Ecological Indicators 68 (2016) 126133

Contents lists available at ScienceDirect

Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind

An indicator highlights seasonal variation in the response of

Lepidoptera communities to warming
B. Martay a, , D.T. Monteith b , M.J. Brewer c , T. Brereton d , C.R. Shortall e ,
J.W. Pearce-Higgins a
a

British Trust for Ornithology, BTO, The Nunnery, Thetford, Norfolk IP24 2PU, United Kingdom
CEH Lancaster, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster, LA1 4AP, United Kingdom
c
Biomathematics & Statistics Scotland, BioSS Ofce, The James Hutton Institute, Craigiebuckler, Aberdeen, AB15 8QH, Scotland, United Kingdom
d
Buttery Conservation, Manor Yard, East Lulworth, Wareham, Dorset, BH20 5QP, United Kingdom
e
Rothamsted Research, West Common, Harpenden, Herts, AL5 2JQ, United Kingdom
b

a r t i c l e

i n f o

Article history:
Received 28 May 2015
Received in revised form 24 January 2016
Accepted 27 January 2016
Available online 5 March 2016
Keywords:
Climate change
Community composition
Indicator
Lepidoptera
Seasonal
Temperature

a b s t r a c t
The impacts of climate change on species and ecosystems are increasingly evident. While these tend
to be clearest with respect to changes in phenology and distribution ranges, there are also important
consequences for population sizes and community structure. There is an urgent need to develop ecological
indicators that can be used to detect climate-driven changes in ecological communities, and identify how
those impacts may vary spatially. Here we describe the development of a new community-based seasonal
climate change indicator that uses national population and weather indices. We test this indicator using
Lepidopteran and co-located weather data collected across a range of UK Environmental Change Network
(ECN) sites. We compare our buttery indicator with estimates derived from an alternative, previously
published metric, the Community Temperature Index (CTI).
First, we quantied the effect of temperature on population growth rates of moths and butteries
(Species Temperature Response, STR) by modelling annual variation in national population indices as
a function of nationally averaged seasonal variation in temperature, using species and weather data
independent of the ECN data. Then, we calculated average STRs for annually summarised species data
from each ECN site, weighted by species abundance, to produce the Community Temperature Response
(CTR). Finally, we tested the extent to which CTR correlated with spatial variation in temperature between
sites and the extent to which temporal variation in CTR tracked both annual and seasonal warming trends.
Mean site CTR was positively correlated with mean site temperature for moths but not butteries.
However, spatial variation in moth communities was well explained by mean site summer temperature
and buttery communities by winter temperature, respectively accounting for 74% and 63% of variation.
Temporal variation in moth and buttery CTR within sites did not vary with the mean annual temperature
but responded to variation in the mean temperature of specic seasons. There were positive correlations
between moth seasonal CTRs and seasonal temperatures in winter, spring and summer; and buttery
seasonal CTRs and seasonal temperatures in winter and summer. Buttery CTR and CTI both correlated
spatially and temporally with winter temperature.
Our results highlight the need for seasonality to be considered when examining the impact of climate
change on communities. Seasonal CTRs may be used to track the impact of changing temperatures on
biodiversity and help identify potential mechanisms by which climate change is affecting communities. In the case of Lepidoptera, our results suggest that future warming may reassemble Lepidoptera
communities.
2016 Elsevier Ltd. All rights reserved.

1. Introduction

Corresponding author. Tel.: +44 01786 466557.

E-mail address: blaise.martay@bto.org (B. Martay).
http://dx.doi.org/10.1016/j.ecolind.2016.01.057
1470-160X/ 2016 Elsevier Ltd. All rights reserved.

The impacts of climate change on species and ecosystems are

increasingly evident (Parmesan and Yohe, 2003; Pearce-Higgins
and Green, 2014). Polewards and uphill range expansions have
been seen across many taxa, accompanied by range retractions at
warm edges (Franco et al., 2006; Hickling et al., 2006; Chen et al.,

B. Martay et al. / Ecological Indicators 68 (2016) 126133

2011). Community changes have been widely observed: generalist

species are increasing in abundance relative to specialists (Warren
et al., 2001; Davey et al., 2012; Le Viol et al., 2012), whilst warmassociated species have tended to increase relative to those that
occupy cooler climates (Devictor et al., 2008, 2012; Jiguet et al.,
2010). To minimise biodiversity loss, there is a considerable need
to identify the species most likely to be vulnerable to climatic
changes (Pacici et al., 2015). These are likely to include: species
with less opportunity or ability to adapt through phenological
plasticity (Vedder et al., 2013), habitat specialists (Davey et al.,
2013), species sensitive to climatic impacts on other species with
which it interacts strongly (Ockendon et al., 2014), and species with
a dispersal barrier polewards of their range extent (Foden et al.,
2013). There is also a need to predict where climate change will
cause the greatest ecosystem disruption.
Tracking the impacts of climate change on species and ecological
communities is complex and challenging, as it potentially involves
effects of multiple climatic, and other, drivers operating through
a range of mechanisms. Despite this, a number of summary indicators have been developed that appear to separate species by
their likely response to warming or climate change, and that can
be used to track the potential impacts of climate change on ecological communities (Thuiller et al., 2005; Devictor et al., 2008;
Gregory et al., 2009). For example, the Species Temperature Index
(STI) describes the mean temperature across a species range, and
is used to derive the Community Temperature Index (CTI), which
is the mean STI across species present at a site, weighted by their
abundance. Changes in CTI through time have been shown to indicate long-term community responses to warming in both birds and
butteries (Devictor et al., 2008, 2012). Alternatively, individual climate envelope models have been used to separate species by their
likely response to climate change. The divergence in population
trends between species projected to increase in response to climate change and those projected to decline then provides another
potential climate change indicator (Gregory et al., 2009). In these
studies, and here, a community is dened as the collection of individuals present at a site within a taxonomic group. Both methods
use spatial information about species distributions to describe the
association of that species with climate, and therefore they are
limited to taxonomic groups with well-understood spatial distributions such as birds, butteries and plants. Furthermore, climatic
niche models have a number of problems which may limit their
ability to predict and indicate climate change impacts (Pearson and
Dawson, 2003; Hampe, 2004). In particular, models based on a climatic envelope take little account of interactions between species
or their ability to shift range (Pearson and Dawson, 2003).
Here we trial an alternative method that characterises species
responses to temperature by modelling temporal variation in
annual abundance as a function of temperature (the Species Temperature Response; STR), rather than using spatial distribution. This
allows metrics of community change to be derived from the average STR at a site in a year, weighted by relative abundance (the
Community Temperature Response; CTR), in a manner similar to
Devictor et al. (2008). Further, this method may also be adapted to
summarise responses across different seasons and thereby provide
an indication of the times of year when species respond most to
climate change impacts, which may be helpful in identifying the
potential mechanism underpinning observed community changes.
If successful, this method may be used where there is long-term
temporal monitoring at any spatial scale (results could be obtained
using only a single site), but relatively little information on species
range extents. UK examples of taxa which meet these criteria
include moths, aphids and bats, and this indicator may therefore
be widely applicable.
In this paper, we used national monitoring data to calculate
STR for a range of moth and buttery species. We then tracked

