Land-use intensity and the effects of organic farming on biodiversity: A hierarchical meta-analysis

Abstract

The benefits of organic farming to biodiversity in agricultural landscapes continue to be hotly debated, emphasizing the importance of precisely quantifying the effect of organic vs. conventional farming.

We conducted an updated hierarchical meta-analysis of studies that compared biodiversity under organic and conventional farming methods, measured as species richness. We calculated effect sizes for 184 observations garnered from 94 studies, and for each study, we obtained three standardized measures reflecting land-use intensity. We investigated the stability of effect sizes through time, publication bias due to the ‘file drawer’ problem, and consider whether the current literature is representative of global organic farming patterns.

On average, organic farming increased species richness by about 30%. This result has been robust over the last 30 years of published studies and shows no sign of diminishing.

Organic farming had a greater effect on biodiversity as the percentage of the landscape consisting of arable fields increased, that is, it is higher in intensively farmed regions. The average effect size and the response to agricultural intensification depend on taxonomic group, functional group and crop type.

There is some evidence for publication bias in the literature; however, our results are robust to its impact. Current studies are heavily biased towards northern and western Europe and North America, while other regions with large areas of organic farming remain poorly investigated.

Synthesis and applications. Our analysis affirms that organic farming has large positive effects on biodiversity compared with conventional farming, but that the effect size varies with the organism group and crop studied, and is greater in landscapes with higher land-use intensity. Decisions about where to site organic farms to maximize biodiversity will, however, depend on the costs as well as the potential benefits. Current studies have been heavily biased towards agricultural systems in the developed world. We recommend that future studies pay greater attention to other regions, in particular, areas with tropical, subtropical and Mediterranean climates, in which very few studies have been conducted.

Full text

REVIEW
Land-use intensity and the effects of organic farming
on biodiversity: a hierarchical meta-analysis
Sean L. Tuck
1
*, Camilla Winqvist
2
,Fl

avia Mota
3
, Johan Ahnstr
€
om
2
,
Lindsay A. Turnbull
1,3†
and Janne Bengtsson
2†
1
Department of Plant Sciences, University of Oxford, Oxford OX1 3RB, UK;
2
Section for Landscape and Soil Ecology,
Department of Ecology, SLU, Box 7044, Uppsala S-750 07, Sweden; and
3
Institute of Evolutionary Biology and
Environmental Studies, University of Zurich, Zurich 8057, Switzerland
Summary
1. The benefits of organic farming to biodiversity in agricultural landscapes continue to be
hotly debated, emphasizing the importance of precisely quan tifying the effect of organic vs.
conventional farming.
2. We conducted an updated hierarchical meta-analysis of studies that compared biodiversity
under organic and conventional farming methods, measured as species richness. We calcu-
lated effect sizes for 184 observations garnered from 94 studies, and for each study, we
obtained three standardized measures reflecting land-use intensity. We invest igated the stabil-
ity of effect sizes through time, publication bias due to the ‘file drawer’ problem, and consider
whether the current literature is representative of global organic farming patterns.
3. On average, organic farming increased species richness by about 30%. This result has been
robust over the last 30 years of published studies and shows no sign of diminishing.
4. Organic farming had a greater effect on biodiversity as the percentage of the landscape
consisting of arable fields increased, that is, it is higher in intensively farmed regions. The
average effect size and the response to agricultural intensification depend on taxonomic
group, functional group and crop type.
5. There is some evidence for publication bias in the literature; however, our results are
robust to its impact. Current studies are heavily biased towards northern and western Euro pe
and North America, while other regions with large areas of organic farming remain poorly
investigated.
6. Synthesis and applications. Our analysis affirms that organ ic farming has large positive
effects on biodiversity compared with conventional farming, but that the effect size varies
with the organism group and crop studied, and is greater in landscapes with higher land-use
intensity. Decisions about where to site organic farms to maximize biod iversity will, however,
depend on the costs as well as the potential benefits. Current studies have been heavily biased
towards agricultural systems in the developed world. We recommend that future studies pay
greater attention to other regions, in particular, areas with tropical, subtropical and Mediter-
ranean climates, in which very few studies have been conducted.
Key-words: agricultural management, diversity, farming systems, landscape complexity, spe-
cies richness
Introduction
Organic farming, in which insecticides, herbicides and
inorganic fertilizers are entirely or largely avoided, is gen-
erally thought to be more environmentally benign than its
conventional farming cousin. However, the overall bene-
fits of organic farming for biodiversity, the environment
in general, human health and food security have been
intensely debated in recent years (Bengtsson, Ahnstr
€
om &
Weibull 2005; Hole et al. 2005; Badgley et al. 2007; Mon-
delaers, Aertsens & Huylenbroeck 2009; Dobermann
*Correspondence author. E-mail: sean.tuck@plants.ox.ac.uk
†
These authors equally contributed to this work.
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited.
Journal of Applied Ecology 2014, 51, 746–755 doi: 10.1111/1365-2664.12219

2012; Reganold 2012; Tuomisto et al. 2012; Winqvist,
Ahnstr
€
om & Bengtsson 2012; Gabriel et al. 2013). The
debate turns on whether or not the decreased yields from
organic farms negate any local benefits, for example, to
biodiversity, that such methods deliver (Seufert, Rama-
nkutty & Foley 2012; but see Badgley et al. 2007). The
logic of this argument runs as follows: lower yields push
up food prices, and as a consequence, more wild or mar-
ginal land is brought into agricultural production. This
wild land is likely to have supported even higher biodiver-
sity than the organic farm; hence, begging the question, is
there an overall cost of organic farming to biodiversity?
Organic farming provides shared benefits to both
humans and wildlife, while conventional farming, at least
in the short term, maximizes yields – thus potentially
sparing wild lands elsewhere – therefore this argument is
often naively framed as ‘land sharing’ vs. ‘land sparing’
(Green et al. 2005; Vandermeer & Perfecto 2007; Fischer
et al. 2008; Phalan et al. 2011; Tscharntke et al. 2012;
Gabriel et al. 2013) although recently the debate has
moved away from such overly simplistic dichotomies. For
example, it has been argued that decisions about land
sparing vs. sharing are contingent on the landscape and
potential yields (Hodgson et al. 2010; Tscharntke et al.
2012; Gabriel et al. 2013). It is also clear that some
organisms are necessary on the farm to support essential
ecosystem services, for example, pollination and pest con-
trol, which contribute to yield. Therefore, species in farm-
land cannot be entirely sacrificed in order to preserve
biodiversity elsewhere. In addition, some species, particu-
larly in Europe where farming has been an integral part
of the landscape for thousands of years, thrive in exten-
sively managed farmland and are clearly threatened by
agricultural intensification (Chamberlain et al. 2000).
These species are an integral part of the European cul-
tural landscape, and their loss has provoked both public
and political outcry, leading the British Government, for
example, to pledge to reverse such declines by 2020. Thus,
organic farming, which generally increases both crop and
landscape heterogeneity, may be one component of a
land-sharing strategy, delivering wider ecosystem services
including amenity and conservation of culturally impor-
tant species (Vandermeer & Perfecto 2007; Gabriel et al.
2013). In this light, quantifying the precise benefits deliv-
ered by organic farming is essential.
While there is a general consensus that organic farming
increases biodiversity when compared to conventional
agriculture, the magnitude of this effect seems to vary
greatly, particularly among organism groups and across
landscapes (Bengtsson, Ahnstr
€
om & Weibull 2005; Bat

