Pivotal effect of early‐winter temperatures and snowfall on population growth of alpineParnassius smintheusbutterflies

Abstract

Geographic range shifts in species' distributions, due to climate change, imply altered dynamics at both their northern and southern range limits, or at upper and lower elevational limits. There is therefore a need to identify specific weather or climate variable(s), and life stages or cohorts on which they act, and how these affect population growth. Identifying such variables permits prediction of population increase or decline under a changing climate, and shifts in a species' geographic range. For relatively well studied groups, such as butterflies, geographic range shifts are well documented, but weather variables and mechanisms causing those shifts are not well known. The Holarctic butterfly genus Parnassius (Papilionidae) inhabits northern and alpine environments subject to variable and extreme weather. As such, Parnassius species are vulnerable not only to long-term changes in average conditions but especially to short-term extreme weather events. We use population growth estimates for the alpine butterfly, Parnassius smintheus, from 21 populations in the Rocky Mountains of Canada over a 20-yr interval combined with techniques of machine learning (randomForests) and parametric modeling to identify the important weather variables determining population growth. We do this to determine the seasons and life stages of P. smintheus most affected by climate change. Extreme minimum and maximum temperatures in November, in combination with November snowfall, affect annual population growth most, more so than do mean temperatures in November, and more so than weather at any other time of year. Populations decline both in years with low extreme minimum temperatures in November and especially in years with high extreme maximum temperatures in November, indicating that overwintering eggs are particularly vulnerable to early-winter weather. Snowfall ameliorates the negative effects of extreme temperatures, particularly for extreme warm events. Results provide insight into biological mechanisms by which overwintering eggs might be affected by early winter weather. Shortterm extreme weather in November, acting on a single pivotal life stage (egg) is a far better predictor of population change of alpine P. smintheus butterflies than is the general index of climate, the Pacific Decadal Oscillation.

Climate and extreme weather independently affect population growth, but neither is a consistently good predictor

“…Based on the extremely positive values of Pacific Decadal Oscillation (PDO) seen during the winter of 2014, the climate model predicted a large population crash in 2015 (Matter and Roland ). In contrast, the model based on November weather more accurately predicted the observed increases in abundance seen in 2015 (Roland and Matter ). Thus, it remains unclear if the weather model is generally a better predictor than the climate model, or if it was simply better for that year.…”

“…Gaps in the temperature record at this site were supplemented with data from Nakiska Ridge, Alberta (50.94° N, 115.19° W), 17 km from the population study site and 26 km from the Little Elbow site, but at a slightly higher elevation (2540 m). Data are highly correlated among the three sites (Roland and Matter ).…”

“…To compare the explanatory and predictive abilities of the climate index and weather variables on population growth, we began with the climate index‐based model of Roland and Matter () and the weather‐based model (Roland and Matter ). The climate index model was arrived at by relating PDO during the butterflies' different life stages and abundance in sub‐populations, to population growth from one year to the next.…”

“…Here, we compare the ability of a climate‐based model and a weather‐based model to describe and predict changes in yearly population sizes. We have shown that both climatic conditions (Roland and Matter ) and extreme weather (Roland and Matter ) are related to the year‐to‐year population growth ( R t ) of the alpine butterfly Parnassius smintheus . Specifically, Roland and Matter () showed that growth was related to mean climate during the overwintering period and population sizes in the previous year ( N t ).…”

Dispersing maleParnassius smintheusbutterflies are more strongly affected by forest matrix than are females

“…Because females are searching for oviposition sites and lay single eggs, they potentially avoid competition and inbreeding among their offspring by emigrating and distributing their progeny more widely (Rausher, ). Similarly, females may also be selecting a wider range of microclimates to “bet hedge” on the survival of overwintering eggs in warmer or colder winters (Brown & Ehrlich, ; Roland & Matter, ).…”