127

changes in the communities of both groups across a network of

sites with biological and environmental recording that included
meteorological data (Environmental Change Network sites; ECN).
We assessed the ability of the CTR metric to predict both spatial and temporal variation in annual and seasonal temperatures.
Agreement between the CTR metric and temporal temperature
variation would indicate rapid responses to climate change, for
example, caused by direct effects on demographic rates or species
interactions. Agreement between the CTR metric and spatial temperature variation would indicate that temperature is an important
determinant of species current spatial distribution and would thus
provide an indication of the sensitivity of communities to longterm climate change. To be effective, an indicator must be able
to measure rapid responses to temporal temperature variation,
and provide evidence that these will lead to long-term community
changes as indicated by spatial patterns in CTR. We compared the
sensitivity of moth and buttery communities to warming by comparing the strength of relationship between the moth and buttery
CTR indicators and spatial and temporal temperature variation.
Finally, we also compared the ability of buttery CTR to spatially
and temporally track temperature with that of the already published buttery CTI (Devictor et al., 2012).
2. Material and methods
2.1. National data
We used national population indices to characterise the extent
to which a species population was positively or negatively correlated with temperatures (species temperature response, STR). Data
required to calculate the STRs were taken from two national monitoring schemes, one for butteries and one for moths. Buttery data
were obtained from the UK Buttery Monitoring Scheme (UKBMS).
We only used buttery data from sites in England because of a bias
towards England in the early years of the survey compared with
the current distribution of sites, which are more widespread. This
restriction reduced the chance of temporal and spatial bias affecting the national trends. There are over 1000 English sites monitored
in the UKBMS. The moth data were derived from thirteen longrunning sites monitored by the Rothamsted Insect Survey (RIS).
Most of these sites are in England and Wales (see Supplementary
Information (SI) Fig. 1 for a map of the RIS moth traps). We only
used data from the thirteen sites that were consistently monitored
from 1978, hence avoiding the need to consider temporal biases in
the data.
Using these data, we calculated population indices for butteries from 1976 to 2011, and for moths from 1975 to 2010 using
TRIM (Pannekoek and van Strien, 1998). Annual counts at sites were
modelled as a function of both categorical site and year effects
using a Poisson General Linear Model (GLM) with a standard log
link accounting for serial correlation and overdispersion (Ter Braak
et al., 1994; Freeman and Newson, 2008). Only species with at least
one record per year were included, allowing us to calculate STR
values for 46 buttery and 265 moth species (see SI Table 1).
We related the national population indices to nationally averaged weather data (see Section 2.2). For this we used data from
Met Ofce UK Climate Projections 2009 (UKCP09) gridded monthly
datasets averaged across England (for butteries) and the UK (for
moths) (Perry and Hollis, 2005).
2.2. Species temperature response
We used a simple model to characterise each species temperature response (STR), i.e. the extent to which their populations
were positively or negatively correlated with mean seasonal

128

B. Martay et al. / Ecological Indicators 68 (2016) 126133

temperatures throughout a year. Annual population growth was

calculated from the national population indices (yt = ln (nt /nt1 ),
where nt is the annual national population index in year t).
Population growth was modelled against annual variation in seasonal temperatures (throughout, we dene seasons as follows:
Winter, DecFeb; Spring, MarMay; Summer, JunAug; Autumn,
SeptNov), in order to account for potentially contrasting responses
to temperature at different times of the year. In order to standardise these responses, we converted temperatures into anomalies (i.e.
the difference between each mean seasonal temperature and mean
temperature in that season across all years in time-series). Four
seasonal temperature anomalies (x1 x4 ) were included as explanatory variables: season 1 was the season that the species was active
in and seasons 24 were the three preceding seasons. Thus, if a
species adult activity period was from July to October then x1 x4
would respectively be the mean autumn, summer, spring and winter temperature anomalies in year t. For moth species, their period
of activity ended in winter for 2.6% of species, 9.8% in spring, 60%
in summer and 27.6% in autumn. For buttery species, 63.6% had
activity periods that ended in summer and 36.4% in autumn.
Thus we modelled:
yt = a + b1 x1 + b2 x2 + b3 x3 + b4 x4 + ct + fyt1 + et

