язык: Английский

---