Declining abundance of beetles, moths and caddisflies in the Netherlands

Abstract

Recently, reports of insect declines prompted concerns with respect to the state of insects at a global level. Here, we present the results of longer‐term insect monitoring from two locations in the Netherlands: nature development area De Kaaistoep and nature reserves near Wijster.
Based on data from insects attracted to light in De Kaaistoep, macro‐moths (macro‐Lepidoptera), beetles (Coleoptera), and caddisflies (Trichoptera) have declined in the mean number of individuals counted per evening over the period of 1997–2017, with annual rates of decline of 3.8, 5.0 and 9.2%, respectively. Other orders appeared stable [true bugs (Hemiptera: Heteroptera and Auchenorrhyncha) and mayflies (Ephemeroptera)] or had uncertainty in their trend estimate [lacewings (Neuroptera)].
Based on 48 pitfall traps near Wijster, ground beetles (Coleoptera: Carabidae) showed a mean annual decline of 4.3% in total numbers over the period of 1985–2016. Nonetheless, declines appeared stronger after 1995.
For macro‐moths, the mean of the trends of individual species was comparable to the annual trend in total numbers. Trends of individual ground beetle species, however, suggest that abundant species performed worse than rare ones.
When translated into biomass estimates, our calculations suggest a reduction in total biomass of approximately 61% for macro‐moths as a group and at least 42% for ground beetles, by extrapolation over a period of 27 years. Heavier ground beetles and macro‐moths did not decline more strongly than lighter species, suggesting that heavy species did not contribute disproportionately to biomass decline.
Our results broadly echo recent reported trends in insect biomass in Germany and elsewhere.

Full text

Declining abundance of beetles, moths and caddisflies
in the Netherlands
CASPAR A. HALLMANN,
1
THEO ZEEGERS,
2
ROEL VAN KLINK,
3,4
RIKJAN VERMEULEN,
4
PAUL VAN WIELINK,
5
HENK SPIJKERS,
6
JURRIËN VAN DEIJK,
7
WOUTER VAN STEENIS
8
and
EELKE JONGEJANS
1
1
Department of Animal Ecology and Physiology, Radboud University, Nijmegen, The
Netherlands,
2
EIS Kenniscentrum Insecten, Leiden, The Netherlands,
3
German Centre for Integrative Biodiversity Research (iDiv),
Leipzig, Germany,
4
Stichting Willem Beijerink Biologisch Station, Loon, The Netherlands,
5
Natuurmuseum Brabant, Tilburg/De
Kaaistoep, The Netherlands,
6
Beatrixstraat 1, Goirle/De Kaaistoep, The Netherlands,
7
Dutch Butterfly Conservation, Wageningen, The
Netherlands and
8
Natuurmonumenten, ‘s-Graveland, The Netherlands
Abstract.1. Recently, reports of insect declines prompted concerns with respect to the
state of insects at a global level. Here, we present the results of longer-term insect mon-
itoring from two locations in the Netherlands: nature development area De Kaaistoep and
nature reserves near Wijster.
2. Based on data from insects attracted to light in De Kaaistoep, macro-moths (macro-
Lepidoptera), beetles (Coleoptera), and caddisflies (Trichoptera) have declined in the
mean number of individuals counted per evening over the period of 1997–2017, with annual
rates of decline of 3.8, 5.0 and 9.2%, respectively. Other orders appeared stable [true bugs
(Hemiptera: Heteroptera and Auchenorrhyncha) and mayflies (Ephemeroptera)] or had
uncertainty in their trend estimate [lacewings (Neuroptera)].
3. Based on 48 pitfall traps near Wijster, ground beetles (Coleoptera: Carabidae)
showed a mean annual decline of 4.3% in total numbers over the period of 1985–2016.
Nonetheless, declines appeared stronger after 1995.
4. For macro-moths, the mean of the trends of individual species was comparable to
the annual trend in total numbers. Trends of individual ground beetle species, however,
suggest that abundant species performed worse than rare ones.
5. When translated into biomass estimates, our calculations suggest a reduction in
total biomass of approximately 61% for macro-moths as a group and at least 42% for
ground beetles, by extrapolation over a period of 27 years. Heavier ground beetles and
macro-moths did not decline more strongly than lighter species, suggesting that heavy
species did not contribute disproportionately to biomass decline.
6. Our results broadly echo recent reported trends in insect biomass in Germany and
elsewhere.
Key words. Beetles, collecting at light, insect declines, macro-moths, pitfall trap, trend
analysis.
Introduction
Insects, despite their huge diversity, and despite their importance
for ecosystem functioning, are generally much less studied than,
for example, birds and mammals. As a consequence, information
on the abundance and trends of insects is largely lacking, and/or
is geographically limited, preventing the assessment of their state
in the landscape (Habel et al., 2019a). Additionally, large-scale
monitoring data exist usually only for species such as butterflies
(Van Dyck et al., 2009; van Strien et al., 2019), dragonflies
(Termaat et al., 2015; 2019), bees (Biesmeijer et al., 2006;
Aguirre-Gutierrez et al., 2016) and moths (Groenendijk & Ellis,
Correspondence: Caspar A. Hallmann, Department of Animal Ecol-
ogy and Physiology, Radboud University, Nijmegen, The Netherlands.
E-mail: c.hallmann@science.ru.nl
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological Society.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits
use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or
adaptations are made.
1
Insect Conservation and Diversity (2019) doi: 10.1111/icad.12377