(1)

in which a is the intercept, b1 to b4 are the regression coefcient for

the temperature covariates in the four seasons, x1 to x4 are the mean
seasonal temperature anomalies, c is the regression coefcient of
a year of survey covariate t and f is the regression coefcient of a
covariate of log population abundance in the previous year (allowing for density dependent growth) and et is a Normally distributed
error term.
We dened a species four seasonal STRs as the coefcients b1
to b4 . The annual STR was taken to be the sum of seasonal STRs
and indicated the change in each species log population abundance
predicted to occur if the temperature rose by 1 C, increasing evenly
across all seasons. Cold-adapted species will thus have a low STR
while warm-adapted species will have a high STR. See the SI Table
1 for a list of species and STR.
2.3. Species temperature index
We also characterised buttery species associations with temperature using the species temperature index (STI) developed by
Devictor et al. (2008). STI was designated as the long-term average temperature of each species range, combining data from
the European atlas of butteries (Settele et al., 2008) with mean
MarchSeptember temperature across each species European
range (Devictor et al., 2012).
2.4. Environmental change network data
Most terrestrial ECN sites were established in 1992 and their
dominant physical characteristics are described by Sier & Monteith,
2016. ECN sites span a wide range of altitudes and hence mean
air temperatures, rendering them particularly useful in assessing
performance of temperature indicators across a relatively wide climatic gradient. Butteries and moths were monitored at twelve
ECN sites (see SI Fig. 1 for map) and we used ECN Lepidoptera
data from 1993 to 2012 to test the CTR and CTI indicators. Full
descriptions of the methods used are available from the ECN
website (moths: ECN, 2015a; butteries: ECN, 2015b). Moth monitoring used light-trap methods established by the Rothamsted
Insect Survey (RIS), trapping moths every night throughout the
year (Woiwod, 1996b; Conrad et al., 2007). Buttery monitoring followed a transect-based method established by the UKBMS
(Woiwod, 1996a): transects of 12 km were established to be representative of the ECN site and were walked weekly between 1 April

and 29 September (at set times and providing certain weather conditions, dened by the protocol, were met). Butteries seen within
5 m of the transect were recorded. Counts of 47 buttery species
(Rennie et al., 2015a) and 552 species of moth (Rennie et al., 2015b)
were recorded on these sites although only species for which we
could produce national population indices were included in the
analysis (see Section 2.1).
In addition to biological recording, dry bulb air temperature was
also recorded at co-located automatic weather stations at each site.
Temperature measurements were made at 5 s intervals and stored
within the ECN database as hourly means. We used these data
(Rennie et al., 2015c) to calculate monthly mean temperatures from
which we then derived seasonal and annual mean temperatures.
In order to relate temperature measurements to the ECN biological
data, annual mean temperatures were calculated from October in
the year prior to sampling to September, the nal month of buttery monitoring. Relatively few moth species are particularly active
in their adult stage beyond September.
Moth trapping was not carried out at one of the ECN sites (Snowdon: 53.07 N, 4.03 W) due to the absence of a secure power supply,
whilst on one site (Cairngorms: 57.12 N, 3.83 W) buttery monitoring only took place for four years, so this site was excluded from
the buttery data. In addition, in this analysis we excluded moth
data from two sites (Alice Holt and Rothamsted) because data from
these sites were included in the 13-site dataset that was used to create the national moth indices which we used to characterise species
response to temperature (SI Fig. 1). Thus data from nine and eleven
ECN sites were used respectively for moths and butteries.
We modelled annual site temperature and annual seasonal temperatures against year using Linear Mixed Models (LMMs) with site
included as a random factor.
2.5. Spatial variation in communities
At each of the ECN sites, we calculated the annual moth and
buttery CTR and buttery CTI from the annual mean STR of
moth and buttery individuals and the annual mean STI of buttery individuals (Devictor et al., 2008). We modelled the mean
moth and buttery CTR and buttery CTI at each site across all
years against the mean site temperature using linear models. We
repeated this for mean seasonal CTRs and seasonal mean site temperatures and modelled buttery CTI against seasonal mean site
temperatures.
2.6. Temperature-related community change within sites
We examined the temperature-related community changes by
regressing changes in CTR with temperature. We used LMMs to
model the moth and buttery CTR anomaly (CTRA = CTR mean site
CTR) against mean annual temperature anomaly (x = mean annual
temperature mean site temperature):
CTRA = a + bx

(2)

where a is the intercept and b is the regression coefcients for the

mean annual temperature anomaly. Site was included as a random factor and we used a method developed by Nakagawa and
Schielzeth (2013) to identify the variation explained by xed factors
in mixed models. We repeated this for seasonal CTR anomalies and
seasonal temperature anomalies. We also repeated this for buttery
CTI and annual and seasonal temperatures.
We tested whether there were long-term trends in community composition driven by temperature change. This was achieved
using the annual CTR, as this provided a single assessment of the
combined impacts of seasonal temperature changes and community responses. We modelled the change in moth and buttery CTR

B. Martay et al. / Ecological Indicators 68 (2016) 126133

129

Table 1
The parameter estimates, standard errors, r2 and P-values of the relationship between mean annual and seasonal moth and buttery CTR, annual buttery CTI and mean site
temperatures.
Temperature

Moth CTR

Buttery CTR

Buttery CTI

Annual

0.026 0.007
r2 = 0.662, P = 0.008
0.013 0.004
r2 = 0.650, P = 0.009
3.52 103 5.80 103
r2 = 0.050, P = 0.564
0.017 0.003
r2 = 0.839, P = 0.001
5.05 103 2.26 103
r2 = 0.007, P = 0.830

4.21 104 1.40 103

r2 = 0.010, P = 0.771
1.92 103 4.39 104
r2 = 0.681, P = 0.002
3.17 104 1.62 103
r2 = 0.004, P = 0.849
1.90 103 1.17 103
r2 = 0.226, P = 0.139
1.49 103 5.87 104
r2 = 0.366, P = 0.049

