How will climate novelty influence ecological forecasts? Using the Quaternary to assess future reliability

Abstract

Future climates are projected to be highly novel relative to recent climates. Climate novelty challenges models that correlate ecological patterns to climate variables and then use these relationships to forecast ecological responses to future climate change. Here, we quantify the magnitude and ecological significance of future climate novelty by comparing it to novel climates over the past 21,000 years in North America. We then use relationships between model performance and climate novelty derived from the fossil pollen record from eastern North America to estimate the expected decrease in predictive skill of ecological forecasting models as future climate novelty increases. We show that, in the high emissions scenario (RCP 8.5) and by late 21st century, future climate novelty is similar to or higher than peak levels of climate novelty over the last 21,000 years. The accuracy of ecological forecasting models is projected to decline steadily over the coming decades in response to increasing climate novelty, although models that incorporate co-occurrences among species may retain somewhat higher predictive skill. In addition to quantifying future climate novelty in the context of late Quaternary climate change, this work underscores the challenges of making reliable forecasts to an increasingly novel future, while highlighting the need to assess potential avenues for improvement, such as increased reliance on geological analogs for future novel climates and improving existing models by pooling data through time and incorporating assemblage-level information.

An evaluation of transferability of ecological niche models

“…Murray et al (2011) found that algorithms performed well when models were projected to regions adjacent to the calibration area. Additionally, Fitzpatrick et al (2018) found declining performance in Figure 5. Sequeira et al (2016) found better transferability of models when calibration data and evaluation data had similar spatio-temporal scales, matching again our findings of increased model transferability in environmentally matching conditions and poorer performance in non-matching conditions.…”

“…Here, novel has the same meaning as 'novel' in two other published methods accounting for environmental dissimilarity (Elith et al 2011, Owens et al 2013. Euclidean distance or Mahalanobis distance; Fitzpatrick et al 2018), our approach to defining environmental novelty is categorical; we note that extrapolation is defined in statistics as making predictions outside training data, so environmental distance or similarity may not directly help distinguish extrapolation from interpolation; further, even in the case of extrapolation, we may not discriminate between less vs more novel conditions, because both scenarios are extrapolation by definition. Compared with other studies that quantify environmental novelty in a continuous manner (e.g.…”

Land‐use and climatic causes of environmental novelty in Wisconsin since 1890

“…, Fitzpatrick et al. ). To be clear, our new results do not call previous assessments of environmental novelty into question; this paper and those were conducted at different spatial scales (regional vs. global) and temporal domains (historical vs. 21st‐century projections).…”

“…), and high novelty is expected to decrease the predictive strength of ecological forecasting models (Fitzpatrick et al. ). However, species, ecosystems, and land managers are faced with a host of concomitant environmental changes that can drive ecological novelty (Martinuzzi et al.…”

Physiology in ecological niche modeling: using zebra mussel's upper thermal tolerance to refine model predictions through Bayesian analysis

“…Another reason for the relative narrowness of native occurrence data is adaptation of introduced populations to previously limiting conditions in the invaded range (Prentis et al 2008, Moran andAlexander 2014). The poor performance of the native model in the global context is analogous to decreased accuracy of a model calibrated with present climate data and projected to future climate data (Moreno-Amat et al 2015, Fitzpatrick et al 2018, since model extrapolation is involved in both scenarios (Qiao et al 2019). However, the reason for the narrower range of native data does not affect the improvement in physiologically informed models, rather our conclusion is supported by incompleteness of native data and effectiveness of physiological information.…”