ary
et al. 2011; Winqvist, Ahnstr
€
om & Bengtsson 2012). Ben-
gtsson, Ahnstr
€
om and Weibull (2005) suggested that the
effects of organic farming on biodiversity were likely to
be greatest in intensively managed agricultural landscapes,
while Tscharntke et al. (2005) argued that agrienviron-
ment schemes would have larger effects in simple than in
complex landscapes. Some of these predictions have been
borne out by individual studies (Rundl
€
of & Smith 2006;
Rundl
€
of, Bengtsson & Smith 2008; Brittain et al. 2010;
Diek
€
otter et al. 2010; Bat

ary et al. 2011; Fischer et al.
2011; Flohre et al. 2011; Winqvist, Ahnstr
€
om & Bengts-
son 2012) and by meta-analysis in which landscapes were
classified as either simple or complex (Bat

ary et al. 2011).
However, different studies have defined ‘simple’ and ‘com-
plex’ in different ways, whereas it would be preferable to
have some more objective, continuous measurement of
land-use intensity with which to test these ideas more
fully.
While there have been previous meta-analyses compar-
ing conventional vs. organic farming and their biodiver-
sity and environmental impacts (Bengtsson, Ahnstr
€
om &
Weibull 2005; Bat

ary et al. 2011; Seufert, Ramankutty &
Foley 2012; Tuomisto et al. 2012), we believe that a new
analysis is still timely. First, previous meta-analyses have
not taken account of the hierarchical structure of the
data; secondly, a large number of new studies have been
published in recent years; and thirdly, we include here
three objective and standardized measures of land-use
intensity and landscape complexity measured on a contin-
uous scale, newly obtained for each of the studies. Using
an extended data set compared with Bengtsson, Ahnstr
€
om
and Weibull (2005), we can therefore ask the following
questions: (i) By how much does organic farming increase
biodiversity compared with conventional agriculture? (ii)
Do the effects of organic farming depend on the organism
or functional group, land-use intensity and structure, and
crop type? (iii) Has the reported effect size of organic
farming on biodiversity decreased or remained stable over
time? (iv) Is there evidence for publication bias in the lit-
erature, either because studies with negligible or negative
effects of organic farming remain unpublished or because
the present studies of organic farming, which are often
performed in Europe or the US (Bat

ary et al. 2011; Winq-
vist, Ahnstr
€
om & Bengtsson 2012), are unrepresentative
of the crops and regions in which organic farming is con-
ducted globally?
Materials and methods
DATA COLLECTION
We started with the species richness data set published in 2005 by
Bengtsson, Ahnstr
€
om and Weibull, which included 27 studies
published before December 2002. We expanded this data set to
include an additional 68 studies published between 2003 and
2011. Some of the additional data (2003–2009) were gathered for
an unpublished Master’s thesis (Mota 2010). Further studies from
2010 to 2011 were added by co-authors Ahnstr
€
om and Winqvist,
finishing the literature search by the end of 2011. The full data
set consists of 94 publications (see Appendix S1 in Supporting
information). When updating the data set of Bengtsson,
Ahnstr
€
om and Weibull (2005), we used the same keywords in ISI
web of knowledge: biodiversity, biological diversity, conventional
farming (agriculture) and organic farming (agriculture). We
searched for additional studies by scanning the bibliographies in
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
Hierarchical meta-analysis of organic farming 747

publications identified from our search. We followed the relevant
literature and discussed with colleagues throughout. Our data set
contains results from technical reports as well as peer-reviewed
journals. Although it is unlikely that the data we present are
complete, we believe these studies are an extensive and represen-
tative sample.
For a publication to be included in the analysis, it had to pro-
vide species richness data (n > 1) in both organic and conven-
tional systems. This could be in the form of raw data or the
mean species richness, standard deviation and sample size in both
farming systems. In some cases, we used other richness data pro-
vided in the publications – for example, Shannon H’ (M
€
ader
et al. 2002; Mart