0.174 0.049
r2 = 0.581, P = 0.006
0.177 0.066
r2 = 0.443, P = 0.025
0.161 0.047
r2 = 0.562, P = 0.008
0.149 0.038
r2 = 0.625, P = 0.004
0.187 0.051
r2 = 0.601, P = 0.005

Winter
Spring
Summer
Autumn

and buttery CTI over time with site as a random factor using LMMs
as above.
3. Results
3.1. Temperature trends over time
Across sites, mean annual air temperature did not change significantly (temperature change = 6.34 103 5.91 103 C year1 ,
t = 1.07, P = 0.285). Winter temperatures were largely stable
(0.021 0.013 C year1 , t = 1.64, P = 0.103) whilst summer temperature declined during the study duration (0.019 0.009,
t = 2.21, P = 0.028). Spring and autumn temperatures increased
(spring temperature change = 0.040 0.008 C year1 , t = 5.05,
P < 0.001; autumn temperature change = 0.026 0.010, t = 2.50,
P = 0.013). The seasonal temperature range across ECN sites and
years were similar across all seasons: ranges in winter, spring,
summer and autumn temperatures were respectively 9.2 C, 8.6 C,
9.1 C and 9.1 C.
3.2. Spatial variation in communities
Mean moth CTR and buttery CTI metrics for each site were
positively correlated with mean site temperature while buttery
site CTR did not vary signicantly with site temperature (Table 1,
Fig. 1: black lines). Spatial variation in moth community composition, indicated by mean site summer and winter CTRs, was related
to mean site summer and winter temperatures respectively while
for butteries, winter and autumn CTRs were related to mean site
winter temperature and autumn temperatures (Table 1). Buttery
CTI correlated positively with all seasonal temperatures. A maximum of 83.9% and 68.1% of spatial variation in moth and buttery
communities based on CTR could be explained by mean seasonal
temperatures, respectively in summer and winter (Table 1, Fig. 2:
black lines). A maximum of 62.5% of variation in buttery CTI could
be explained by variation in summer temperatures.
3.3. Temperature-related community change within sites
We assessed the ability of the CTR indicator to track annual
variation in temperature by correlating CTR anomalies with temperature anomalies. Moth and buttery CTR and buttery CTI did
not vary signicantly with annual site temperature, although P < 0.1
for moth CTR and buttery CTI (Table 2, Fig. 1: coloured lines).
We also examined the relationship between seasonal CTR
anomalies and seasonal temperature anomalies. We found that
moth winter, spring and summer CTR varied signicantly and
positively with respective seasonal temperatures. Buttery winter
and summer CTR varied signicantly and positively with winter
and summer temperatures while buttery CTI varied positively
with winter temperatures (Table 2, for plots of all seasonal CTRs
and annual CTI varying with seasonal temperatures see SI Fig. 2).

A maximum of 10.5% and 7.2% of temporal variation in moth and

buttery CTR could be explained by variation in seasonal temperatures, respectively spring and summer temperatures (Fig. 2,
coloured lines). A maximum of 2.8% of variation in buttery CTI
could be explained by winter temperature.
Over time, moth CTR remained stable (1.04 103
8.26 104 , t = 1.26, P = 0.21), while buttery CTR and CTI
declined (CTR: 2.98 104 6.27 105 , t = 4.75, P < 0.001;
CTI: 1.48 104 5.32 105 , t = 2.79, P = 0.006).
4. Discussion
4.1. Indicating the impact of climate change in Lepidoptera
communities
We tested the ability of a novel indicator of climate change
impacts, Community Temperature Response (CTR) to identify
temperature-related Lepidoptera community changes. The CTR
indicator calculated from correlations between communities and
annual temperatures, did not track temporal variation in temperatures from Lepidoptera communities (Table 2, Fig. 1) or spatial
variation in temperatures from buttery communities. It was possible to predict mean site temperature from moth annual CTR across
a mean annual temperature gradient of 4.710.8 C (Table 1), indicating that northern or upland moth communities associated with
colder climates consist of species whose populations have more
negative relationships with temperature (negative STR), when integrated across seasons, than communities found in warmer climates.
This is consistent with previous results for birds suggesting that
spatial variation in bird population trends can be linked to climate
(e.g. Jiguet et al., 2010), but is to our knowledge the rst time this
has been identied in moths. The lack of equivalent relationship for
butteries may suggest a weaker relationship between their distribution and population responses to temperature (see also Palmer
et al., 2015), although we found that buttery communities occupying colder sites did contain more cold-associated species, as
measured by STI.
There were signicant spatial associations between temperature and seasonal CTR indicators for moths in winter and summer,
and butteries in winter and autumn. Mean winter temperature
at sites could account for 68.1% of the spatial variation in buttery
winter CTR and the mean summer temperature could account for
83.9% of spatial variation in moth summer CTR (Table 1, Fig. 2). This
may suggest that moth and buttery communities, rather than representing the summed effect of warming throughout the year, may
be particularly driven by variation in species responses to winter
and summer, and winter and autumn temperature respectively.
Temporal uctuations in these seasonal indicators partially
support this inference, with annual uctuations in seasonal CTR
indicators for both butteries and moths positively correlated with
summer and winter temperatures, and for moths also with spring
temperature (Table 2, Fig. 2). Thus, annual variation in winter

130

B. Martay et al. / Ecological Indicators 68 (2016) 126133

Fig. 1. (a) Moth and (b) buttery annual Community Temperature Response (CTR) and (c) buttery Community Temperature Index (CTI) plotted against annual average
temperature across ECN sites between 1993 and 2012. Each site is shown with a unique symbol and colour combination: the colour indicates the average temperature at the
site with the coldest site shown in dark blue and the warmest site in red. The black crosses show the average site temperature and average site CTR (or CTI) and the black
lines show the between-site relationship between them. The coloured lines show the within-site relationship between CTR/CTI and temperature. (For interpretation of the
references to colour in this gure legend, the reader is referred to the web version of this article.)