2011; Habel et al., 2019b), taxonomically limiting the inference
that can be made over the state entomofauna in general. Never-
theless, studies on these species largely reveal patterns of decline
in abundance over recent decades (Sánchez-Bayo & Wyckhuys,
2019), with reports on insect declines coming from tropics
(Lister & Garcia, 2018; Janzen & Hallwachs, 2019), to the arctic
(Gillespie et al., 2019). Recently, a large decline in flying insect
biomass was reported for German lowland nature reserves
(Hallmann et al., 2017; Schuch et al., 2019) prompting concerns
with respect to the state of insects at a global level. In response to
the findings in Germany, and commissioned by the Dutch minis-
try of environment and agriculture, Kleijn et al. (2018) identified
a list of existing data sets potentially suitable to derive trends for
insects species in the Netherlands, and to allow for comparison
to the German case. Here, we use two long-term data sets (each
from a single location or area, using different approaches) cover-
ing a wide range of insect families, to provide further insights
into trends in insect abundance in the Netherlands, the trends in
their biomass, and to examine trend variation along species-
specific traits.
Analysis of insect trends over time poses significant chal-
lenges. First, it is often hard to differentiate long-term trends
from natural cycles (Fewster et al., 2000; Benton et al., 2002),
particularly in the absence of prolonged sampling over many
years. Secondly, seasonal activity of the insects plays a signifi-
cant role in the numbers trapped, particularly when species have
multiple generations and peaks throughout the year. Thirdly,
weather variation, possibly at multiple time spans and with var-
iable time lags, influences the population dynamics and activity
of the insects (Johnson, 1969; Jonason et al., 2014; van Wielink,
2017a,b). Hence, sampling characteristics such as timing (both
in the season and during the day) and duration of sampling,
can play important roles in the numbers caught, and hence trend
estimates. If meaningful trends of insect numbers are to be
derived, such sampling characteristics need to be accounted for
in the analyses.
To contribute to answering the question whether the abun-
dance and biomass of insects is declining in the Netherlands,
we report here on insect trends in two longer-term data sets,
while correcting for sampling and weather aspects, and assess
the relative performance of the various insect orders. For the
most well-studied and most species-rich orders, beetles and
macro-moths, we also report trends per species, and we examine
trend variation along a number of species traits as a means to pin-
point potential drivers of trends in abundance (e.g. Potocký
et al., 2018; Habel et al., 2019c). For instance, these analyses
will show whether insect species associated with certain types
of host plants or specific habitats decline more than other insect
species. On the other hand, if species trends show no relation-
ships to species traits, pressure factors would be suspect that
affect all types of insects in the same way. Additionally, based
on general weight-length relationships (Sabo et al., 2002;
García-Barros, 2015), we attempt to derive estimates of trends
in total biomass, in order to compare these to the recently
reported trends in flying insect biomass in Germany (Hallmann
et al., 2017). Our specific research objectives were
1 to assess the trends in abundance of various insects at the
species and order level,
2 to assess the trend in biomass of macro-moths and ground
beetles, and
3 to assess how species-trends vary along species-specific
trait axes.
Materials and methods
Data were collected at two groups of sites: De Kaaistoep and
Wijster. For each site, we describe the sampling protocols, data
set and statistical analysis. A summary description of available
data is given in Table 1. In addition, we obtained data from
two KNMI weather stations (for De Kaaistoep data: weather sta-
tion Gilze-Rijen, for Wijster data: weather station Eelde, at,
respectively, 3.6 and 40 km from trapping locations), from
which we extracted relevant parameters for effect analysis on
insect numbers, as well as for correcting trends.
Collecting at light in De Kaaistoep
De Kaaistoep is a 330 ha managed natural area consisting of
heathland, pine forest and grassland. It was established in 1994
on former arable land. Information about the location and man-
agement history can be found in the study by Felix and van Wie-
link (2008). Insects were attracted by light in combination with a
Table 1. Summary of the data used in the analyses. For each insect order included in this study, we show the number of years, sites and individuals that
were used in the present analysis. Data from the Kaaistoep were collected at light, while data from Wijster using pitfall traps.
Order Family Location Sites Years Samples Individuals Species
Lepidoptera Kaaistoep 1 21: 1997:2017 497 nights 54 492 477(178)
Coleoptera Kaaistoep 1 21: 1997:2017 572 nights 257 793 123(76)
Trichoptera Kaaistoep 1 10: 2006,2009:2017 261 nights 33 540
Ephemeroptera Kaaistoep 1 10: 2006,2009:2017 255 nights 9713
Neuroptera Kaaistoep 1 10: 2006,2009:2017 258 nights 936
Hemiptera Kaaistoep 1 10: 2006,2009:2017 258 nights 49 747
Coleoptera Carabidae Wijster 48 26: 1986:1997,2002:2003,2005:2016 26 years 264 986 156(98)
Coleoptera Carabidae Wijster 31 16: 2002:2017 15 672 weeks 99 075
Separate species trends are performed on a subset of species for which enough data were available (numbers indicated between brackets).
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
2 Caspar A. Hallmann et al.

white cloth (Supporting Information Fig. S1) over a period of
3.3 h per trapping night, normally starting around sunset
(Fig. 1c). During this sampling period, individuals of the various
insect taxa were counted, or were estimated in the case of large
numbers. All macro-moths were always counted and identified,
while for other groups of insects, between 25 and 100% were
collected for identification. Further details of the sampling proto-
col are given in the study by van Wielink and Spijkers (2013).
Data in the presentanalysis have been collected during 628 trap-
ping nights between 1997 and 2017, on average 30 evenings per
year (10–77). Data were available for the period of 1997–2017
for macro-moths (Lepidoptera), beetles (Coleoptera) and ground
beetles (Carabidae), while for caddisflies (Trichoptera), lacewings
(Neuroptera), true bugs (Hemiptera-Heteroptera and Hemiptera-
Auchenorrhyncha) and mayflies (Ephemeroptera) data were avail-
able only for the years 2006 and 2009–2017. Of the large number
of Coleoptera, only ground beetles, ladybirds and carrion beetles
were identified to species up to 2017, accounting for 48 000 of
239 000 beetle specimens.
As it is known that the environmental conditions (like temper-
ature) during each trapping night influenced the number of
insects caught, we aimed to include relevant covariates in our
analyses. Information about the timing and duration of sampling
were available for 91.2% of the nights (n= 574), and lacking
more in the first few years of sampling than later on. The number
of sampling hours per night varied little among years (Fig. 1a)
but did increase from an average of 3.1 h (1997–2009) to an
average of 3.8 h per night after 2010 (F= 48.98, d.f. = 572,
P<0.001; Fig. 1b). Timing of onset of sampling was roughly
at sunset throughout the years, with the exception of the first
few years in which sampling started on average up to half an hour
after sunset (Fig. 1c,d). The starting time of sampling correlated
significantly (R
2
= 96.6%, df = 514, P< 0.001) with the evaluated
sunset moment for the specified location (Meeus, 1991; Bivand &
Lewin-Koh, 2015). Additionally, the slope of the linear relation-
ship between the startingand sunset moments did not deviate sig-
nificantly from one (F=0.809,P= 0.369), and the intercept did
not deviate from zero (F=1.568,P=0.211).
To analyse trends for each order (or species) k, we modelled
the counts in year t and on day d using generalised additive
models (GAMs; Wood, 2006) and assuming a negative-binomial
distribution (White & Bennetts, 1996) and a log link to the pre-
dictors. GAMs were deemed more appropriate than generalised
linear models, as insects counts vary considerably throughout
the year, often with multiple peaks (i.e. generations), as well as
between years (i.e. nonlinear dynamics). We constructed six
basic models, differing in how the year covariate is treated (lin-
ear, non-linear, and categorical), and if the weather covariates
were included or not (Supporting Information Table S1). We
considered linear as well as non-linear trends over time, as well
as an annual index (the latter for visual assessment). Addition-
ally, in all models, we included a smooth seasonal component
(a) (b)
(c) (d)
Fig. 1. Sampling characteristics for De Kaaistoep data set. (a) Number of sampling hours per evening plotted against day of the year (1 = 1 January).
(b) Number of sampling hours per evening per year. (c) Start of sampling relative to sunset (in hours) versus day of the year. (d) Start of sampling relative
to sunset per year.
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
Declining abundance of beetles, moths and caddisflies in the Netherlands 3

