Discussion
For butterflies in Massachusetts, changes in abundance were most
strongly associated with changes in flight period. Species that have
extended their flight period have increased in abundance relative to
those that have not. Past studies that have evaluated the effect of
warming temperatures have generally demonstrated that species advancing
phenological events under warming environmental conditions are generally
increasing in abundance or have positive demographic growth (Cleland et
al., 2014; Hulme, 2011; Ozgul et al. 2010; Willis et al., 2008; Møller
et al., 2008). Species which fail to change their phenology, however,
appear to be declining (MacGregor et al. 2019; Cleland et al., 2014;
Willis et al., 2008; Møller et al., 2008). In many ways, our results
parallel a recent study of Lepidoptera in the Britain (MacGregor et al.,
2019). In both studies, changes in some phenological metric were
associated with increases in abundance, especially in multivoltine
species. However, in contrast to Britain, trends in the mean timing of
activity of Massachusetts butterflies were not strongly associated with
trends in abundance. Instead, changes in the flight period were the best
predictor of trends in abundance. In Massachusetts, changes in flight
period were driven more by extended late-season activity than advanced
onset of activity, which also contrasts with past work in Britain (Roy
and Sparks, 2000). Our results highlight the importance of evaluating
the entire temporal distribution of a phenological phenomenon, as
opposed to simply using trends in mean observation dates.
One possible mechanistic explanation for the association between
extended flight period and increased abundance is an increase in the
number of generations per year. In other insect species, climate warming
may allow for faster larval development and the production of additional
direct-developing generations before adverse winter conditions
(MacGregor et al., 2019; Miton & Ferrenberg, 2012; Altermatt, 2009;
Tobin et al., 2008). Visual inspection of representative species in our
data suggest that at least some populations are adding, at least, a
partial generation (Fig. 1b, c & e). For example, Satyrdotes
appalachia (Appalachian Brown, Fig. 1b) may be transitioning from
univoltine to bivoltine in our region (Stichter, 2015). If additional
generations are the causal mechanism between changes in flight period
and increased abundance in multivoltine species, it would suggest that
beneficial effects of higher population growth rates (Kerr et al., 2020)
and/or more rapid evolution (Chevin et al., 2010) outweigh the potential
costs of developmental traps (Van Dyck et al., 2015; Levy et al., 2015).
In general, multivoltine butterfly species seem to have higher
population growth rates than univoltine species in north-temperate
climates (MacGregor et al., 2019; Wepprich et al., 2019; Breed et al.,
2013). In one study from a Mediterranean climate, however, multivoltine
species showed significantly steeper declines than obligate uni- or
bivoltine species. In this ecosystem, seasonal adversity in summer
months via drought may have led to a developmental trap (Melero et al.,
2016).
Past studies of butterflies in Massachusetts (Breed et al., 2013) and of
other Lepidoptera (Pöyry, 2009; Parmesan et al., 1999) have revealed
strong patterns of poleward range shifting in response to climate
change. In Massachusetts, southern species are increasing in relative
abundance, and northern species tend to be declining (Breed et al.
2013). Our structural equation model suggests one causal pathway by
which southern species are increasing in relative abundance to some
extent because of their capacity to extend their overall flight period.
Although the SEM supports variability in voltinism as the strongest
driver of trends in abundance, there were also direct effects of
species’ range type on trends in abundance. These effects presumably
reflect other limiting factors such as physiological tolerance or trends
in interacting species. For example, Nymphalis vaualbum (Compton
Tortoiseshell, Fig. 1a) is a northern species that has recently declined
rapidly in the state of Massachusetts. The abrupt shift in abundance
suggests a physiological threshold could have been crossed in the past
decade.
For at least two species in our data, changes in flight period appeared
to be due in part to changes in the ability to overwinter in
Massachusetts, as opposed to voltinism per se. Poanes
zabulon Zabulon skipper (Fig. 1b) and Battus philenor Pipevine
swallowtail (Fig. 1c) were formerly considered vagrants. In other words,
in the past, individuals of these species moved annually to
Massachusetts from populations further south during the summer, but
never established resident populations (Stichter, 2015). It seems likely
that these species can now successfully overwinter in Massachusetts, as
shown by an apparent early season generation (see Fig. 1b & c). These
trends contrast with two other migratory species, Vanessa cardui (Painted Lady, Fig. 1d) and Danaus plexippus (Monarch, Fig. 1f),
both which are non-resident, multi-generational, obligatory migrants,
far north of overwintering sites. Both species have delays (not
advances) in the onset of spring activity in Massachusetts, and only V. cardui showed delayed end of flight activity, while there was
no substantial trend in the end of D. plexippus flight in
Massachusetts. Both species showed negligible changes in relative
abundance in Massachusetts.
Across the community of butterflies in Massachusetts, trends in
abundance estimated from counts, as well as trends from the list length
analysis, were centered on zero. In other words, approximately equal
numbers of Massachusetts species are increasing in abundance as are
declining in abundance. Our results contrast with negative average
trends reported across butterfly communities in four other regions:
Ohio, USA (Wepprich et al. 2019), California, USA (Forister et al.
2011), the UK (Brereton et al. 2018), and Spain (Melero et al. 2016).
Unlike Ohio and California, both of which have agricultural development
at industrial scales, Massachusetts may be subject to lower levels of
agricultural land use intensification. Systematic monitoring rather than ad hoc citizen science observations would be needed to rigorously
test whether butterfly communities in Massachusetts are truly more
stable in abundance when compared to other regions. Nonetheless, our
data suggest a hopeful hypothesis that butterflies may be able to adjust
to climate change, if freed from the additional pressure of
anthropogenic changes in land use.
Our results highlight the potential for phenological changes to affect
population dynamics, and the general importance of phenological
flexibility for insect population dynamics in changing environments.
Past studies of phenology have rarely evaluated shifts in the end of
activity (Zipf al., 2017; Gallinat et al., 2015), despite the
demographic importance of late-season events such as entry into diapause
(Kerr et al. 2020). Unlike traditional metrics of phenological change,
quantile regression (Cade & Noon, 2003) enabled us to estimate trends
in both the onset and end of flight across Massachusetts butterflies. In
Massachusetts, changes in the total flight period and end of activity
were associated more strongly with trends in abundance than onset or
mean trends. In Lepidoptera, a diversity of cues and mechanisms, both
genetic and plastic (Lindestad et al., 2019; Kozak et al., 2019), shape
patterns of phenology and voltinism. Understanding these mechanisms
would be a valuable direction for future research if our goal is to
predict changes in phenology and abundance in new systems. This case
study highlights how phenological changes relate to population dynamics,
and the importance of phenological changes for population viability in
changing environments.