Fig. 2. Seasonal community temperature responses (CTR) in (a) summer for moths and (b) winter for butteries and (c) buttery CTI, plotted against mean summer and
winter temperature across ECN sites between 1993 and 2012. The seasons shown here are the seasons that explained most temporal and spatial variation in moth and
buttery community composition. Each site is shown with a unique symbol and colour combination: the colour indicates the average temperature at the site with the coldest
site shown in dark blue and the warmest site in red. The black crosses show the average seasonal site temperature and average seasonal site CTR and the black lines show the
between-site relationship between them. The coloured lines show the within-site relationship between seasonal CTR and temperature. (For interpretation of the references
to colour in this gure legend, the reader is referred to the web version of this article.)

Table 2
The parameter estimates, standard errors, r2 values and P-values of the relationship between annual and seasonal moth and buttery CTR anomalies, buttery CTI anomaly
and annual and seasonal temperature anomalies.
Temperature
Annual
Winter
Spring
Summer
Autumn

Moth CTR
0.020 0.010
r2 = 0.022, P = 0.058
5.36 103 1.93 103
r2 = 0.046, P = 0.006
0.019 0.004
r2 = 0.1058, P < 0.001
0.015 0.004
r2 = 0.070, P < 0.001
2.43 103 2.41 103
r2 = 0.002, P = 0.552

temperature is clearly correlated with both spatial and temporal

variation in CTR in both groups, as is summer temperature for
moths. Interestingly, however the slope of the relationship between
CTR and temperature in only equivalent in space and time with
respect to the summer CTR for moths; for winter temperature there

Buttery CTR
4

Buttery CTI
3

6.99 10 7.95 10
r2 = 0.004, P = 0.380
6.53 104 1.66 104
r2 = 0.072, P < 0.001
5.40 104 4.53 104
r2 = 0.007, P = 0.234
2.96 103 8.60 104
r2 = 0.054, P = 0.001
8.97 105 3.87 104
r2 < 0.001, P = 0.817

0.052 0.029
r2 = 0.015, P = 0.082
0.032 0.013
r2 = 0.028, P = 0.018
6.10 103 1.93 102
r2 < 0.001, P = 0.752
1.36 103 1.88 102
r2 < 0.001, P = 0.942
3.44 103 1.62 102
r2 < 0.001, P = 0.846

is a much stronger spatial than temporal gradient. The effects of

spring temperature on moth communities and summer temperature upon buttery communities occurred in the absence of any
signicant spatial relationship between CTR and climate. Further,
these seasonal CTRs accounted for relatively small amounts of

B. Martay et al. / Ecological Indicators 68 (2016) 126133

variation in community composition through time: seasonal temperatures accounted for a maximum of 10.5% of variation in moth
communities and 7.2% of variation in buttery communities.
These results suggest that different processes govern spatial and
temporal variation in buttery and moth populations, although
as the results for summer temperature and moths, and winter
temperature and both groups show, they can be linked. It is also
worth noting that despite the signicant seasonal correlations with
temperature, our models explained only a small proportion of variation in community composition through time.
There was no signicant long-term trend in moth annual CTR,
whilst the buttery annual CTR trend was signicantly negative.
It is likely that these results are a combination of the complex interactions between seasonal CTR and variable temperature
trends between seasons, potentially exacerbated by unknown nonclimatic factors also driving these trends (Clavero et al., 2011;
Barnagaud et al., 2012). Thus, summer temperature declined in
the period we examined while winter temperatures remained stable, which, based on the seasonal CTR responses for butteries,
could account for the counter-intuitive decline in buttery CTR
through time. Although there was no signicant long-term trend
in moth CTR, given contrasting spring and summer temperature
trends through the time-period studied, both of which affected
annual variation in CTR, this is again to be expected. These results
suggest that both moth and buttery communities are likely to
undergo signicant future long-term changes in response to predicted future warming across all seasons, which, in the presence of
a strong warming trend, we would expect our measures of CTR to
track, at least in some seasons.
4.2. Assessing CTR as a climate change indicator
In this study we compared the ability of our CTR indicators
to track buttery community change in response to temperature
with CTI, a climate change indicator based on a species spatial association with temperature (Devictor et al., 2008, 2012).
There was a strong correlation between CTI and all measures of
spatial variation in temperature (Table 1), which is unsurprising
given the indicator is based upon the spatial association of species
with spring and summer temperature. Neither indicator tracked
annual variation in temperature, but both indicators tracked winter temperatures effectively, both spatially (Table 1) and temporally
(Table 2), although as already outlined, the spatial association was
stronger than the temporal (Fig. 2b and c). Both indicators appear to
track temporal variation in temperature with broadly similar levels
of accuracy although winter temperature could explain marginally
more spatial and temporal variation in buttery CTR (respectively
explaining 68.1% and 7.2% of variation) than CTI (respectively 44.3%
and 2.8%). This suggests that our measures of CTR may be a useful addition to the range of climate change indicators available,
although further validation is required.
Our method for quantifying STR relies on the assumption that
populations respond linearly to temperature. We would not, however, expect a linear response across the full climatic range of
most species; populations of a species are more likely to correlate positively and negatively with temperature at the cold and
warm edges of its range respectively. For many species, the range
of temperature variation across the UK may be small enough to
only cover a portion of its climatic range and thus be adequately
modelled with a linear relationship between population and temperature. However, our models would identify species with small
climatic ranges centred in the UK as unresponsive to climate. This
is likely to reduce the sensitivity of the indicator, and will limit
the spatial extent in which this indicator can be used. If STRs were
calculated using continental-scale data then many temperaturesensitive species would be classied as unresponsive. Alternative

