Flexible species distribution modelling methods perform well on spatially separated testing data

Global Ecology and Biogeography, Journal Year: 2023, Volume and Issue: 32(3), P. 369 - 383

Published: Jan. 27, 2023

Abstract Aim To assess whether flexible species distribution models that perform well at nearby testing locations still strongly when evaluated on spatially separated data. Location Australian Wet Tropics (AWT), Ontario, Canada (CAN), north‐east New South Wales, Australia (NSW), Zealand (NZ), five countries of America (SA), and Switzerland (SWI). Time period Most data were collected between 1950 2000. Major taxa studied Birds, mammals, plants reptiles. Methods We compared 10 modelling methods with varying flexibility in terms the allowed complexity their fitted functions [boosted regression trees (BRT), generalized additive model (GAM), multivariate adaptive splines (MARS), maximum entropy (MaxEnt), support vector machine (SVM), variants linear (GLM) random forest (RF), an Ensemble model]. used established practices for selection to avoid overfitting, including parameter tuning learning methods. Models trained presence–background 171 tested presence–absence Training using both spatial partitioning, latter based 75‐km blocks. calculated average performance mean rank (focussing area under receiver operating characteristic precision‐recall gain curves, correlation) assessed statistical significance differences them. Results The ranking did not change strongest predictive nonparametric known be flexible. An ensemble formed by averaging predictions pre‐selected was best followed MaxEnt a variant forest. Main conclusions Whilst some modellers expect limited simple smooth predict better data, we found no evidence blocks 75 km. conclude are tuned enough overfitting effective predicting distinct areas.

Language: Английский

---

Adapting machine learning for environmental spatial data - A review

Ecological Informatics, Journal Year: 2024, Volume and Issue: 81, P. 102634 - 102634

Published: May 11, 2024

Large-scale modeling of environmental variables is an increasingly complex but necessary task. In this paper, we review the literature on using machine learning to cope with challenges associated spatial autocorrelation. Our focus was studies in which researchers predicted a supervised regression algorithm that accounted for autocorrelation any part pipeline from data exploration model validation. Methods included explicit covariates, splitting training–testing, calculations, and independent exploratory analysis. Authors most often analysis had no impact values. We concluded there seems be overall systematic approach how account models. selected studies, appropriate method depended specific characteristics study. Using covariates training-testing provided more insights into method's applicability. summarize these provide considerations selecting method.

Language: Английский

---

Model‐based variance partitioning for statistical ecology

Ecological Monographs, Journal Year: 2025, Volume and Issue: 95(1)

Published: Jan. 15, 2025

Abstract Variance partitioning is a common tool for statistical analysis and interpretation in both observational experimental studies ecology. Its popularity has led to proliferation of methods with sometimes confusing or contradicting interpretations. Here, we present variance model‐based Bayesian framework as general summarizing interpreting regression‐like models produce additional insight on ecological compared what traditional parameter inference these its own can reveal. For example, propose predictive extend sample‐based analyses whole populations scenarios. We also encompass within between ecologically relevant subgroups the observations, population interest, provide information how relative roles processes underlying study system may vary depending environmental context. discuss role correlated covariates random effects highlight uncertainty quantification partitioning. To showcase utility our approach, case comprising simple occupancy model metapopulation Glanville fritillary butterfly. As result, demonstrate rigorous gain more from data.

Language: Английский

---

The use of GEDI canopy structure for explaining variation in tree species richness in natural forests

Environmental Research Letters, Journal Year: 2022, Volume and Issue: 17(4), P. 045003 - 045003

Published: Feb. 24, 2022

Variables describing the abiotic environment (e.g. climate, topography or biogeographic history) have a long tradition of use as predictors tree species richness patterns. However, these variables may capture variations in related to but not those that are soil type forest disturbance. Canopy structure has previously been shown provide information on variation richness, with generally increasing larger canopy heights and denser foliage. The is increasingly relevant availability such data from Global Ecosystem Dynamics Investigation (GEDI), lidar mission onboard International Space Station. In this analysis we show GEDI explains up 66% natural forests without history recent disturbance across globe. portion overlaps (up 80%) explained by environmental biogeographical variables. Our results relationships between one side climate other straightforward initially expected, should be further investigated both disturbed forests.

Language: Английский

---

Using automated passive acoustic monitoring to measure changes in bird and bat vocal activity around hedgerows of different ages

Biological Conservation, Journal Year: 2024, Volume and Issue: 296, P. 110722 - 110722

Published: July 19, 2024

Hedgerows are a semi-natural habitat that supports farmland biodiversity by providing food, shelter, and connectivity. Hedgerow planting goals have been set across many countries in Europe agri-environment schemes (AES) play key role reaching these targets. Passive acoustic monitoring using automated vocalisation identification (automated PAM), offers valuable opportunity to assess changes following AES implementation simple, community-level metrics, such as vocal activity of birds bats. To evaluate whether could be used indicate the effectiveness hedgerow future result-based or hybrid schemes, we surveyed twenty-four hedgerows England classified into chrono-sequence three age categories (New, Young, Old). We recorded 4466 h over course 30 days measured bird bat BirdNET for Kaleidoscope Vocal all birds, bats were modelled with predictors hedgerow, habitat, weather conditions occurring from maturity. show an increase Young Old compared New ones highlight elements surrounding landscape should considered when evaluating on communities. found high precision low species-level observations, argue may novel link payment component PAM results, incentivising effective management farmers landowners.

