Citizen science monitoring demonstrates dramatic declines of monarch butterflies in western North America

Highlights

• The monarch population in western North America is in sharp decline.

• We estimated population trends from spatially and temporally erratic sampling data.

• Current trends indicate an extinction risk of 72% in 20 years and 86% in 50 years.

• Reversing current trends requires achieving higher than historic average growth rates.

Abstract

Count-based PVA allows researchers to assess patterns of population change through time and to evaluate future persistence. We combined state-space models and citizen science data to evaluate viability of the western population of monarch butterflies over 36 years. A key feature of our analysis was combining irregular sampling from multiple sites to obtain a single estimate of total abundance using state-space models. The average population growth rate was negative, u = − 0.0762 (λ = 0.927), average abundance in the 2000s was < 5% of average abundance in the 1980s, and current quasi-extinction risk is 72% within 20 years. Despite wide confidence intervals in some parameter estimates, western monarch monitoring data provide unambiguous evidence for dramatic population declines. To obtain viable populations, managers could target historic abundance and high enough growth rates to avoid near-term extinction.

Introduction

From time to time, widespread species decline in abundance so much that they appear to be at risk of extinction. Assessing such declines in the context of historic observations and yearly fluctuations, however, presents a challenge: Are observed declines sufficient to substantially increase extinction risk? Abundance data from long-term monitoring allow us to quantitatively evaluate this question. Count-based population viability analysis (PVA; Dennis et al., 1991, Morris and Doak, 2002, Fieberg and Ellner, 2000) estimates two parameters from monitoring data: a density-independent annual rate of population growth or decline, and year-to-year variation in this population growth rate, i.e., environmental stochasticity. These parameters, combined with current population size, can be used to predict extinction risk (Morris and Doak, 2002) and evaluate the magnitude of changes needed to ensure persistence (e.g., Molano-Flores and Bell, 2012).

Here, we used count-based PVA to evaluate the current status and future prospects of monarch butterflies (Danaus plexippus plexippus) in western North America. Like many at-risk species, systematic monitoring of this population began after dramatic declines had already been noticed. Therefore, it has been difficult to assess the status of the western monarch population with respect to historic abundance. We addressed this concern by finding appropriate statistical models to integrate irregular sampling during the 1980s and 1990s with more systematic monitoring during the past 20 years. Our case study highlights how modern statistical tools can help us make use of long-term monitoring data collected by citizen scientists for status assessment of at-risk species.

Monarchs, well-known for their distinctive migration from their breeding range to overwintering sites in Mexico and coastal California, were once common throughout most of North America. Recently, the viability of eastern monarchs, which overwinter in Mexico, has received considerable attention (Inamine et al., 2016, Pleasants et al., 2016, Semmens et al., 2016). Western monarchs, which breed west of the Rocky Mountains and are considered a distinct population from eastern monarchs, have been largely ignored in the literature and popular press. Most western monarchs overwinter in wooded groves along coastal California, with limited numbers overwintering in Mexico (Morris et al., 2015, Yang et al., 2016). As with eastern monarch declines in Mexico, changes in California overwintering populations may indicate threats occurring in breeding states, or coastal overwintering habitat loss and degradation, and beg the question of population viability.

For several decades, volunteers have been counting overwintering monarchs in coastal California. The data consist of overwintering butterfly counts in individual groves, each of which represents a subset of the entire breeding population. The vagaries of the data, e.g., year-to-year fluctuations in population size, added or missing sites across years, and count errors, require sophisticated methods of analysis only recently available to ecologists. State-space models account for noisy data by separating observation error from processes of population growth and environmental stochasticity (DeValpine and Hastings, 2002, Holmes et al., 2012). Here, we used state-space models to estimate the western monarch population growth rate from spatially and temporally erratic sampling data, and show how short-term population fluctuations can have long-term consequences in a species of conservation concern.