Confidence and consistency in discrimination: A new family of evaluation metrics for potential distribution models

Ecological Modelling, Journal Year: 2024, Volume and Issue: 491, P. 110667 - 110667

Published: March 11, 2024

Potential distribution models (PDMs) are widely applied to understand and predict biogeographic patterns. PDM evaluation, however, presents major challenges, including (1) matches of predictions with observed absences presences being treated similarly (2) treatment predicted falling outside the observations as errors, while a motivation PDMs is identify such locations. Our aim was construct family model performance metrics measure reliability transferability providing solutions problems mentioned above. Instead binarisation, reclassified into three categories for evaluation: certain negative, uncertain positive predictions. Model tested solely within known reduce effect unoccupied but suitable sites registered absences. Metrics were developed both cases: when target modelling identification potential presence locations evaluation equally. The new measures offer optimised models. On one hand, proposed concentrate on presences. Thus, typically large amount do not inflate metric values. other treat all mismatches errors thus allow exploitation information in mismatches, too. Besides theoretical background, we also provide R package calculating measures. We field simulation data. Both field-based simulation-based case studies underlined that capture different aspect than traditional metrics, AUC, TSS, sensitivity specificity. conclude our help whether process could preferences object well enough reliably find further or stay reliable transferred space time.

Language: Английский

---

Variable impacts of land-based climate mitigation on habitat area for vertebrate diversity

Science, Journal Year: 2025, Volume and Issue: 387(6732), P. 420 - 425

Published: Jan. 23, 2025

Pathways to achieving net zero carbon emissions commonly involve deploying reforestation, afforestation, and bioenergy crops across millions of hectares land. It is often assumed that by helping mitigate climate change, these strategies indirectly benefit biodiversity. Here, we modeled the habitat requirements 14,234 vertebrate species show impact on species’ area tends not arise through mitigation, but rather conversion. Across locations, reforestation provide more both land-cover change whereas loss from afforestation cropping typically outweighs mitigation benefits. This work shows how where land-based can be deployed without inadvertently reducing for global

Language: Английский

---

Steppe-land birds under global change: insights from the Eurasian Stone-curlew (Burhinus oedicnemus) in the Western Palearctic

Global Ecology and Conservation, Journal Year: 2025, Volume and Issue: unknown, P. e03478 - e03478

Published: Feb. 1, 2025

Language: Английский

---

Trait‐based projections of climate change effects on global biome distributions

Diversity and Distributions, Journal Year: 2021, Volume and Issue: 28(1), P. 25 - 37

Published: Nov. 9, 2021

Abstract Aim Climate change will likely modify the global distribution of biomes, but magnitude is debated. Here, we followed a trait‐based, statistical approach to model influence climate on biomes. Location Global. Methods We predicted plant community mean specific leaf area (SLA), height and wood density as function soil characteristics using an ensemble models. Then, probability occurrence biomes three traits with classification model. Finally, projected changes in corresponding biome distributions 2070 for low (RCP 2.6; +1.2°C) extreme 8.5; +3.5°C) future scenarios. Results estimated that under scenario (sub)tropical expand (forest by 18%–22%, grassland 9%–14% xeric shrubland 5%–8%), whereas tundra temperate broadleaved mixed forests contract 30%–34% 16%–21%, respectively. Our results also indicate over 70%–75% current grasslands shift northwards. These become amplified which lose more than half its extent. Main conclusions considerable imminent alterations possibly major consequences life Earth. The level accuracy our given limited input data insights how trait–environment relationships can suggest trait‐based correlative approaches are promising tool forecast vegetation provide independent, complementary line evidence next process‐based

Language: Английский

---

Uncertainty matters: ascertaining where specimens in natural history collections come from and its implications for predicting species distributions

Ecography, Journal Year: 2022, Volume and Issue: 2022(9)

Published: June 14, 2022

Natural history collections (NHCs) represent an enormous and largely untapped wealth of information on the Earth's biota, made available through GBIF as digital preserved specimen records. Precise knowledge where specimens were collected is paramount to rigorous ecological studies, especially in field species distribution modelling. Here, we present a first comprehensive analysis georeferencing quality for all records served by GBIF, illustrate impact that coordinate uncertainty may have predicted potential distributions. We used analyse availability coordinates associated spatial across geography, resolution, taxonomy, publishing institutions collection time. three plant their native ranges different parts world show found 38% 180+ million provide only 18% uncertainty. Georeferencing determined more country than taxonomic group. Distinct practices are determinant implicit characteristics difficulty specimens. Availability contrasts regions. Uncertainty values not normally distributed but peak at very distinct values, which can be traced back specific regions world. leads wide spectrum range sizes when modelling distributions, potentially affecting conclusions biogeographical climate change studies. In summary, digitised fraction world's NHCs far from optimal terms mainly depends hosted. A collective effort between communities around NHC institutions, research data infrastructure needed bring par with its importance relevance research.

Language: Английский

---