[γs(d)] and a quadratic component for sampling duration
(h + h2), as we expected non-linear responses to sampling dura-
tion. Weather covariates included mean temperature, sum of pre-
cipitation, mean relative moisture content and mean wind speed.
Additionally, as response variables may have a convex relation-
ship (e.g. optima) to weather variables, we also included qua-
dratic effects. Each weather covariate in the design matrix
W(including the squared values) was standardised to a zero
mean and unit variance. The different models were compared
by the Akaike’s information criterion (AIC) (Burnham & Ander-
son, 2003), a measure of parsimony that tries to balance the
amount of deviance explained and the number of parameters.
Pitfall traps near Wijster
A long-term monitoring program using pitfall traps was
started at the Wijster Biological Station (and continued by the
Foundation Willem Beijerink Biological Station) in two nature
reserves in the province of Drenthe: National Park Dwingelder-
veld and the fragmented, but increasingly reconnected Hullen-
zand. In these reserves restoration measures, mainly in the
form of topsoil removal and reconnection, were carried out dur-
ing the early 1990s. The pitfall data have been collected between
1959 and 2016 at in total 48 unique locations (mean = 9, range
4–19 operating locations per year). The locations consisted
mainly of heathlands, with some forest sites, a forest edge and
an abandoned crop field. At each location, three square pitfall
traps (25 ×25 cm) were installed (Supporting Information
Fig. S2): one lethal funnel trap with a 3% formaldehyde solution
and two live traps. The traps at each location were spaced 10 m
apart. Caught ground beetles (Coleoptera: Carabidae) have been
identified at weekly intervals. Further details on the sampling pro-
tocol and the area are givenin the study by den Boer and van Dijk
(1994). Because we are only interested in recent trends in insect
abundances, and because sampling protocols were not consistent
in the early years, we only used data collected since 1986. We per-
formed two types of analyses: we first used the annual sums per
species and location for the period of 1986–2016 (Table 1), and
secondly, the weekly sums per species and location that have been
fully digitised and checked: 2002–2017.
Annual totals 1986–2016. In total, 7778 records of
species-specific counts were used in the present analyses, which
amounted to 264 986 individual ground beetles. For 20 records,
we used multiple imputation (Onkelinx et al., 2017) to derive
more reliable estimates for suspected erroneous counts. This
method is based on the correlation structure between years and
between other species. Note that in the years 1998–2001, no
monitoring took place, and 2004 was omitted because of incom-
plete catches. We used GAMs to model the annual community
abundance and counts per species (based on annual totals) with
a negative-binomial distribution and a log link. We treated trap
location as a random effect by making use of the random effects
as smooth terms (Wood, 2006; 2008). We considered six basic
models depending on how the year covariate is treated, and if
weather covariates are included or not (Supporting Information
Table S2). We considered both linear and nonlinear trends over
time, as well as an annual index (the latter for visual assess-
ments). Weather covariates included mean temperature, sum of
precipitation, mean relative moisture content, and mean wind
speed, over the spring months in each year (March to May),
and separately over the summer months (June to August). Addi-
tionally, we also included quadratic effects of each variable.
Each weather covariate in W(including the squared values)
was standardised to a zero mean and unit variance.
The number of years each location was sampled varied between
1 and 22, with 19 of the locations only sampled in 1 year and
10 locationsonly sampled in 2 years. To assess whether our trend
estimates were affected by including locations with limited years
of sampling, we repeated the analysis by only including locations
in our models when the number of years sampled exceeded a par-
ticular threshold. This threshold varied between 2 and 10 years,
and, for each repetition, we computed the annual trend coefficient
from model M1, along with the standard error.
Weekly counts 2002–2017. For the years for which weekly
data were available, the catches at weekly intervals were ana-
lysed to observe how weather patterns and seasonal variation
might account for some of the inter-annual variation in ground
beetle abundances. Here too, we used GAMs with a negative
binomial error structure, and a log link. We used modelling for-
mulations with a seasonal component (a cubic cyclic spline for
all models), a random effect for trap location (for all models),
and an inter-annual component that was specified either as a cat-
egorical variable, as a linear trend, or as a smooth thin plate
covariate. Additionally, we evaluated effects of temperature
and precipitation in half of the models, yielding in total six differ-
ent model formulations (Supporting Information Table S3).
Location was included in all models as a random effect.
Biomass estimation
Insect monitoring at De Kaaistoep and Wijster is based on
counts of individuals per species or higher taxa, while weighing
of insects is not part of the monitoring protocol. Yet, we deemed
it interesting to try to compare our abundance trends to recent
findings of insect biomass declines in Germany (Hallmann
et al., 2017). We therefore tried to translate species-specific
counts into total biomass estimates. For that purpose we used
known species length measurements and known relationships
of length to weight (Sabo et al., 2002; García-Barros, 2015).
For the Carabidae in the Wijster data set, we used the minimum
and maximum body length as stated in the Dutch ground beetles
field guide (Boeken et al., 2002). Per species we averaged the
minimum and maximum lengths, and used these averages to esti-
mate mass per specimen (k), using the mass-length relationship
determined by Sabo et al. (2002) for terrestrial insects:
massk=0:032 ×length2:63
kð1Þ
where mass is in mg and length in mm.
For the macro-moths at De Kaaistoep site, we used species-
specific minimum and maximum lengths of the front wings,
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
4 Caspar A. Hallmann et al.

which is the only size measure provided at the website of the
Dutch Butterfly Conservation (assessed 11 April 2018). Again
we averaged the minimum and maximum lengths (sometimes
sex-specific) per species, but now used a Lepidoptera-specific
mass–length relationship. García-Barros (2015) measured the
mass (mg) and front wing lengths (mm) of 665 specimens. As
García-Barros only reported the means and sample sizes per
superfamily (his Supporting Information 5), we analysed those
summary data in a log–log regression analysis with sample size
as the weight of the records. Superfamily-specific residuals (ϵ
k
)
of this regression analysis were stored. The fitted model was then
used to estimate the mass of marco-Lepidoptera species based on
its average front wing length and the superfamily it belongs to:
massk= exp −5:144 + 3:018 ×log lengthk
ðÞ+ϵk
ðÞð2Þ
where, for instance, the effect sizes (ϵ
k
) of Noctuoidea and Geo-
metroidea were 0.218 and −0.126, respectively.
In order to calculate the reduction in biomass over the years,
we used the sum of individual species weights (Bt) estimated
for a particular year t (for ground beetles in the Wijster data
set) or day d (for macro-moths in De Kaaistoep data set):
Bt=XK
k=1Bk,tð3Þ
and where B
k,t
=Y
k,t
×mass
k
, i.e. numbers counted per species
(Y
k,t
) multiplied by their estimated mean mass.
We ran GAMs on the resulting responses, using a Gaussian
distribution and log–link relationship to the covariates. For De
Kaaistoep data, we used the formulation of model M4
(Supporting Information Table S1) and for the Wijster data
model A1 (Supporting Information Table S2).
Trend classification
We classified order- and species-specific trends in abundance
and biomass, based on estimates of the annual trends coefficient
ρand on its significance. The trend coefficients represent the
annual intrinsic rate of population change, or equivalently, the
natural logarithm of the mean annual multiplication factor. Trend
coefficients close to zero (−0.025 < ρ< 0.025) were interpreted
as indicators of stable population trends, while more negative ρ
associated with P-values larger than 0.05 were classified as
‘uncertain declines’. Declines were labelled ‘severe’when sig-
nificant ρvalues were lower than −0.05. More information on
these trend classifications can be found in the Supporting Infor-
mation Table S4.
Species traits
We examined variation in species mean log annual trend in rela-
tion to ecological traits, for macro-moths in De Kaaistoep, as well
as ground beetles from Drenthe. For macro-moths, trait data were
assembled from existing literature, and include voltinism (five clas-
ses: one generation per year, one or two generations, two generations,
two or three generations, and three generation per year; Waring &
Townsend,2015), wintering strategy (fourclasses: as egg, caterpil-
lar, pupa or adult; Ebert, 2005; Waring & Townsend, 2015 ),
host plant type (six classes: grass, herb, trees and shrubs, trees
only, diverse, and other; Waring & Townsend, 2015), host plant
specificity(threeclasses:monophagous,oligophagousand polyph-
agous; Waring& Townsend, 2015), rarity (five classes: rare to very
common, Ellis et al., 2013), and the log of species weight (see the
aforementionedexplanation).Host plant type class ‘other’included
several species of heath, and mosses and lichens. Using data from
Habel et al. (2019c), we also examined the effects of Ellenberg
values of the host plants of macro-moths, and major habitat
type, on mean annual species trends. This was done for a subset of
the species that overlapped between the present study and the one
of Habel et al. (2019c), and for which trends were estimable
(N= 146 out of 178 species trend estimates).
For ground beetles in Drenthe, we derived species traits from
Turin (2000), while reducing the number of categories for sev-
eral traits in some variables. We used three categorical trait vari-
ables, namely: flight ability [macropterous (i.e. having large
wings), brachypterous (i.e. having reduced wings), dimorphic
or polymorphic], habitat specialisation [four classes: from steno-
topic (i.e. specialised to one or few habitats)to very eurytopic
(i.e. habitat generalist)], distribution type (four classes marginal,
submarginal, sub central and central), and the log of species
weight. Habitat specialisation was condensed from numeric scale
(2–10) into the four mentioned classes as follows: 2–4 stenotopic,
5–6 less stenotopic, 7–8 less eurytopic, and 9–10 eurytopic. The
original rankings simply resemble the number of types of habitat
each species has been found in the Netherlands.
To examine the effects of the traits, we regressed the intrinsic
rate of increase to the aforementioned traits using generalised
least squares. As we expected greater residual variation in low-
density species because of higher demographic stochasticity
(i.e. heteroscedasticity), we specified the variance around the
mean (V(y)) as an exponential function of the log of mean species
abundance as:
VyðÞ=σ2exp 2 ×φ×log yðÞðÞð4Þ
where φis an to be estimated parameter measuring the decline in
variance with increasing species abundance. Starting with a
global model (all traits as covariates) and using a stepwise dele-
tion of insignificant terms, we derived the most parsimonious
models for each group.
Results
Collecting at light at De Kaaistoep
Across insect orders, models including weather variables
always prevailed over models without weather variables
(Supporting Information Table S5). Across orders, sampling
duration was significantly positively related to the number of
insects counted. Given the increase in sampling duration from
an average of 3.1 h in the period of 1997–2006 to an average
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
Declining abundance of beetles, moths and caddisflies in the Netherlands 5