Language: Английский

---

Accounting for temporal change in multiple biodiversity patterns improves the inference of metacommunity processes

Ecology, Journal Year: 2022, Volume and Issue: 103(6)

Published: March 21, 2022

In metacommunity ecology, a major focus has been on combining observational and analytical approaches to identify the role of critical assembly processes, such as dispersal limitation environmental filtering, but this work largely ignored temporal community dynamics. Here, we develop "virtual ecologist" approach evaluate processes by simulating metacommunities varying in three main processes: density-independent responses abiotic conditions, density-dependent biotic interactions, dispersal. We then calculate number commonly used summary statistics structure space time use random forests their utility for inferring strength these processes. find that (i) both spatial data are necessary disentangle based test, including measured through increases explanatory power up 59% compared cases where only variation is considered; (ii) studied can be distinguished with different descriptors; (iii) each statistic differently sensitive sampling effort. Including repeated observations over was essential particularly Some most useful include coefficient species abundances metrics incorporate relative (evenness) species. conclude combination methods probably understand underlie time, recognize results will modified when other or used.

Language: Английский

---

The effects of longitudinal fragmentation on riverine beta diversity are modulated by fragmentation intensity

The Science of The Total Environment, Journal Year: 2023, Volume and Issue: 903, P. 166703 - 166703

Published: Sept. 6, 2023

The loss of longitudinal connectivity affects river systems globally, being one the leading causes freshwater biodiversity crisis. Barriers alter dispersal aquatic organisms and limit exchange species between local communities, disrupting metacommunity dynamics. However, interplay losses due to dams other drivers structure, such as configuration network, needs be explored. In this paper, we analyzed response fish communities network position fragmentation induced by while controlling for human pressures environmental gradients. We studied three large European catchments covering a gradient: Upper Danube (Austrian section), Ebro (Spain), Odra/Oder (Poland). quantified through reach-scaled indices that account barriers along dendritic capacity organisms. used generalized linear models explain richness Local Contributions Beta Diversity (LCBD) multilinear regressions on distance matrix describe its Replacement Richness Difference components. Results show was not affected fragmentation. Network centrality metrics were relevant beta diversity with lower (Ebro, Odra), strong predictors catchment higher (Danube). conclude in highly fragmented catchments, effects centrality/isolation could masked dam metapopulation dynamics can strongly altered barriers, restoration (i.e. natural gradient) is urgent prevent extinctions.

Language: Английский

---

Nineteenth-century land use shapes the current occurrence of some plant species, but weakly affects the richness and total composition of Central European grasslands

Landscape Ecology, Journal Year: 2025, Volume and Issue: 40(1)

Published: Jan. 13, 2025

Language: Английский

---

Leveraging synergies between UAV and Landsat 8 sensors to evaluate the impact of pale lichen biomass on land surface temperature in heath tundra ecosystems

The Science of The Total Environment, Journal Year: 2025, Volume and Issue: 969, P. 178982 - 178982

Published: March 1, 2025

Pale terricolous lichens are a vital component of Arctic ecosystems, significantly contributing to carbon balance, energy regulation, and serving as primary food source for reindeer. Their characteristically high albedo also impacts land surface temperature (LST) dynamics across various spatial scales. However, remote sensing is challenging due their complex spectral signatures large variations in coverage biomass even within local landscape This study evaluates the influence pale on LST at scales by integrating RGB, multispectral, thermal infrared imagery from an Unmanned Aerial Vehicle (UAV) with multi-temporal Landsat 8 data. An Extreme Gradient Boosting algorithm was employed map lichen biomass, areal extent, occurrence major plant functional types sub-arctic heath tundra Jávrrešduottar Sieiddečearru areas Finland-Norway border. Generalized Additive Models (GAMs) were used elucidate factors affecting LST. The UAV model accurately predicted (R2 0.63) vascular vegetation cover 0.70). GAMs revealed that regimes, increased leading decreased LST, effect more pronounced scale (deviance explained 47.26 % 65.8 models, respectively). identified second most important variable both scales, elevation being variable. research demonstrates capability UAV-derived models capture heterogeneous fine-scale structure ecosystems. Furthermore, it underscores effectiveness combining resolution temporal satellite platforms. Finally, this highlights pivotal role showcases how advanced techniques can be ecological monitoring management.

Language: Английский

---

Environmental constraints and diffusion shaped the global transition to food production

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 10, 2025

The transition from foraging to plant cultivation represents the most important shift in economic history of early Holocene societies. This process unfolded independently different regions globe, resulting varied assemblages, strategies, dietary practices, and landscape modifications. To investigate drivers this transition, we employed a machine-learning approach. Using Random Survival Forest, analyze comprehensive dataset radiocarbon dates linked first adoption domesticated plants, coupled with environmental predictors. Our findings indicate strong spatial autocorrelation timing agricultural adoption, underscoring role diffusion contact between regions. Region-specific bioclimatic factors emerged as influential: Americas, mean temperature seasonality were critical, while Southwest Asia Europe, seasonal variation precipitation relative held greater importance. These results suggest that facilitated spread practices shaped by local conditions, it was not possible determine set universal drivers. Thus, emergence food production influenced combination cultural transmission, leaving specific determinants for each region's an open question further study.

Language: Английский

---