131

methods to quantify CTR using non-linear models would need to

be considered in order to use this CTR indicator across continental
scales.
It is worth noting that we modelled species responses to temperature with a density dependence covariate included in the
model. This means that we did not distinguish between species
equally sensitive to temperature but with differing density dependence relationships. We would expect species with a high density
dependent component to their models to have shallower long-term
population trends than those with low density dependence. The
density dependence relationship is likely to vary spatially and temporally: as ranges shift, populations will face extinction at the warm
edges of their range irrespective of average density dependence
relationships. Further research into the effect of density dependence on the predicted population changes from the STR models is
required. The CTI method also does not take density dependence
into account.
4.3. Mechanisms of seasonal temperature impact
Our results highlight the importance of variation in seasonal
temperatures in inuencing moth and buttery population uctuations, and suggest that warming during the winter may have the
greatest impact upon buttery species and a signicant impact on
moth species. It should be noted that these results could be an artefact of the time period analysed and sites considered and further
research across a wider range of sites and periods would be benecial. However, the temperature variation in winter across sites and
years was very similar to other seasons (see Section 3.1) so there
was no evidence that our results are unrepresentative. A range
of mechanisms may drive this community response; species may
have differential tolerance to cold and warm winters. Stuhldreher
et al. (2014) and Conrad et al. (2002) found that mortality increased
and populations declined in Erebia medusa and Arctia caja respectively after warm winters, particularly in the lowlands. Declines
could be due to increased metabolism, reducing winter energy
reserves (Mercader and Scriber, 2008), or increased vulnerability
to pathogen attack, such as from fungi. Our results suggest that a
response to winter temperature may be seen more widely across
Lepidoptera communities but further investigation is required.
Moth communities were also inuenced by summer temperature. Many studies have identied a positive relationship between
Lepidoptera population abundance and summer temperature (Roy
et al., 2001; Pyry et al., 2011). Temporal variation in buttery
CTR was inuenced by summer temperature, while spatial variation was not. Conversely, spatial variation in buttery CTI was
inuenced by summer temperature, while temporal variation was
not. High temperatures may increase buttery activity which could
lead to false positive correlations between summer temperatures
(when most butteries were active) and populations, and therefore
an inated summer STI. Where this is the case we would expect a
positive temporal relationship between summer CTR and summer
temperature without a spatial relationship (as activity increases do
not necessarily lead to population increases). This was reected in
our results. CTI calculations are not inuenced by summer activity
levels, so the positive spatial correlation between CTI and summer
temperature may reect a real population effect of summer temperature on buttery communities. If we assume that species with
populations that increase in warm years also generally have high
STIs, then a temporal correlation between CTI and summer temperature may be prevented because of increased activity in some
species which do not show increases in population abundance and
thus do not have high STIs. The moth indicators appear not to be
affected by this problem. Our results indicate that more research
is required to disentangle the effect of summer temperature on
buttery activity and population abundance.

132

B. Martay et al. / Ecological Indicators 68 (2016) 126133

Temporal variation in moth communities was also inuenced

by spring temperature and a number of mechanisms relating
Lepidopteran population abundance to spring temperature have
been proposed. High spring temperature can lead to earlier emergence (Sparks and Yates, 1997) and thus population increases
due to increased voltinism (Altermatt, 2010). It can also lead to
an increased metabolic rate, draining energy reserves, and hence
smaller, poorer quality adults (Mercader and Scriber, 2008). Further
research into the mechanisms and timing behind the temperaturedriven community changes in Lepidoptera is required.
4.4. Implications
Our results highlight the effect of temperature on Lepidoptera
communities. Although contrasting seasonal temperature trends
on ECN sites appear to have masked long-term annual changes in
community metrics relative to annual temperature, it is clear that
annual variation in winter temperature affected buttery communities, and winter and summer temperatures similarly had impacts
upon moth communities. Whilst we found evidence that temperature trends inuenced Lepidoptera communities at these sites,
we do not yet know the magnitude of these impacts in driving
populations, relative to other factors such as land-use change. Butteries may currently be more inuenced by habitat factors than
climate change in Britain (Oliver et al., 2012), as found for farmland
birds (Eglington and Pearce-Higgins, 2012), while moths have similarly been identied as particularly threatened in the agriculturally
intensive lowlands (Fox et al., 2014). Further work is required to
quantify the role of climate change in driving long-term population trends in both species groups, particularly given the highly
signicant long-term decline in moth populations that has been
observed in southern Britain (Conrad et al., 2002). This is particularly the case given the relatively weak contribution of temperature
to annual variation in both CTR and CTI measures. There is mounting evidence that population declines of climate-vulnerable species
can be mitigated through habitat protection and creation (Oliver
et al., 2013; Virkkala et al., 2014). This CTR method could be used
to examine spatial variation in climate change impacts, enabling
future research into the interactions between habitat and climate
change impacts, which could inform climate change adaptation
strategies.
Butteries and birds have been the predominant taxa used to
track the impacts of climate change on ecological communities
(Thomas, 2005; Gregory et al., 2009; Davey et al., 2013). On the basis
of our results, we suggest that moths should also be considered
potential candidate indicators of climate change impacts on biodiversity, and may indeed be more sensitive than these other groups.
Similar CTR-based approaches applied to other taxa for which longterm temporal population data exist would enable other species to
be identied, where the effects of temperature on their population
growth may be used to track community changes in response to
warming.
Acknowledgements
We would like to thank Lorna Sherrin for providing Environmental Change Network data and advising on its use, and the site
managers of the ECN terrestrial sites for initial generation of the
biological and meteorological data. We would also like to thank
the British Trust for Ornithology for funding part of this work.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.ecolind.2016.
01.057.