of 3.8 h in 2009–2017 (Fig. 1b), fitted trends over the study
period were slightly lower when correcting for sampling dura-
tion (Supporting Information Fig. S3). Hence, we derived annual
trends while accounting for weather variables and sampling
duration (see Supporting Information Table S6 for coefficients).
Trends of the abundance of six insect orders (based on an
annual index, a linear and a non-linear trend) are depicted in
Fig. 2. Following correction for sampling duration and weather
effects, and based on the overall mean (linear) estimates, true
bugs (Hemiptera-Heteroptera and Hemiptera-Auchenorrhyncha)
appeared to be stable, and lacewings (Neuroptera) appeared to
decline but not significantly so, and hence their trend was consid-
ered to be uncertain. In contrast, caddisflies (Trichoptera),
mayflies (Ephemeroptera), beetles (Coleoptera) and moths
(macro-Lepidoptera) showed significant negative coefficients.
The linear trends per order are summarised in Table 2. Because
apparent declines in Trichoptera and Ephemeroptera might have
been dominated by high counts in 2006, we re-analysed
these trends while excluding data from 2006. For mayflies, the
trend coefficient changed both magnitude and sign (ρ= 0.010,
se = 0.058, P-value = 0.87), and we therefore labelled the trend of
this insect order to be stable. For caddisflies, the trend became
slightly less negative when the year 2006 was omitted but remained
significantly negative (ρ=−0.070, se = 0.033, P-value = 0.033).
Trends of macro-moth species were variable, with on average a
decline of 4% per year (Fig. 3a). The largest group of species
(38%) showed a declining trend, while only 5% showed an increase
and the remainder of the species had stable or insignificant trends
(Fig. 3b). Declines of individual species were positively, but not
significantly, related to mean abundance (mean number of individ-
uals per trapping night; t-value = 0.861, P-value = 0.392).
Within the 76 beetle species for which enough data was avail-
able to analyse population trends, the average annual decline was
estimated to be −0.05, with 38% of the species showing a signif-
icant (and severe) decline, while 12% of the species significantly
increased (Figure 3). The species-rich family of ground beetles
(Carabidae) dominated these results, with numeric declines
(totals within family) of ground beetles declining steeper
(ρ=−0.090, se = 0.021, P-value<0.001) than those of ladybirds
(Coccinellidae, excluding the invasive exotic Harmonia axyri-
dis,ρ=−0.029, se = 0.012, P-value = 0.001), whereas carrion
beetles (Silphidae, n= 4) were found to significantly increase
(ρ= 0.035, se = 0.016, P-value = 0.003). Within ground beetles,
average species declines amounted to 6.8% per year, and
although species-specific trends were highly variable, a large
proportion of these species showed significantly declining trends
(44.1%), and only few (6.8%) showed increases (Supporting
Information Figure S4).
Pitfall traps near Wijster
In total, 156 species of ground beetles were found in the pitfall
traps. Year totals of specimens over all species of ground beetles
showed a declining pattern regardless of the considered model.
Although non-linear trends explained year totals significantly
better than linear models (AICnl = 3768.26, df = 35.54 versus
AICl = 3773.63, d.f = 33.48). Models considering weather
variables did not improve model fit, regardless of whether they
were measured over spring (March to May) or summer (June
to August). Hence, we present trends based on models that omit
weather effects. The linear trend coefficient was significantly
negative (ρ=−0.044, se = 0.006, P−value < 0.001, 4% decline
per year, Fig.4). Results of the non-linear trend model however
showed that the trend initially increased, followed by a decline
starting after 1995 (Fig. 4). The linear annual trend since 1995
showed even steeper declines (ρ=−0.060, se = 0.009,
P< 0.001), implying a 6% annual decline since 1995.
Furthermore, the trend estimates were affected by the mini-
mum number of years that a given location was sampled. While
the main analysis included all locations, including only locations
with more than 2 years of sampling resulted in a slightly more
negative trend coefficient of ρ=−0.051 (se = 0.005), i.e. 5%
annual decline rate. Restricting the analysis to the 12 locations
with at least four sampling years made the trend even more neg-
ative (5.5% annual rate of decline, Supporting Information
Figure S5).
Among 127 ground beetle species with sufficient data, the
average of the species trends (based on year totals) amounted
to a 7% decline per year (Fig. 3a), which is more negative than
the trend of the year totals. Most species (42.5%) showed declin-
ing (most of which severe declines) trends, while 29.4% of the
species showed stable or uncertain trends and 8.5% of species
showed significantly positive trends (Fig. 3b).
Trend estimates as obtained from our analysis of the weekly
counts of all ground beetles combined (over the years
2002–2017, see Methods) were similar but more negative to that
of the year totals over the longer period. In these seasonal ana-
lyses, models with weather variables did outperform models
without such variables (Supporting Information Table S7). On
the contrary, the mean annual trend coefficient did not differ
much between these models. Based on the weather-corrected
annual trend coefficient, we estimated the annual decline at an
average of 7.41% (ρ=−0.077, se = 0.002, P< 0.001) for the
period of 2002–2017 (Supporting Information Figure S6).
Trends in estimated insect biomass
For the macro-moths at De Kaaistoep site, our calculations cul-
minated in an estimation of ‘severe decline’for total biomass
(ρ=−0.036, se = 0.006, P<0.001,i.e.−3.3%, se = 0.52 mg/year;
Fig. 5a). For the ground beetles near Wijster, we estimated the
average decline in total biomass to be 2% (se = 0.48) annually
(Fig. 5b), which is considerably less than that of numbers per spe-
cies or total sums of individuals. Nevertheless, considering only
the period after 1995, the rate of decline in biomass appeared a
lot more severe (ρ=−0.0414, se = 0.006, P< 0.001), implying
an on average 4.1% (se = 0.53) decline per year.
Species traits
For both macro-moths and ground beetles, accounting for
heteroscedasticity provided a significant better fit to the data
(log-likelihood ratio of 19.91, P< 0.0001, for macro-moths,
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
6 Caspar A. Hallmann et al.