ınez-S

anchez 2008) or richness of taxa higher
than species level (e.g. Galv

an et al. 2009; Crowder et al. 2010).
Unfortunately, many of the published studies do not meet these
criteria and therefore did not provide sufficient data to be useful
in a meta-analysis (see Appendix S2, Table S3 Supporting infor-
mation).
Organic agriculture is normally defined as any farming system
where the use of pesticides, herbicides and synthetic fertilizers is
prohibited or strictly limited. Organic farms often have other dif-
ferences, for example they tend to use more complex crop rota-
tions as a weed- and pest-control strategy and use animal
manure, green manure or compost in place of synthetic fertilizers.
Conventional systems, however, use pesticides and inorganic fer-
tilizers to various degrees and often use simplified crop rotations
and fewer crops. Due to the broad range of farming systems that
can be grouped within organic and conventional definitions, the
two farming systems are likely to differ between and within stud-
ies. However, despite these potential differences, we did not fur-
ther subdivide farming systems to avoid using more than two
treatments in the meta-analysis.
For each effect size, we extracted taxonomic and functional
data on the study organism(s). We also recorded (i) the sampling
unit of the species richness data (e.g. numbers per trap or tran-
sect), (ii) the sampling scale (plot, field or farm) and (iii) the crop
type. Data on species richness were extracted from the text, tables
or figures in publications using the program
GETDATA GRAPH DIGI-
TIZER
2.25 (Fedorov 2013) when necessary. Other measures of
variation presented in publications were converted to standard
deviations.
The information on taxonomic groups was used to create cate-
gorical covariates for different higher taxonomic units and eco-
logical functions. For taxonomic groups, we classified species as:
arthropods, birds, microbes and plants. Data on earthworms,
mammals, nematodes and protozoa were excluded from this
analysis due to small sample sizes (n < 5). For functional groups,
we classified species as producers (plants), herbivores, pollinators
(as adults), predators, soil-living decomposers and others (includ-
ing omnivores and organisms with variable or unknown func-
tional characteristics). The functional classification is based on
the idea that different organism groups may contribute to differ-
ent ecosystem services. We acknowledge that considerable uncer-
tainty about ecological function exists for several groups: carabid
beetles, for example, are often considered to provide pest control
(
€
Ostman, Ekbom & Bengtsson 2001, 2003), but many species are
known to be at least partly herbivorous seed eaters (Jonason
et al. 2013).
We also separated the data according to crop type. Given the
data, we were able to identify the following crop types: cereals,
grassland (usually permanent or semi-permanent leys or
pastures), mixed crops (comparison made across several different
crops), orchards, vegetable crops and miscellaneous (i.e. not spec-
ified precisely in the original study). Many studies include multi-
ple records for different organism groups or crop types on the
same farm. These were treated as distinct within-study observa-
tions and used to calculate separate effect sizes for subgroups. As
a result, our data set of 94 studies was subdivided into 184 obser-
vations (see Statistical analysis for more details).
LAND-USE INTENSITY METRICS
Three metrics of land-use intensity were collected using Google
Earth (2013). We conducted new landscape analyses for all
included studies in order to provide continuous standardized
measures of land-use intensity and complexity. We distinguished
between different land-use types: field (annual and perennial
crops, ley, grazed ley), pasture (perennial grassland used for graz-
ing), forest (including clear-cuts), wetland, water, rural, urban
and permanent line elements (e.g. ditches, hedge rows, roads
etc.). Using these land cover classifications, we calculated (i) %
arable fields – the proportion of the landscape covered by arable
fields; (ii) number of habitats – the number of distinguishable
habitats found in the landscape; and (iii) average field size – the
average size of arable fields in the landscape. The percentage of
arable fields is a measure of land-use intensity, while the number
of habitats represents landscape complexity. However, an inten-
sively farmed region is likely to include fewer habitats than a
more extensively farmed area. The average field size may reflect
the overall extent of farming on the landscape but, depending on
local farming practices, not necessarily farming intensity.
To calculate the three metrics, we first identified a standardized
sampling space at each location based on descriptions in the ori-
ginal publications. Where coordinates were not provided, we
identified an area that we were confident, included the study area
based on descriptions in the text. We then identified a central
measuring point, making sure it was placed in a landscape with
agricultural fields, and the radius (in metres) defining the appro-
priate area for sampling around this point. If no information
about the area of the study region was available, we visually
examined the Google Map image and set the radius so that the
included landscape was representative of its complexity (and simi-
lar to the landscape closest to the central point). We then ran-
domly placed five 1-km transects within this study region. The
positions of the five transects were defined by sets of three ran-
domly generated numbers. The first number, randomly selected
between 0 (central measuring point) and the radius of the study
region, denoted how many metres from the central point the
starting point of each transect would be situated. The second
number specified the angle (degrees), defining the direction rela-
tive to the central point for which the start point of the transect
should be placed. Combined, these two random numbers created
a bearing, from the centre of the study region, that defined the
transect location. The final number would randomly select
between 0, 45, 90, and 180 degrees to specify the angle at which
the transect should be drawn, 500 m to each side of the start
point. Transects were not allowed to cross. Our measures of land-
scape complexity and land-use intensity were calculated for each
transect, extracted directly from Google Earth and input to our
data base and averaged to give mean values of each metric for
each study or substudy region. The transects sampled were line
transects with no surrounding buffer.
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
748 S. L. Tuck et al.