References
Altermatt, F., 2010. Climatic warming increases voltinism in European butteries
and moths. Proc. R. Soc. B: Biol. Sci. 277, 12811287.
Barnagaud, J.Y., Devictor, V., Jiguet, F., Barbet-Massin, M., Le Viol, I., Archaux, F., 2012.
Relating habitat and climatic niches in birds. PLos ONE 7, e32819.
Chen, I.-C., Hill, J.K., Ohlemller, R., Roy, D.B., Thomas, C.D., 2011. Rapid range shifts of
species associated with high levels of climate warming. Science 333, 10241026.
Clavero, M., Villero, D., Brotons, L., 2011. Climate change or land use dynamics: do we know what climate change indicators indicate? PLoS ONE 6,
e18581.
Conrad, K.F., Woiwod, I.P., Perry, J.N., 2002. Long-term decline in abundance and
distribution of the garden tiger moth (Arctia caja) in Great Britain. Biol. Conserv.
106, 329337.
Conrad, K.F., Fox, R., Woiwod, I.P., 2007. Monitoring biodiversity: measuring longterm changes in insect abundance. In: Stewart, A.J.A., New, T.R., Lewis, O.T. (Eds.),
Insect Conservation Biology. CABI Publishing, Wallingford, UK, pp. 203225.
Davey, C.M., Chamberlain, D.E., Newson, S.E., Noble, D.G., Johnston, A., 2012. Rise of
the generalists: evidence for climate driven homogenization in avian communities. Glob. Ecol. Biogeogr. 21, 568578.
Davey, C.M., Devictor, V., Jonzn, N., Lindstrm, ., Smith, H.G., 2013. Impact of climate change on communities: revealing species contribution. J. Anim. Ecol. 82,
551561.
Devictor, V., Julliard, R., Couvet, D., Jiguet, F., 2008. Birds are tracking climate warming, but not fast enough. Proc. R. Soc. B: Biol. Sci. 275, 27432748.
Devictor, V., van Swaay, C., Brereton, T., Brotons, L., Chamberlain, D., Heliola, J., et al.,
2012. Differences in the climatic debts of birds and butteries at a continental
scale. Nat. Clim. Change 2, 121124.
ECN, 2015a. ECN moth recording protocol, http://www.ecn.ac.uk/measurements/
terrestrial/i/im (accessed 18.11.15).
ECN, http://www.ecn.ac.uk/measurements/terrestrial/i/ib (accessed 18.11.15)
2015b. ECN buttery recording protocol.
Eglington, S.M., Pearce-Higgins, J.W., 2012. Disentangling the relative importance
of changes in climate and land-use intensity in driving recent bird population
trends. PLoS ONE 7, e30407.
Foden, W.B., Butchart, S.H.M., Stuart, S.N., Vi, J.-C., Akcakaya, H.R., Angulo, A., et al.,
2013. Identifying the worlds most climate change vulnerable species: a systematic trait-based assessment of all birds, amphibians and corals. PLOS ONE 8,
e65427.
Fox, R., Oliver, T.H., Harrower, C., Parsons, M.S., Thomas, C.D., Roy, D.B., 2014. Longterm changes to the frequency of occurrence of British moths are consistent
with opposing and synergistic effects of climate and land-use changes. J. Appl.
Ecol. 514, 949957.
Franco, A.M.A., Hill, J.K., Kitschke, C., Collingham, Y.C., Roy, D.B., Fox, R., et al., 2006.
Impacts of climate warming and habitat loss on extinctions at species lowlatitude range boundaries. Glob. Change Biol. 12, 15451553.
Freeman, S.N., Newson, S.E., 2008. On a log-linear approach to detecting ecological
interactions in monitored populations. Ibis 150, 250258.

Gregory, R.D., Willis, S.G., Jiguet, F., Vorsek, P., Klvanov,

A., van Strien, A., et al., 2009.
An indicator of the impact of climatic change on European bird populations. PLoS
ONE 4, e4678.
Hampe, A., 2004. Bioclimate envelope models: what they detect and what they hide.
Glob. Ecol. Biogeogr. 13, 469471.
Hickling, R., Roy, D.B., Hill, J.K., Fox, R., Thomas, C.D., 2006. The distributions of a
wide range of taxonomic groups are expanding polewards. Glob. Change Biol.
12, 450455.
Jiguet, F., Gregory, R.D., Devictor, V., Green, R.E., Vorsek, P., Van Strien, A., et al., 2010.
Population trends of European common birds are predicted by characteristics
of their climatic niche. Glob. Change Biol. 16, 497505.
Le Viol, I., Jiguet, F., Brotons, L., Herrando, S., Lindstrm, ., Pearce-Higgins, J.W.,
et al., 2012. More and more generalists: two decades of changes in the European
avifauna. Biol. Lett. 8, 780782.
Mercader, R.J., Scriber, J.M., 2008. Asymmetrical thermal constraints on the parapatric species boundaries of two widespread generalist butteries. Ecol. Entomol.
33, 537545.
Nakagawa, S., Schielzeth, H., 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol. Evol. 4,
133142.
Ockendon, N., Baker, D.J., Carr, J.A., White, E.C., Almond, R.E., Amano, T., et al., 2014.
Mechanisms underpinning climatic impacts on natural populations: altered
species interactions are more important than direct effects. Glob. Change Biol.
20, 22212229.
Oliver, T.H., Brereton, T., Roy, D.B., 2013. Population resilience to an extreme drought
is inuenced by habitat area and fragmentation in the local landscape. Ecography
36, 579586.
Oliver, T.H., Thomas, C.D., Hill, J.K., Brereton, T., Roy, D.B., 2012. Habitat associations of thermophilous butteries are reduced despite climatic warming. Glob.
Change Biol. 18, 27202729.
Pacici, M., Foden, W.B., Visconti, P., Watson, J.E., Butchart, S.H., Kovacs, K.M., et al.,
2015. Assessing species vulnerability to climate change. Nat. Clim. Change 5,
215224.
Palmer, G., Hill, J.K., Brereton, T.M., Brooks, D.R., Chapman, J.W., Fox, R., et al., 2015.
Individualistic sensitivities and exposure to climate change explain variation in
species distribution and abundance changes. Sci. Adv. 1, e1400220.
Pannekoek, J., van Strien, A., 1998. TRIM 2.0 for Windows (Trends & Indices for
Monitoring data). Voorburg, Statistics Netherlands.