and 25.99, P< 0.0001, for ground beetles) and hence was
retained in all models. Analysis of the trends of macro-moths
in relation to traits showed that out of the covariates considered,
only host plant type explained a significant amount of variation
(Supporting Information Table S8), with species depending on
grass, herbs or diverse host plant species declining most
(Fig. 6, Supporting Information Table S9). Additional analysis
based on a subset of the macro-moth species in relation to Ellen-
berg values of the host plants (data from Habel et al., 2019c) did
not reveal any significant effects of the predictors (Ellenberg
values for nitrogen, pH, light, continentality, humidity and tem-
perature; Supporting Information Table S10).
For ground beetles in the Wijster area, lower intrinsic rates
were observed among species that are considered in the Nether-
lands to be in the margin or sub-margin of their distribution,
among very stenotopic (i.e. restricted to few types of habitats)
or very eurytopic species (i.e. habitat generalists), among lighter
species, and among xerophilic (i.e. occurring in dry habitats)
species (Fig. 7; Supporting Information Tables S11 and S12).
Discussion
We reported trends of six insect orders collected at light in De
Kaaistoep, and one family of beetles in the Wijster region.
Macro-moths, caddisflies, beetles and its subset of ground bee-
tles at De Kaaistoep, and ground beetles near Wijster, showed
severe declines. Only true bugs and mayflies appeared to be sta-
ble, while the negative trend for lacewings was statistically not
significant. The majority of macro-moths (macro-Lepidoptera)
are attracted to light, as are mayflies (Ephemeroptera) and
caddisflies (Trichoptera), and hence are expected to be well
represented in the data obtained by collection at light in De
Kaaistoep. Similarly, the Wijster pitfall dataset, with 127of
395 species observed in the Netherlands, can be considered
as representative for ground beetles (Coleoptera: Carabidae)
species present in the Netherlands.
Amid recent reports of broad insect decline in German nature
reserves (Hallmann et al., 2017; Habel & Schmitt, 2018; Hom-
burg et al., 2019; Schuch et al., 2019), concerns with respect to
2006 2008 2010 2012 2014 2016
0
10
20
30
40
50
60
70
abundance
2006 2008 2010 2012 2014 2016
0
10
20
30
40
50
60
70
2006 2008 2010 2012 2014 2016
0.0
0.5
1.0
1.5
2.0
2006 2008 2010 2012 2014 2016
0
5
10
15
20
2000 2005 2010 2015
0
50
100
150
200
Year
2000 2005 2010 2015
0
20
40
60
80
100
Year
Trichoptera
(a) (b)
(c) (d)
(e) (f)
Hemiptera
Neuroptera Ephemeroptera
Coleoptera Lepidoptera
Fig. 2. Trends in numbers counted per evening of six orders of insects at De Kaaistoep. For each order, the annual indices (points, model M3), and esti-
mates of the linear (orange, model M4) and non-linear (blue, model M5) trends are given. Evidence for non-linearity is only apparent in Neuroptera, Ephe-
meroptera and Coleoptera, while for the remainder of the orders models M4 and M5 are indistinguishable. [Color figure can be viewed at
wileyonlinelibrary.com]
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
Declining abundance of beetles, moths and caddisflies in the Netherlands 7

the state of Dutch entomofauna have been raised (Kleijn et al.,
2018). Previous results from country-wide analyses in moths
(Groenendijk & Ellis, 2011) and butterflies (van Swaay et al.,
2018) showed a drop in absolute numbers of 37% over 30, and
40% in 25 years, respectively. Our analysis, covering a compara-
tively wider range of insect species (over 1700 species, i.e. 9%,
out of the 19 254 known insect species in the Netherlands), and
showing broad declines for most orders investigated, are likely to
be indicative to a broader group of insects in these areas, reinforcing
the concerns with respect to the state of insects in the Netherlands.
Yet, since only two areas are included in this analysis, it is hard to
generalise to the national level, and we urge caution with extrapolat-
ing conclusions from these results to broader spatial levels.
On average, annual trends of macro-moths were negative
(totals: −3.9%, mean species −4%) suggesting a proportionally
uniform decline rate across abundance classes of this taxon.
Since no relation was found between weight of the species and
their annual trend, we conclude that the biomass reduction
Table 2. Trend evaluation per insect order. For each order, we provide
the annual trend coefficient (log of average annual population growth
rate) of model M
4
, along with its standard error between brackets, as well
as a translation into the percentage decline per year.
Insect order
Annual trend
coefficient (ρ) % Decline P-value
Trend
evaluation
Lepidoptera −0.040 (0.006) 3.9 <0.001 Decline
Coleoptera −0.048 (0.010) 4.7 <0.001 Decline
Trichoptera −0.096 (0.021) 9.2 <0.001 Severe decline
Ephemeroptera −0.128 (0.037) 12.0 0.001 Decline
(uncertain)
Neuroptera −0.047 (0.029) 4.6 0.108 Decline
(uncertain)
Hemiptera −0.006 (0.022) 0.6 0.789 Stable
See Supporting Information Table S4 for the scheme of the significance
evaluation of the trends. See the main text for a discussion about the
uncertainty concerning the Ephemeroptera trend.
Fig. 4. Trends in tota lnum bers of ground beetles (Coleoptera: Carabidae) in
pitfalls near Wijster. [Color figure can be viewed at wileyonlinelibrary.com]
(a)
(b)
Fig. 3. Log of annual trend coefficient (ρ) of species of macro-moths (n= 178) and beetles in De Kaaistoep (n= 76) as well as ground beetles in Wijster
(n= 130). (a) Barplots depicting trend classifications. (b) Distribution of trend coefficients. [Color figure can be viewed at wileyonlinelibrary.com]
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
8 Caspar A. Hallmann et al.