In some cases, several studies had been conducted in the same
area, in which case the same landscape data were used. When
publications that garnered multiple observations had been con-
ducted in multiple regions, and data specified per region, we col-
lected landscape data per region. If the study region was not
specified at all – but only the country – we used the mean values
of all other studies in that country. The Google Maps analysed
were always the most recent images available. This represents one
caveat in our landscape analysis: for older studies, there is a time
lag between the date the study was conducted and the date our
landscape data were collected. Many of the early studies were
conducted in Europe, a region that we would expect to show the
least landscape change (in agricultural areas) over the relevant
time span.
STATISTICAL ANALYSIS
Our effect size is the log response ratio, which quantifies the pro-
portional difference between mean species richness in conven-
tional and organic farming (Hedges, Gurevitch & Curtis 1999).
On the log scale, an effect size of 0 means no difference and a
positive value means that the organic farm has higher species
richness than the conventional farm. The log response ratio dis-
plays bias at small sample sizes, when the normal approximation
to the distribution of the effect size deviates from the exact distri-
bution. To assess the appropriateness of this approximation,
√n∙l/r for both mean values within each effect size should be
generally >3 (Hedges, Gurevitch & Curtis 1999). In our data set,
only 10% of effect sizes fall below 3, while ~71% of scores exceed
6, and hence, the log response ratio is appropriate.
Our analysis was carried out using
R 3.0.1 (R Development
Core Team 2013) with the
R package metahdep (Stevens & Taylor
2009). The models were fitted to the data using the function
metahdep.HBLM. We analysed 184 separate observations for sub-
groups within studies – that is, different taxonomic groups or
crop types. A random effect was used to account for differences
across studies, for example, among farming systems included
within organic and conventional groups. A grand mean effect
size, across subgroups, was calculated using an intercept model
(Borenstein et al. 2009). Variables of interest, selected a priori,
were included in a metaregression to see whether they explained
any differences in biodiversity on organic vs. conventional farms.
These variables were functional groups, taxonomic groups, the
three landscape measures (see Land-use intensity metrics), crop
types and scale of sampling (plot, field, or farm). Uncertainty in
the regression coefficients was quantified using 95% credible
intervals. Credible intervals were calculated by multiplying the
posterior standard error of the coefficients by the 95% point of a
t-distribution with N–p degrees of freedom. We estimated hetero-
geneity between effect sizes, s
2
. This estimates the proportion of
between-studies variance that is true variance, as opposed to
within-study sampling error. This heterogeneity measure was used
to estimate I
2
, the proportion of total variance that is due to true
heterogeneity among effect sizes (Higgins & Thompson 2002).
There is hierarchical dependence between multiple observations
within studies. Having several effect sizes obtained from the same
publication violates the assumption that effect sizes are indepen-
dent. A publication-level random effect allowed us to account for
the dependency of multiple within-study observations. The non-
independence among effect sizes gathered from the same publica-
tion was defined by specifying a covariance structure in the
study-specific random deviations, as parameterized by s
2
(Stevens
& Taylor 2009). Defining dependence groups meant that a large
group of within-study effect sizes with extreme effect sizes was
down-weighted, preventing them from having a dominant effect
on the overall result. By incorporating this hierarchical variance
structure, we could disentangle important differences between
organisms and crop types without assuming independence of
observations.
The potential for bias in published results in the literature to
skew synthesized results is seen as a common limitation for meta-
analyses (Borenstein et al. 2009; Gillman & Wright 2010). There
are two ways that bias could be introduced: (i) a tendency for
only ‘significantly positive’ results to be published – the ‘file
drawer problem’ or (ii) studies are not representative of the popu-
lation – that is, there are evidence gaps in the literature, where
the question has not been investigated in certain contexts. A sim-
ple ecological example would be a lack of studies representing a
system relative to its global importance; this is a bias produced
by consensus in the literature that is not founded on a representa-
tive sample of reality. Meta-analysis can provide a general quan-
titative synthesis. It should also describe bias in the literature and
indicate where that bias may lie. We investigated bias in both the
forms described above.
To investigate bias in the ‘file drawer’ context, we characterized
funnel plot asymmetry in the data. The funnel plot is based on
the assumption that studies with smaller sample sizes (and hence
higher sampling variance) are more likely to be skewed, because
they have lower statistical power; hence, negative and low-effect
results from small-sample studies are missing from the literature.
To produce a funnel plot from our hierarchical model, we plotted
the residuals against precision (inverse sampling standard error;
Nakagawa & Santos 2012). In combination with this funnel plot,
we conducted a trim and fill assessment, whereby it is assumed
that skew is due to publication bias and compensates for this by
‘filling in’ new effect sizes until the skew in the residuals is cor-
rected for. To investigate this further, we conducted a cumulative
meta-analysis – in which studies are progressively added to the
data set in the order of increasing sampling variance – which
qualitatively shows how quickly the overall mean stabilizes and
whether the final estimate is strongly affected by the less reliable
studies. We also estimated the slope of the relationship between
sampling variance and effect size. Combining these diagnostics
allowed us to explore asymmetry in the data and then, under the
assumption that this is due to publication bias, assess its impact
on our result. The cumulative meta-analysis approach was used
to assess change in the overall effect size over time by progres-
sively adding studies in order of publication year, and again by
estimating the slope of the relationship between publication year
and effect size. To investigate ‘evidence gap’ bias, we compared
our data set with global data on the area of organic farming
across the globe and for different crop types, collected from the
FAO website [Food & Agriculture Organization of the United
Nations (FAOSTAT) 2013]. We used this comparison to discuss
how representative the current literature is of global organic
farming trends.
Results
The overall mean log response ratio was 0296 (95% CI:
0231–0361); this indicates that species richness on
organic farms is on average 34% (95% CI: 26–43) higher
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
Hierarchical meta-analysis of organic farming 749

than conventional. The estimated standard deviation of
the true effect sizes, s, was 0304 (variance for
s = 00004). This true variance among effect sizes com-
prised an overwhelming proportion of total variance
(I
2
= 974%). These results reveal substantial heterogene-
ity among effect sizes, although many studies showed a
large positive effect of organic farming on biodiversity rel-
ative to conventional farming. The estimate for hierarchi-
cal dependence was positive, meaning that the covariance
among within-publication effect sizes will downweight
large groups of effect sizes that would otherwise have an
excessive effect on the overall result.
We found large differences in the effect of organic
farming on different taxonomic and functional groups
(Fig. 1a,b; Table S2, Supporting information). For exam-
ple, among taxonomic groups, plants benefited the most
from organic farming (Fig. 1b). Arthropods, birds and
microbes also showed a substantial positive effect. Disag-
gregating organisms into functional groups showed a vari-
ety of responses: among functional groups, the largest
effect size was found for pollinators while decomposers
showed little effect (Fig. 1a). The crop types showed vary-
ing responses, with large positive effect sizes in cereals
and mixed farming, and moderate positive effect sizes for
all others (Fig. 1c).
The percentage arable fields had a positive effect on
the magnitude of the effect size (slope log(RR) = 0442,
95% CI: 0089 to 0973; Fig. 2). To assess the sensitiv-
ity of this slope estimate to the largest (‘outlying’) effect
sizes, we removed the four data points with log(RR) >2
and reperformed the analysis; there was a small reduc-
tion in the slope estimate (0396). Other landscape met-
rics had slope estimates close to zero (number of habitats:
log(RR) = 0006, 95% CI: 0019 to 0031; average field
size: log(RR) = 0001, 95% CI: 0001 to 0002). When
the percentage of arable fields was fitted as an interac-
tion with functional group, there was substantial hetero-
geneity in the resulting slopes. However, there was
significant uncertainty in these estimates, possibly due to
small sample sizes within some functional groups; thus,
we choose to report this result qualitatively: increasing
landscape intensity affected the magnitude of the effect
size in the order: herbivores > ‘other’ > predators > pro-
ducers > decomposers > pollinators. The sampling scale
of species richness observation did not appreciably
change the effect size (farm = 0249, 95% CI: 0 161 –
0338; treatment contrasts with farm scale: field = 0139,
95% CI: 0002 to 0279; plot = 0017, 95% CI:
0222 to 0187).
The representation of different crop types in the meta-
analysis was comparable with the global FAO statistics;
there were similar proportions of cereals, vegetables and
(a)
(b)
(c)
Fig. 1. The difference in species richness (%) on organic farms,
relative to conventional, classified: (a) by functional group (n:
decomposers = 19, herbivores = 6, other = 27, pollinators = 21,
predators = 49, producers = 62), (b) by organism group (n: ar-
thropods = 89, birds = 17, microbes = 6, plants = 62) and (c) by
crop types (n: cereals = 100, grasses = 13, mixed = 40,
orchard = 9, unspecified = 6, vegetables = 16). The grand mean is
shown in black, accompanied by the black line. The dashed lines
show the zero line. 95% credible intervals are calculated from
posterior standard errors.
Fig. 2. The relationship between the effect size and the propor-
tion of the landscape covered by arable fields showing a regres-
sion slope with 95% confidence intervals.
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
750 S. L. Tuck et al.

