Natural niche truncation hinders accurate prediction of species’ past and future responses to changing climates

Adam B. Smith, Center for Conservation and Sustainable Development, Missouri Botanical Garden, St. Louis, MO and David Henderson, Center for Conservation and Sustainable Development, Missouri Botanical Garden, Saint Louis, MO

Background/Question/Methods
How much are species’ observable niches naturally truncated, and how does this affect the reliability of estimating species’ responses to past and future climates? Niches truncation can arise “naturally” when a species’ range abuts a barrier such as a coastline; if the range contains and occupies a location with an extreme environment; or when dispersal limitation, biotic interactions, or disturbance preclude occupancy of all suitable climates. Truncation can also occur “unnaturally” when inadequate sampling compromises representation of the niche. Regardless of the cause, niche truncation can make predicting species’ responses to past or future conditions unreliable because it requires extrapolating into novel environmental conditions. Here we assess the prevalence of risky niche truncation due to natural dispersal barriers for 44 species in the genus Asclepias (milkweeds) in North America. We measured truncation as the proportion of a species’ niche perimeter that “faced” climates that do not currently exist. Extrapolation is required when a truncated border abuts a future climate that does not currently exist. For truncated sections of niche perimeters, we scored subsections according to whether extrapolations into novel climate space would likely be unreliable or reliable based on local slopes in niche occupancy near the niche periphery.
Results/Conclusions
On average, 35±15% (mean ± SE) of niche perimeters are currently truncated, meaning they abut environmental space that does not currently exist. Of this truncated perimeter, 48±29% faced future climates that do not currently exist but do under RCP 6.0 for the 2070s. Thus, about half of predictions would require extrapolating trends into novel climate space. Extrapolating into future climate space would be highly unreliable for 19±20% of truncated niche perimeters, but nominally reliable for 30±21% (the remainder of truncated perimeter does not “face” future climates so does not require extrapolation). When projecting species’ niches back to 21 Kybp, the ratio of unreliable to reliable extrapolation increases with time from the present. Natural niche truncation was common where species occupy extreme environments on the edge of available niche space and along coastlines. As a result, correlative ecological niche models, mechanistic models that rely on present-day environments for model parameterization, space-for-time studies, and even transplant experiments will be hindered in how well then can ascertain many species’ past and future responses to environmental change. Natural niche truncation cannot be ameliorated by better sampling of existing niches, so reliably predicting past the point of truncation will remain a challenge for ecologists to overcome.