(−3.3% per year) is shared proportionally among macro-moth
species, with declines in abundant species naturally accounting
for a larger extent of the biomass decline. Annual decline in total
biomass of ground beetles (based on pitfall data), however, was
less negative than the average of the individual species trend
(totals −6%, mean species trend −7%, biomass −4%). Addition-
ally, following corrections of several traits, a positive effect was
found of weight on species trend (Fig. 7c). Here, the less
abundant and smaller species showed stronger declines than
common or larger species, giving rise to a much lower decline
rate in biomass as compared to the numerical declines. These
results imply that the declines in insect biomass, although indic-
ative to diversity loss, may not always show a one-to-one corre-
spondence to numerical declines (Homburg et al., 2019).
Identifying causes of insect population change was beyond the
scope of this study. However, both areas are nature reserves
managed with the prime aim to protect and restore biodiversity.
In the Wijster region, our data series start a few years prior to
1995, where a peak in numbers (and species) of beetles occurred
following restoration of degraded heath. It is possible that, for
example, succession from open ground to more closed heath/
forest over time may have impacted ground beetle communities.
The more negative trends among specialised xerophilic species
support this hypothesis. However, lowered trends were also
observed among lighter species, and among habitat specialist
(i.e. stenotopic) species, implying that succession is not the sole
driver of decline here. Similarly, in De Kaaistoep, changes since
the 1990s in management of forests and the transformation of the
agricultural area into a more natural landscape, together with
drying of grassland parcels have possibly affected macro-moth
and other insect taxa. Indeed, species depending on grass and
herb host plants seemed to be affected more severely in this area.
Elsewhere (e.g. Habel et al., 2019c) succession also has been
found to be important in shaping moth communities. It has to
be noted, however, that due to the attraction by light, species
(e.g. moths) are drawn into the study site from a wider area. As
such, our results may represent the surrounding environment as
well as the local conditions. With the recent notions that biodi-
versity loss occurs at a landscape scale (Habel & Schmitt,
2018) and that more generalist and abundant species are equally
Fig. 5. Biomass trend of (a) macro-moths (Lepidoptera) per trapping night at De Kaaistoep and (b) ground beetles (Coleoptera: Carabidae) per year from
pitfalls near Wijster. For each order, the annual indices (points), and estimates of the linear (orange) and non-linear (blue,) trends are given. Evidence for
non-linearity is only apparent in Ground beetles, while for the remainder of the macro-moths the estimated trends of the two species are indistinguishable.
[Color figure can be viewed at wileyonlinelibrary.com]
Fig. 6. Mean log annual trend coefficient (ρ, +95% confidence levels) of
macro-moth species (in De Kaaistoep) with various types of host plants.
The number of macro-moth species are indicated for each host plant
category.
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
Declining abundance of beetles, moths and caddisflies in the Netherlands 9

affected as rare species, it may well be that our results regarding
macro-moths reflect landscape health, rather than ‘only’site-
specific conditions.
Sometimes a decline or increase can be made very plausible.
The decline of Coccinellidae, for example, could be explained
by the introduction of the invasive ladybird Harmonia axyri-
dis,first noted on the illuminated screen in 2003 and rapidly
increasing in the following years (van Wielink, 2017a, b).
On the other hand, the increase in carrion-beetles (Silphidae)
can be explained by carrion experiments done at approxi-
mately 25 m from the light source in the period of 2015,
2016 and 2017. The significant decline in caddisflies (but not
mayflies), being aquatic species, is surprising at first sight,
because water quality is thought to have improved locally over
recent years, with sensitive aquatic species (e.g. larvae of Odo-
nata) showed positive population trends in a stream in De
Kaaistoep about 1 km from the collection site (van Wielink &
Spijkers, 2012). However, for dragonflies, Termaat and van
Strien (2015) report a decline starting around 2008, quite similar
to our results. It would require insect and environmental data
from multiple sites to tease apart potentially positive effects of
improved water quality and negative effects from other environ-
mental factors (such as eutrophication) and pollutants (including
pesticides; Zahrádková et al., 2009; Nakanishi et al., 2018).
Additional analyses integrating besides species traits, also habi-
tat and landscape changes (e.g. road traffic, Martin et al., 2018),
are likely to increase our understanding of the present declines
observed, and help delimit for which part these can be attributed
to anthropogenic (e.g. nitrogen deposition and pesticide leach-
ing) or natural (e.g. succession) factors.
In both data sets, the counts of individuals are a reflection of
both abundance and activity of species. This implies that the
numbers caught cannot be translated into a (relative) measure
of abundance directly, but require accounting for effects of
seasonality, phenology and weather. Moreover, inter-annual
cyclic or erratic patterns in abundance of some species compli-
cates the interpretation of trends, particularly so when shorter-
term data underlie the calculations. Here, weather data and the
inclusion of seasonality have improved the fit of the models
for all orders examined in both areas. For three of the orders
in De Kaaistoep, models with an annual index (a categorical
covariate) were selected over linear or nonlinear (spline)
(a) (b)
(c) (d)
Fig. 7. Mean log annual trend coefficient (ρ, +95% confidence levels) of ground beetle species (in the Wijster region) for different levels of
(a) distribution (ranging from species in the Netherlands being in the margin of their distribution to more in the center), (b) specialisation (ranging from
stenotopic habitat specialist to eurytopic habitat generalists), (c) weight and (d) Turin (2000) classification depending on preferred habitats [i.-
e. hydrophylous, no preference, forest, xerophylous (adapted to dry conditions)].
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
10 Caspar A. Hallmann et al.

models, while second best were usually the nonlinear models.
These results show the challenges associated with the erratic
temporal behaviour of some insect populations, and the need
for more complex models to accommodate sources of variation
and bias. Despite our efforts, there is room for improvement in
the trend calculations, for example, by incorporating species-
specific detection probabilities, for which we currently do not
have sufficient information. Hence, we cannot rule out that
changes in species-specific detection rates and community
composition may be for a small part responsible for the decline
rates observed.
Comparison of the presented abundance trends with the Ger-
man (−76% in biomass) and Puerto Rico (−98% in abundance)
results (Hallmann et al., 2017; Lister & Garcia, 2018) remains
difficult because insect traps vary widely in which groups of
insects are sampled (Russo et al., 2011). The methods used in
this study, collecting at light and pitfall traps, both sampled dif-
ferent species and numbers than the malaise traps that were
deployed by the Krefeld Entomological Society in Germany,
or sweep-netting and sticky-traps as applied in Puerto Rico.
Furthermore, in the German study, total biomass of all insects
caught in the Malaise traps was analysed, while here we focus
on counts of important insect orders. Still, we made an attempt
to compare our results with the reported 76% decline in total
insect biomass over 27 year (Hallmann et al., 2017). To do
so, we estimated total biomass for macro-moths in De Kaais-
toep and ground beetles near Wijster based on the assumption
that published species-specific sizes and general size-weight
relationships would be accurate enough to not affect the bio-
mass estimates in a distorting way. For macro-moths, the bio-
mass reductions amounted to 3.3% per year. Over an
extrapolated period of 27 years, this amounted to a reduction
of 61%, which is close to (but less than) the reported declines
in Germany for total flying insect biomass. Ground beetles of
the Wijster data set also showed a negative biomass trend,
although at a less strong rate (mean = 2% per year). Over a
period of 27 years, this would amount to 42% reduction in total
biomass. Additionally, after 1995, the average rate of decline in
biomass was more severe (4.1%), which, over a period of
27 years, would amount to 67%. Even higher rates of decline
can be found depending on which locations are included
(i.e. including only long series of locations results in more neg-
ative annual trends, Supporting Information Fig. S5). Given the
latter, our results for the ground beetles in the heathlands and
forests near Wijster are likely to be conservative. While we
lacked the required species-specific information to estimate
biomass trends for the other insect orders, the variable trends
at the order level (e.g. severe decline in caddisflies, stable in
true bugs) suggests that not all insect orders might have contrib-
uted equally to the decline in total insect biomass as suggested
in the Krefeld study. Note, however, that elsewhere in Germany
true bugs did show strong declines (Schuch et al., 2019), sug-
gesting that the present trends of true bugs might not be indic-
ative for large-scale trends. Future research will hopefully
disentangle these contributions by various insect groups in a
quantitative analysis, which should also shed more light on
the factors that are most instrumental in causing insect numbers
and biomass to decrease this much.
Conclusions
In Dutch nature reserves, insects, particularly macro-moths,
ground beetles and caddisflies, appear to be in considerable
decline according to the studied datasets, as are lacewings,
albeit with less certainty. Together with recent reports on but-
terflies (van Swaay et al., 2018) at the national level, the lim-
ited information that is available suggests that many insect
species in the Netherlands are in decline too (but not all,
e.g. Termaat et al., 2015), similar (but a little less negatively)
to the trends reported for the German nature areas (Hallmann
et al., 2017) or in other regions (Lister & Garcia, 2018, Sán-
chez-Bayo & Wyckhuys, 2019). As such, we suggest that the
declines in insects may be a widespread phenomenon not lim-
ited to nature areas in Germany only. The fact that these stud-
ies are based on data collected using different approaches
strengthens this conclusion. Moreover, with exception of
Hemiptera, our results suggest similar rates of decline as
reported in a recent literature review study (Sánchez-Bayo &
Wyckhuys, 2019).
Standardised networks to monitor the state of insects in the
Netherlands are largely absent, or limited to few species
groups only. Including a relatively broad spectrum of insect
species, this study shows many species being in severe decline,
but also few species increasing, and some groups being
affected less or not at all. More detailed monitoring and eco-
logical studies are thus required to shed light on the actual
causes of decline. Structural funding and facilitation for devel-
oping such monitoring networks, possibly using citizen sci-
ence, is highly required at the moment, as this would provide
the information necessary to assess the state of entomofauna
in the Netherlands, investigate drivers and to develop conser-
vation guidelines. Further work should concentrate on formu-
lating and testing plausible causes for the declines observed
presently.
Acknowledgements
We are grateful for the permission to perform insect monitoring
on land owned by TWM Gronden BV and Natuurmonumenten.
We are grateful for Natuurmonumenten, the Uyttenboogaart-
Eliasen Foundation and the Netherlands Organisation for Sci-
entific Research (NWO-grant 841.11.007) for financial
support. We are also grateful to Piet den Boer, who started
the Wijster data collecting in 1959 already and Alje Woltering
for field assistance in the Wijster region. Guido Stooker
assisted in the macro-moth monitoring at De Kaaistoep in
2016 and 2017. Ron Felix identified part of the Carabidae of
De Kaaistoep. Willem Ellis entered data (1995–2015) in the
Noctua database. Hans Turin helped with the trait database
for ground beetles.
Conflict of Interest
All authors declare that they have no conflict of interest.
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
Declining abundance of beetles, moths and caddisflies in the Netherlands 11