orchards (fruit; Fig. 3a), although fibre and oil crops were
underrepresented. The geographical representation in our
data set, however, showed much less congruence
(Fig. 3b): Western and Northern Europe, and to some
degree North America, were highly overrepresented, while
studies were largely lacking from most other geographical
regions, especially Asia, Africa and Australia.
The funnel plot (Fig. 4a) showed some positive bias. A
trim and fill assessment of how publication bias could
impact our inference, after correcting for positive funnel
plot skew, produced a negligible reduction in the effect
size (00001, three studies added). This suggests that, if
publication bias is evident, the reported effect size is
robust to its impact. Investigating further, the cumulative
meta-analysis of effect sizes sorted by sampling variance
showed that less reliable studies caused the grand mean to
increase, but not drastically so (Fig. 4b). If we assume
that this was due to publication bias then the most con-
servative effect size estimate is 0190 (95% CI: 0135–
0246), which still corresponds to a >20% increased spe-
cies richness on organic farms. This was the minimum
value obtained from the cumulative plot and was reached
after c. 80 observations (out of 184) were included. This
reduced effect size did not greatly alter our interpretation
of the magnitude of organic farming’s positive effect on
biodiversity. The relationship between sampling variance
and the effect size had a positive slope (0022, 95% CI:
0056 to 0101), which confirms the positive association
seen in Fig. 4.
The cumulative meta-analysis plot for data sorted by
publication date (Fig. 4c) showed that the grand mean
effect size estimated from our model was robust over
time, although, interestingly, many of the earliest studies
reported very high effect sizes. The lack of change with
time was supported by a slope estimate close to zero
(0003, 95% CI: 0007 to 0013).
Discussion
Our updated meta-analysis shows that organic farming on
average increases biodiversity (measured as species rich-
ness) by about one-third relative to conventional farming.
This result has been robust over the last 30 years of pub-
lished studies and shows no sign of diminishing. Organic
farming is therefore a tried and tested method for increas-
ing biodiversity on farmlands and may help to reverse the
continued declines of formerly common species in devel-
oped nations (Burns et al. 2013). Similar results have been
previously obtained (Bengtsson, Ahnstr
€
om & Weibull
2005; Fuller et al. 2005; Hole et al. 2005; Bat

ary et al.
2011; Garratt, Wright & Leather 2011), but our study is
the most up to date, deals with the hierarchical structure
of multiple within-publication effect sizes and includes
standardized measures of land-use intensity and heteroge-
neity across all studies.
In other areas of biology and medicine, it has been
noted that, with the addition of further evidence, effect
sizes concerning a particular question often decrease over
time (Jennions & Møller 2002). This is thought to occur
because of initial publication bias against non-significant
or negative results that is eventually corrected. The effect
size in our new study is slightly lower than the one
reported in Bengtsson, Ahnstr
€
om and Weibull (2005);
however, our analysis reveals that the grand mean effect
size is robust over time (Fig. 4c). There is therefore no
sign of a dwindling effect size with the addition of further
evidence. This implies that the increase in diversity with
organic farming that we report here is robust, given the
C
er
eal
Grass
Mixed Misc.
Orchard
Veg
Cereal
Oil crops
Fibre crops
Pulses
Fruit
Veg
NAmerica
CAmerica
SAmerica
NEurope
S Europe
WC Europe
NZ-Australia
Africa
Asia
NAmerica
CAmerica
SAmerica
NEurope
SEurope
WC Europe
NZ-Australia
Meta-analysis data FAO data
Fig. 3. Top row: proportions of different
crop types present in the meta-analysis
data set compared with the frequency of
the most commonly grown organic crops
world-wide. Bottom row: geographical ori-
gin of studies in the meta-analysis data set
compared with the area under organic pro-
duction in different regions of the world.
FAO data obtained from their website
(FAOSTAT 2013).
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
Hierarchical meta-analysis of organic farming 751

choice of crops and study areas included (see below for a
discussion of the representativeness of our study).
LAND-USE INTENSITY EFFECTS
Many authors have speculated on and investigated the
importance of landscape characteristics in shaping the
likely effect of organic farming on biodiversity (Bengts-
son, Ahnstr
€
om & Weibull 2005; Rundl
€
of & Smith 2006;
Rundl
€
of, Bengtsson & Smith 2008; Rundl
€
of, Nilsson &
Smith 2008; Bat

ary et al. 2011). Here, we calculated
three standardized measures of land-use intensity and
heterogeneity for all studies: the proportion of arable
fields, the typical field size and the number of habitats.
Only the proportion of arable fields in the landscape had
any significant overall effect. The difference in diversity
between organic and conventional farming generally
increased with increasing proportion of arable fields,
although there was large variation around the estimated
slope. Some of this variance may be due to different
responses between functional groups (Bat

ary et al. 2011).
The slope of this relationship decreased in the order:
decomposers > ‘other’ > predators > herbivores > produc-
ers > pollinators, suggesting that the effect of organic
farming on predators is greater in intensively managed
landscapes, whereas the effect of organic farming on
pollinators does not increase much with land-use inten-
sity. These differences may be due to the importance of
local actions relative to regional actions and to the move-
ment of organisms and chemicals across the landscape.
For example, some pollinators are known to be sensitive
to certain pesticides (Goulson 2013), leading to an EU
moratorium on neonicotinoids. If an organic farmer
refrains from using pesticides, then local pollinator rich-
ness might increase; however, given that these chemicals
might drift substantially, and that pollinators on an
organic farm will likely visit neighbouring farms, the
impact of this local action might have no more effect in
an intensively managed landscape compared with an
extensive one.
ORGANISM GROUPS, CROP TYPES AND SPATIAL
SCALE
We expected that the magnitude of the positive effect of
organic farming would vary among organism groups, as
this has been found repeatedly (Bengtsson, Ahnstr
€
om &
Weibull 2005; Fuller et al. 2005; Bat