B. Martay et al. / Ecological Indicators 68 (2016) 126133

Parmesan, C., Yohe, G., 2003. A globally coherent ngerprint of climate change
impacts across natural systems. Nature 421, 3742.
Pearce-Higgins, J.W., Green, R.E., 2014. Birds and Climate Change: Impacts and Conservation Responses. Cambridge University Press.
Pearson, R.G., Dawson, T.P., 2003. Predicting the impacts of climate change on the
distribution of species: are bioclimate envelope models useful? Glob. Ecol. Biogeogr. 12, 361371.
Perry, M., Hollis, D., 2005. The generation of monthly gridded datasets for a range of
climatic variables over the UK. Int. J. Climatol. 25, 10411054.
Pyry, J., Leinonen, R., Sderman, G., Nieminen, M., Heikkinen, R.K., Carter, T.R., 2011.
Climate-induced increase of moth multivoltinism in boreal regions. Glob. Ecol.
Biogeogr. 20, 289298.
Rennie, S., Adamson, J., Anderson, R., Andrews, C., Bater, J., Bayeld, N., et al., 2015a.
UK Environmental Change Network (ECN) buttery data: 19932012. NERC
Environmental Information Data Centre, http://dx.doi.org/10.5285/dad68dc49e36-457e-aa3a-e7eb816fb508.
Rennie, S., Adamson, J., Anderson, R., Andrews, C., Bater, J., Bayeld, N., et al., 2015b.
UK Environmental Change Network (ECN) moth data: 19922012. NERC Environmental Information Data Centre, http://dx.doi.org/10.5285/9cf26366-6ce14e5e-b648-05345a43e15d.
Rennie, S., Adamson, J., Anderson, R., Andrews, C., Bater, J., Bayeld, N., et al., 2015c.
UK Environmental Change Network (ECN) meteorology data: 19922012. NERC
Environmental Information Data Centre, http://dx.doi.org/10.5285/e1d33b37f1d4-4234-a0d5-8bf4e657f653.
Roy, D.B., Rothery, P., Moss, D., Pollard, E., Thomas, J.A., 2001. Buttery numbers
and weather: predicting historical trends in abundance and the future effects of
climate change. J. Anim. Ecol. 70, 201217.
Settele, J., Kudrna, O., Harpke, A., Khn, I., Van Swaay, C., Verovnik, R., et al., 2008.
Climatic risk atlas of European butteries. Soa-Moscow: Pensoft.
Sier, A.R.J., Monteith, D.T., 2016. The UK Environmental Change Network after
Twenty Years of Integrated Ecosystem Assessment: Key Findings and Future
Perspectives. Ecol. Indic. 68, 112.

133

Sparks, T.H., Yates, T.J., 1997. The effect of spring temperature on the appearance
dates of British butteries 18831993. Ecography 20, 368374.
Stuhldreher, G., Hermann, G., Fartmann, T., 2014. Cold-adapted species in a warming
world an explorative study on the impact of high winter temperatures on a
continental buttery. Entomol. Exp. Appl. 151, 270279.
Ter Braak, C.J.F., van Strien, A.J., Meijer, R., Verstrael, T.J., 1994. Analysis of monitoring data with many missing values: which method? In: Hagemeijer, E.J.M.,
Verstrael, T.J. (Eds.), Bird Numbers 1992. Distribution, Monitoring And Ecological Aspects. Proceedings of the 12th International Conference of IBCC and EOAC,
Noordwijkerhout, The Netherlands. Statistics Netherlands, Voorburg/Heerlen &
SOVON, Beek-Ubbergen, pp. 663673.
Thomas, J.A., 2005. Monitoring change in the abundance and distribution of insects
using butteries and other indicator groups. Philos. Trans. R. Soc. B: Biol. Sci.
360, 339357.
Thuiller, W., Lavorel, S., Arajo, M.B., 2005. Niche properties and geographical extent
as predictors of species sensitivity to climate change. Glob. Ecol. Biogeogr. 14,
347357.
Vedder, O., Bouwhuis, S., Sheldon, B.C., 2013. Quantitative assessment of the importance of phenotypic plasticity in adaptation to climate change in wild bird
populations. PLOS Biol. 11, e1001605.
Virkkala, R., Pyry, J., Heikkinen, R.K., Lehikoinen, A., Valkama, J., 2014. Protected
areas alleviate climate change effects on northern bird species of conservation
concern. Ecol. Evol. 4, 29913003.
Warren, M.S., Hill, J.K., Thomas, J.A., Asher, J., Fox, R., Huntley, B., et al., 2001. Rapid
responses of British butteries to opposing forces of climate and habitat change.
Nature 414, 6569.
Woiwod, I.P., 1996a. IB protocol: butteries. In: Sykes, J.M., Lane, A.M.J. (Eds.), The
United Kingdom Environmental Change Network: Protocols for Standard Measurements at Terrestrial Sites. The Stationery Ofce.
Woiwod, I.P., 1996b. IM protocol: moths. In: Sykes, J.M., Lane, A.M.J. (Eds.), The
United Kingdom Environmental Change Network: Protocols for Standard Measurements at Terrestrial Sites. The Stationery Ofce.