Supporting information
Additional supporting information may be found online in the
Supporting Information section at the end of the article.
Appendix S1: Supporting Information
References
Aguirre-Gutiérrez, J., Kissling, W.D., Carvalheiro, L.G.,
WallisDeVries, M.F., Franzén, M. & Biesmeijer, J.C. (2016) Func-
tional traits help to explain half-century long shifts in pollinator distri-
butions. Scientific Reports,6, 24451.
Benton, T.G., Bryant, D.M., Cole, L. & Crick, H.Q. (2002) Linking agri-
cultural practice to insect and bird populations: a historical study over
three decades. Journal of Applied Ecology,39, 673–687.
Biesmeijer, J.C., Roberts, S.P.M., Reemer, M., Ohlemüller, R.,
Edwards, M., Peeters, T., Schaffers, A.P., Potts, S.G., Kleukers, R.,
Thomas, C.D., Settele, J. & Kunin, W.E. (2006) Parallel declines in
pollinators and insect-pollinated plants in Britain and The Nether-
lands. Science,313, 351–354.
Bivand, R. & Lewin-Koh, N. 2015. maptools: Tools for Reading and
Handling Spatial Objects, 2014. R package version 0.8-29.
Boeken, M., Desender, K., Drost, B., van Gijzen, T., Koese, B.,
Muilwijk, J., Turin, H. & Vermeulen, R. (2002) De loopkevers van
Nederland en Vlaanderen (Coleoptera, Carabidae). Jeugdbondsuit-
geverij, Nederland.
Burnham, K.P. & Anderson, D.R. 2003. Model Selection and Multimo-
del Inference: A Practical Information-Theoretic Approach. Springer
Science & Business Media.
den Boer, P.J. & van Dijk, T.S. (1994). Carabid beetles in a changing
environment. 94-6, Wageningen Agricultural University.
Ebert, G. (2005) Die Schmetterlinge Baden-Württembergs. Eugen Ulmer
KG, Stuttgart, Germany.
Ellis, W.N., Groenendijk, D., Groenendijk, M.M., Huigens, M.E.,
Jansen, M.G.M., van der Meulen, J., van Nieukerken, E.J. & de
Vos, R. (2013) Nachtvlinders belicht: dynamisch, belangrijk, bed-
reigd. De Vlinderstichting. Werkgroep Vlinderfaunistiek, Leiden,
Wageningen en.
Felix, R. & van Wielink, P. (2008) On the biology of Calodromius bifas-
ciatus and related species in De Kaaistoep (Coleoptera: Carabidae).
Entomologische Berichten,68, 198–209.
Fewster, R.M., Buckland, S.T., Siriwardena, G.M., Baillie, S.R. &
Wilson, J.D. (2000) Analysis of population trends for farmland birds
using generalized additive models. Ecology,81, 1970–1984.
García-Barros, E. (2015) Multivariate indices as estimates of dry body
weight for comparative study of body size in Lepidoptera. Nota Lepi-
dopterologica,38,59–74.
Gillespie, M.A., Alfredsson, M., Barrio, I.C., Bowden, J.J.,
Convey,P.,Culler,L.E.,Coulson,S.J.,Krogh,P.H.,Koltz,A.
M., Koponen, S., Loboda, S., Marusik, Y., Sandström, J.P.,
Sikes, D.S. & Høye, T.T. (2019) Status and trends of terrestrial
arthropod abundance and diversity in the North Atlantic region
of the Arctic. Ambio. https://doi.org/10.1007/s13280-019-
01162-5
Groenendijk, D. & Ellis, W. (2011) The state of the Dutch larger moth
fauna. Journal of Insect Conservation,15,95–101.
Habel, J.C. & Schmitt, T. (2018) Vanishing of the common species:
empty habitats and the role of genetic diversity. Biological Conserva-
tion,218, 211–216.
Habel, J.C., Samways, M.J. & Schmitt, T. (2019a) Mitigating the precip-
itous decline of terrestrial European insects: requirements for a new
strategy. Biodiversity and Conservation,28, 1343–1360.
Habel, J.C., Ulrich, W., Biburger, N., Seibold, S. & Schmitt, T. (2019b)
Agricultural intensification drives butterfly decline. Insect Conserva-
tion and Diversity,12, 289–295.
Habel, J.C., Segerer, A.H., Ulrich, W. & Schmitt, T. (2019c) Succession
matters: Community shifts in moths over three decades increases mul-
tifunctionality in intermediate successional stages. Scientific Reports,
9, 5586.
Hallmann, C.A., Sorg, M., Jongejans, E., Siepel, H., Hofland, N., Schwan, H.,
Stenmans, W., Müller, A., Sumser, H., Hörren, T., Goulson, D. & de
Kroon, H. (2017) More than 75 percent decline over 27 years in total flying
insect biomass in protected areas. PLoS ONE,12, e0185809.
Homburg, K., Drees, C., Boutaud, E., Nolte, D., Schuett, W.,
Zumstein, P., Ruschkowski, E. & Assmann, T. (2019) Where have
all the beetles gone? Long-term study reveals carabid species decline
in a nature reserve in Northern Germany. Insect Conservation and
Diversity,12, 268–277.
Janzen, D.H. & Hallwachs, W. (2019) Perspective: where might be many
tropical insects? Biological Conservation,233, 102–108.
Johnson, C.G. (1969) Migration and Dispersal of Insects by Flight.
Methuen and Co. Ltd., London.
Jonason, D., Franzen, M. & Ranius, T. (2014) Surveying moths using
light traps: effects of weather and time of year. PLoS ONE,9,
e92453.
Kleijn, D., Bink, R.J., ter Braak, C.J.F., van Grunsven, R., Ozinga, W.A.,
Roessink, I., Scheper, J.A., Schmidt, A.M., Wallis de Vries, M.F.,
Wegman, R., van der Zee, F.F. & Zeegers, T. (2018) Achteruitgang
insectenpopulaties in Nederland: trends, oorzaken en kennislacunes.
Wageningen Environmental Research rapport,2871.
Lister, B.C. & Garcia, A. (2018) Climate-driven declines in arthropod
abundance restructure a rainforest food web. Proceedings of the
National Academy of Sciences of the USA,115, E10397–E10406.
Martin, A.E., Graham, S.L., Henry, M., Pervin, E. & Fahrig, L. (2018)
Flying insect abundance declines with increasing road traffic. Insect
Conservation and Diversity,11, 608–613.
Meeus, J.H. (1991) Astronomical Algorithms. Willmann-Bell, Rich-
mond, Virginia.
Nakanishi, K., Yokomizo, H. & Hayashi, T.I. (2018) Were the sharp
declines of dragonfly populations in the 1990s in Japan caused by
fipronil and imidacloprid? An analysis of Hill’s causality for the case
of Sympetrum frequens.Environmental Science and Pollution
Research,25, 35352–35364.
Onkelinx, T., Devos, K. & Quataert, P. (2017) Working with popula-
tion totals in the presence of missing data comparing imputation
methods in terms of bias and precision. Journal of Ornithology,
158,603–615.
Potocký, P., Barto
nová, A., Beneš, J., Zapletal, M. & Konvi
cka, M.
(2018) Life-history traits of Central European moths: gradients of var-
iation and their association with rarity and threats. Insect Conservation
and Diversity,11, 493–505.
Russo, L., Stehouwer, R., Heberling, J.M. & Shea, K. (2011) The com-
posite insect trap: an innovative combination trap for biologically
diverse sampling. PLoS ONE,6, e21079.
Sabo, J.L., Bastow, J.L. & Power, M.E. (2002) Length–mass relation-
ships for adult aquatic and terrestrial invertebrates in a California
watershed. Journal of the North American Benthological Society,21,
336–343.
Sánchez-Bayo, F. & Wyckhuys, K.A. (2019) Worldwide decline of
the entomofauna: a review of its drivers. Biological Conservation,
232,8–27.
Schuch, S., Meyer, S., Bock, J., van Klink, R. & Wesche, K. (2019)
Drastic biomass loss in leafhopper and planthopper populations of
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
12 Caspar A. Hallmann et al.