ary et al. 2011; Garr-
att, Wright & Leather 2011; Winqvist et al. 2011; Winqvist,
Ahnstr
€
om & Bengtsson 2012). As in previous studies, we
found that plants benefited most from organic farming,
probably because of restricted herbicide use (Roschewitz
et al. 2005; Rundl
€
of, Edlund & Smith 2010). Arthropods,
birds and microbes also benefited, with varying levels of
estimated confidence. Accordingly, most functional groups
– herbivores, pollinators, predators and producers – were
more diverse in organic farming, with the exception of
decomposers. The lack of positive effects on decomposers,
which are mostly soil fauna, is surprising given that there
are positive effects of organic farming on soil conditions
and soil carbon (M
€
ader et al. 2002; Gattinger et al. 2012).
This may be because variation in soil type and structure is
more important for soil organisms than the farming system
itself. Such interactions between factors influencing the
diversity and abundance of soil organisms would repay
more investigation. The strong positive effects of organic
farming on herbivores and pollinators are consistent with
other studies (Rundl
€
of & Smith 2006; Holzschuh, Steffan-
Dewenter & Tscharntke 2008; Rundl
€
of, Bengtsson & Smith
2008; Garratt, Wright & Leather 2011).
We found significant differences in the effect of organic
farming among crop types. In cereal fields, which com-
prised >50% of the studies, organic farming had large
Sampling variance
Publication date
Sample size (cumulative total)
Effect size (log response ratio) Effect size (log response ratio)
(a)
(b) (c)
Fig. 4. (a) Funnel plot showing asymmetry in the spread of resid-
uals around the mean, created using the R package meta (Schwar-
zer 2010). The dashed line shows 95% confidence limits. (b)
Cumulative meta-analysis forest plot of data sorted by increasing
sampling variance. (c) Cumulative meta-analysis forest plot of
data sorted by increasing publication date.
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
752 S. L. Tuck et al.

effects, significantly higher than in vegetable crops and
orchards (Fig. 1c). This might reflect the intensive man-
agement of conventional cereal crops, with repeated appli-
cations of inorganic fertilizers and fungicides. The effect
size in both vegetable crops and orchards, although posi-
tive, did not differ significantly from zero, but this could
be due to small sample sizes. A lower but still significant
effect was found in grasslands (pastures and permanent or
semi-permanent leys), which are generally not so inten-
sively managed. The number of studies in grasslands, veg-
etables and orchards was quite low, and we recommend
that these crops are given more attention in the future.
In a previous meta-analysis (Bengtsson, Ahnstr
€
om &
Weibull 2005), small-scale studies (on the plot or single
field scale) showed much larger effect sizes than studies
on larger spatial scales. However, we found negligible dif-
ferences across scales. This suggests that the general bene-
fit of organic farming is robust across sampling scales, in
contrast to recent work that suggests that this benefit
diminishes at larger scales (Gabriel et al. 2010; Crowder
et al. 2012). The previous meta-analysis result may have
been due to small sample size or publication bias, which
highlights the importance of updating meta-analyses with
additional evidence. We note that most of the recent stud-
ies have been conducted at the farm scale, which is the
most relevant scale for evaluating both organic farming as
an agrienvironmental scheme for biodiversity, and for the
sustainability of farming systems in general.
PUBLICATION BIAS
The funnel plot suggests a positively biased spread of
effect sizes (Fig. 4a), which could be interpreted as a ten-
dency for studies showing large positive effects of conven-
tional farming on biodiversity to remain unpublished.
However, an alternative interpretation may be that large
positive effects of organic farming occur occasionally,
while large positive effects of conventional farming are
exceptionally unlikely. This seems reasonable given the
nonlinear nature of many natural processes, for example
population growth, which could occasionally fuel very
large impacts of not controlling certain groups of organ-
isms. In any case, the positive bias is slight and has been
shown to not affect our result.
Previous studies of organic farming on biodiversity
have been strongly biased towards temperate Western and
Northern Europe and North America (Fig. 3), that is,
intensive farming systems in developed countries. There is
extremely limited data available from other areas of the
world, for example, Eastern Europe, Asia, Africa, Central
and Southern America, a bias also noted by Bat

ary et al.
(2011), Martin, Blossey and Ellis (2012), and Randall and
James (2012). We therefore recommend that studies of
organic farming practices on diversity in tropical and sub-
tropical areas (e.g. Deb 2009; Zhang et al. 2013) should
receive high priority. It is, for example, surprising that
there are no studies on organic bananas or cacao, despite
these products being widely available in European super-
markets. Mediterranean climates are also underrepre-
sented, although a few studies from California
(Drinkwater et al. 1995; Letourneau & Bothwell 2008;
Kremen, Iles & Bacon 2012) and South Africa (Kehinde
& Samways 2012) exist.
THE ORGANIC CONTROVERSY
The yields from organic farms are generally lower than
conventional yields, although some controversy exists con-
cerning the size of this effect and whether it is more
prominent in developed countries (Badgley et al. 2007; De
Ponti, Rijk & van Ittersum 2012; Dobermann 2012; Rega-
nold 2012; Seufert, Ramankutty & Foley 2012). As out-
lined in the introduction, this implies a potential trade-off
between biodiversity and crop yields. For example, Gab-
riel et al. (2013) in a study of cereal crops in Southern
England concluded that the benefits of organic farming to
biodiversity were entirely bought at the cost of reduced
yield. They further suggested that the lower yields of
organic farming may therefore have the unfortunate result
of increasing the total area of land under agricultural pro-
duction. However, there are other, often unmeasured,
potential positive environmental benefits of organic farm-
ing. For example, nitrogen and phosphorus pollution
caused by leaching from intensively managed fields is still
a major problem in many countries and incurs significant
costs to society (Heathwaite, Sharpley & Gburek 2000).
An overall evaluation of organic farming in relation to
crop yields therefore needs to account for the effects of
farming practice on a wider range of environmental fac-
tors (Mondelaers, Aertsens & Huylenbroeck 2009; Sand-
hu, Wratten & Cullen 2010; Gattinger et al. 2012;
Bommarco, Kleijn & Potts 2013).
SYNTHESIS AND RECOMMENDATIONS
This analysis affirms that organic farming usually has
large positive effects on average species richness compared
with conventional farming. Given the large areas of land
currently under agricultural production, organic methods
could undoubtedly play a major role in halting the contin-
ued loss of diversity from industrialized nations. The
effect of organic farming varied with the organism group
and crop studied, and with the proportion of arable land
in the surrounding landscape. We found larger effects in
cereals, among plants and pollinators, and in landscapes
with higher land-use intensity. Despite the fact that
organic farming has been suggested to have large effects
on soil conditions, its effects on soil organisms were
ambiguous and in general understudied. Finally, it is clear
that three decades of studying the effects of organic farm-
ing on biodiversity have been heavily biased towards agri-
cultural systems in the developed world, especially Europe
and North America. We therefore recommend that other
regions and agricultural systems are given much greater
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
Hierarchical meta-analysis of organic farming 753