various grasslands in Germany within six decades. Natur und Land-
schaft,94, 141–145.
Termaat, T., van Grunsven, R.H., Plate, C.L. & van Strien, A.J. (2015)
Strong recovery of dragonflies in recent decades in The Netherlands.
Freshwater Science,34, 1094–1104.
Termaat, T. & van Strien, A. (2015) Libellen, de grootste winst voorbij?
Vlinders,2,10–12.
Termaat, T., van Strien, A.J., van Grunsven, R.H.A., de Knijf, G.,
Bjelke, U., Burbach, K., Conze, K.-J., Goffart, P., Hepper, D.,
Kalkman, V.J., Motte, G., Prins, M.D., Prunier, F., Sparrow, D., van
den Top, G.G., Vanappelghem, C., Winterholler, M. &
WallisDeVries, M.F. (2019) Distribution trends of European dragon-
flies under climate change. Diversity and Distributions (in press), 25,
936–950.
Turin, H. (2000) De Nederlandse Loopkevers—Verspreiding en
oecologie, Vol. 3. Nationaal Natuurhistorisch Museum Naturalis,
Leiden.
Van Dyck, H., van Strien, A.J., Maes, D. & van Swaay, C.A.M. (2009)
Declines in common, widespread butterflies in a landscape under
intense human use. Conservation Biology,23, 957–965.
van Strien, A.J., van Swaay, C.A., van Strien-van Liempt, W.T., Poot, M.
J. & WallisDeVries, M.F. (2019) Over a century of data reveal more
than 80% decline in butterflies in The Netherlands. Biological Conser-
vation,234, 116–122.
van Swaay, C., Bos, G., van Grunsven, R., Kok, J., Huskens, K., van
Deijk, J. & Poot, M. (2018). Vlinders en libellen geteld. Jaarver-
slag 2017. Rapport VS2018.006, De Vlinderstichting,
Wageningen.
van Wielink, P.S. (2017a) Harmonia axyridis (Coleoptera: Coccinelli-
dae): 13 jaar gevolgd met lichtvangsten in De Kaaistoep, Noord-Bra-
bant. Entomologische Berichten,77,97–105.
van Wielink, P.S. (2017b) Negentien jaar lichtvangsten van lieveheers-
beestjes in De Kaaistoep (Coleoptera: Coccinellidae). Entomologische
Berichten,77, 127–139.
van Wielink, P.S. & Spijkers, H. (2012). Fauna in de Oude en Poppelsche
Leij. Pages 41–54 in Peeters, T., van Eck, A. & Cramer T., editors.
Natuurstudie in De Kaaistoep Verslag 2012, 18e onderzoeksjaar.
TWM Gronden BV, Natuurmuseum Brabant and KNNV-afdeling
Tilburg, Tilburg.
van Wielink, P. & Spijkers, H. (2013) Insects nightly attracted to light at
a single site in De Kaaistoep, The Netherlands. Orders, families and
species identified in 1995–2011. Entomologische Berichten,73,
200–214.
Waring, P. & Townsend, M. (2015) Nachtvlinders, veldgids met alle in
Nederland en België voorkomende soorten. Tirion Uitgevers, Baarn
<www.lepidoptera.se>, 18th April 2019.
White, G.C. & Bennetts, R.E. (1996) Analysis of frequency count data
using the negative binomial distribution. Ecology,77, 2549–2557.
Wood, S.N. (2006) Generalized Additive Models: an Introduction with
R. Chapman and Hall/CRC.
Wood, S.N. (2008) Fast stable direct fitting and smoothness selection for
generalized additive models. Journal of the Royal Statistical Society:
Series B,70, 495–518.
Zahrádková, S., Soldán, T., Bojková, J., Helešic, J., Janovska, H. &
Sroka, P. 2009. Distribution and biology of mayflies (Ephemeroptera)
of The Czech Republic: present status and perspectives. Aquatic Insects
31(sup1):629–652.
Accepted 12 August 2019
Editor: Simon Leather; Associate Editor: Christopher Hassall
© 2019 The Authors. Insect Conservation and Diversity published by John Wiley & Sons Ltd on behalf of Royal Entomological
Society., Insect Conservation and Diversity, doi: 10.1111/icad.12377
Declining abundance of beetles, moths and caddisflies in the Netherlands 13