attention. In particular, more studies are needed in tropi-
cal, subtropical and Mediterranean climates. Studies at
any scale would be beneficial: at the farm scale because
this is the economic unit of farming, and at the landscape
scale because this is the scale at which many organisms
respond. This would allow a more balanced and globally
relevant assessment of organic farming effects on biodi-
versity, ecosystem services, food production and agricul-
tural sustainability.
Acknowledgements
We thank authors who performed the reviewed studies (see Appendix S1
Supporting inform ation) and who corresponded with us during data col-
lection, and FAO for the publication of their data. We also thank Peter
Bat

ary and one anonymous reviewer for constructive feedback, and John
Stevens for help with calculating residuals. The work was funded by the
Swedish Research Council FORMAS and the Ekhaga Foundation. SLT
was supported by UK Natural Environment Research Council.
Author contributions
SLT, JB and LAT designed the analyses and drafted the
paper; SLT performed the analyses; JA, CW and FM col-
lected the data. The original idea for the study emerged
from discussions between JB, LAT, JA, FM and CW.
Data accessibility
Meta-analysis data and R script: DRYAD entry doi:
105061/dryad.609t7 (Tuck et al. 2014). FAO statistics on
organic farming coverage: FAOSTAT (2013).
References
Badgley, C., Moghtader, J., Quintero, E., Zakem, E., Chappell, M.J.,
Avil

es-V

azquez, K., Samulon, A. & Perfecto, I. (2007) Organic agricul-
ture and the global food supply. Renewable Agriculture and Food Sys-
tems, 22,86–108.
Bat

ary, P., B

aldi, A., Kleijn, D. & Tscharntke, T. (2011) Landscape-mod-
erated biodiversity effects of agri-environmental management: a
meta-analysis. Proceedings of the Royal Society B: Biological Sciences,
278, 1894–1902.
Bengtsson, J., Ahnstr
€
om, J. & Weibull, A.-C. (2005) The effects of organic
agriculture on biodiversity and abundance: a meta-analysis. Journal of
Applied Ecology, 42, 261–269.
Bommarco, R., Kleijn, D. & Potts, S.G. (2013) Ecological intensification:
harnessing ecosystem services for food security. Trends in Ecology &
Evolution, 28, 230–238.
Borenstein, M., Hedges, L.V., Higgins, J.P.T. & Rothstein, H.R. (2009)
Introduction to Meta-Analysis. John Wiley & Sons, Ltd, Chichester, UK.
Brittain, C., Bommarco, R., Vighi, M., Settele, J. & Potts, S.G. (2010)
Organic farming in isolated landscapes does not benefit flower-visiting
insects and pollination. Biological Conservation, 143, 1860–1867.
Burns, F., Eaton, M.A., Gregory, R.D., Al Fulaij, N., August, T.A., Biggs,
J. et al. (2013) State of nature report. The State of Nature partnership.
Chamberlain, D.E., Fuller, R.J., Bunce, R.G.H., Duckworth, J.C. &
Shrubb, M. (2000) Changes in the abundance of farmland birds in rela-
tion to the timing of agricultural intensification in England and Wales.
Journal of Applied Ecology, 37, 771–788.
Crowder, D.W., Northfield, T.D., Strand, M.R. & Snyder, W.E. (2010)
Organic agriculture promotes evenness and natural pest control. Nature,
466, 109–112.
Crowder, D.W., Northfield, T.D., Gomulkiewicz, R. & Snyder, W.E.
(2012) Conserving and promoting evenness: organic farming and fire--
based wildland management as case studies. Ecology, 93, 2001–2007.
De Ponti, T., Rijk, B. & van Ittersum, M.K. (2012) The crop yield gap
between organic and conventional agriculture. Agricultural Systems,
108,1–9.
Deb, D. (2009) Biodiversity and complexity of rice farm ecosystems: an
empirical assessment. Open Ecology Journal
, 2, 112–129.
Diek
€
otter, T., Wamser, S., Wolters, V. & Birkhofer, K. (2010) Landscape
and management effects on structure and function of soil arthropod
communities in winter wheat. Agriculture, Ecosystems & Environment,
137, 108–112.
Dobermann, A. (2012) Getting back to the field. Nature, 485, 176–177.
Drinkwater, L.E., Letourneau, D.K., Workneh, F., van Bruggen, A.H.C.
& Shennan, C. (1995) Fundamental differences between conventional
and organic tomato agroecosystems in California. Ecological Applica-
tions, 5, 1098.
Fedorov, S. (2013) GetData Graph Digitizer, v2.25. Moscow, Russia.
http://getdata-graph-digitizer.com/
Fischer, J., Brosi, B., Daily, G.C., Ehrlich, P.R., Goldman, R., Goldstein,
J. et al. (2008) Should agricultural policies encourage land sparing or
wildlife-friendly farming? Frontiers in Ecology and the Environment, 6,
380–385.
Fischer, C., Flohre, A., Clement, L.W., Bat

ary, P., Weisser, W.W.,
Tscharntke, T. & Thies, C. (2011) Mixed effects of landscape structure
and farming practice on bird diversity. Agriculture, Ecosystems & Envi-
ronment, 141, 119–125.
Flohre, A., Rudnick, M., Traser, G., Tscharntke, T. & Eggers, T. (2011)
Does soil biota benefit from organic farming in complex vs. simple land-
scapes? Agriculture, Ecosystems & Environment, 141, 210–214.
Food and Agriculture Organization of the United Nations (FAOSTAT).
(2013) http://faostat3.fao.org/home/index.html#DOWNLOAD.
Fuller, R.J., Norton, L.R., Feber, R.E., Johnson, P.J., Chamberlain, D.E.,
Joys, A.C. et al. (2005) Benefits of organic farming to biodiversity vary
among taxa. Biology Letters, 1 , 431–434.
Gabriel, D., Sait, S.M., Hodgson, J.A., Schmutz, U., Kunin, W.E. & Ben-
ton, T.G. (2010) Scale matters: the impact of organic farming on biodi-
versity at different spatial scales. Ecology Letters, 13, 858–869.
Gabriel, D., Sait, S.M., Kunin, W.E. & Benton, T.G. (2013) Food produc-
tion vs. biodiversity: comparing organic and conventional agriculture.
Journal of Applied Ecology, 50, 355–364.
Galv

an, G.A., Par

adi, I., Burger, K., Baar, J., Kuyper, T.W., Scholten,
O.E. & Kik, C. (2009) Molecular diversity of arbuscular mycorrhizal
fungi in onion roots from organic and conventional farming systems in
the Netherlands. Mycorrhiza, 19, 317–328.
Garratt, M.P.D., Wright, D.J. & Leather, S.R. (2011) The effects of farm-
ing system and fertilisers on pests and natural enemies: a synthesis of
current research. Agriculture, Ecosystems & Environment, 141, 261–270.
Gattinger, A., Muller, A., Haeni, M., Skinner, C., Fliessbach, A., Buch-
mann, N. et al. (2012) Enhanced top soil carbon stocks under organic
farming. Proceedings of the National Academy of Sciences, 109, 18226–
18231.
Gillman, L.N. & Wright, S.D. (2010) Mega mistakes in meta-analyses:
devil in the detail. Ecology, 91, 2550–2552.
Google Earth. (2013) http://www.google.co.uk/intl/en_uk/earth/.
Goulson, D. (2013) An overview of the environmental risks posed by neo-
nicotinoid insecticides. Journal of Applied Ecology, 50, 977–987.
Green, R.E., Cornell, S.J., Scharlemann, J.P.W. & Balmford, A. (2005)
Farming and the fate of wild nature. Science, 307, 550–555.
Heathwaite, L., Sharpley, A. & Gburek, W. (2000) A conceptual approach
for integrating phosphorus and nitrogen management at watershed
scales. Journal of Environmental Quality, 29, 158–166.
Hedges, L.V., Gurevitch, J. & Curtis, P.S. (1999) The meta-analysis of
response ratios in experimental ecology. Ecology, 80, 1150–1156.
Higgins, J.P.T. & Thompson, S.G. (2002) Quantifying heterogeneity in a
meta-analysis. Statistics in Medicine, 21, 1539–1558.
Hodgson, J.A., Kunin, W.E., Thomas, C.D., Benton, T.G. & Gabriel, D.
(2010) Comparing organic farming and land sparing: optimizing yield
and butterfly populations at a landscape scale. Ecology Letters, 13,
1358–1367.
Hole, D.G., Perkins, A.J., Wilson, J.D., Alexander, I.H., Grice, P.V. &
Evans, A.D. (2005) Does organic farming benefit biodiversity? Biological
Conservation, 122, 113–130.
Holzschuh, A., Steffan-Dewenter, I. & Tscharntke, T. (2008) Agricultural
landscapes with organic crops support higher pollinator diversity. Oikos,
117, 354
–361.
Jennions, M.D. & Møller, A.P. (2002) Relationships fade with time: a
meta-analysis of temporal trends in publication in ecology and evolu-
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
754 S. L. Tuck et al.

tion. Proceedings of the Royal Society of London Series B: Biological
Sciences, 269,43–48.
Jonason, D., Smith, H.G., Bengtsson, J. & Birkhofer, K. (2013) Landscape
simplification promotes weed seed predation by carabid beetles (Coleop-
tera: Carabidae). Landscape Ecology, 28, 487–494.
Kehinde, T. & Samways, M.J. (2012) Endemic pollinator response to
organic vs. conventional farming and landscape context in the Cape
Floristic Region biodiversity hotspot. Agriculture, Ecosystems & Envi-
ronment, 146, 162–167.
Kremen, C., Iles, A. & Bacon, C. (2012) Diversified farming systems: an
agroecological, systems-based alternative to modern industrial agricul-
ture. Ecology and Society, 17, 44.
Letourneau, D.K. & Bothwell, S.G. (2008) Comparison of organic and
conventional farms: challenging ecologists to make biodiversity func-
tional. Frontiers in Ecology and the Environment, 6, 430–438.
M
€
ader, P., Fliessbach, A., Dubois, D., Gunst, L., Fried, P. & Niggli, U.
(2002) Soil fertility and biodiversity in organic farming. Science, 296,
1694–1697.
Martin, L.J., Blossey, B. & Ellis, E. (2012) Mapping where ecologists
work: biases in the global distribution of terrestrial ecological observa-
tions. Frontiers in Ecology and the Environment, 10, 195–201.
Mart

ınez-S

anchez, J.C. (2008) The role of organic production in biodiver-
sity conservation in shade coffee plantations. PhD Thesis, Department
of Biology, University of Washington, USA.
Mondelaers, K., Aertsens, J. & Huylenbroeck, G.V. (2009) A meta-analy-
sis of the differences in environmental impacts between organic and
conventional farming. British Food Journal, 111, 1098–1119.
Mota, F.K.P. (2010) The effects of organic farming on biodiversity: a
meta-analysis. Unpublished Master’s thesis, Institute of Evolutionary
Biology and Environmental Studies, University of Z
€
urich, Z
€
urich, Swit-
zerland.
Nakagawa, S. & Santos, E.S.A. (2012) Methodol ogical issues and
advances in biological meta-analysis. Evolutionary Ecology, 26, 1253–
1274.
€
Ostman,
€
O., Ekbom, B. & Bengtsson, J. (2001) Landscape heterogeneity
and farming practice influence biological control. Basic and Applied
Ecology,
2, 365 –371.
€
Ostman,
€
O., Ekbom, B. & Bengtsson, J. (2003) Yield increase attributable
to aphid predation by ground-living polyphagous natural enemies in
spring barley in Sweden. Ecological Economics, 45, 149–158.
Phalan, B., Onial, M., Balmford, A. & Green, R.E. (2011) Reconciling
food production and biodiversity conservation: land sharing and land
sparing compared. Science, 333, 1289–1291.
R Development Core Team. (2013) R: A Language and Environment for
Statistical Computing. R Foundation for Statistical Computing, Vienna,
Austria.
Randall, N.P. & James, K.L. (2012) The effectiveness of integrated farm
management, organic farming and agri-environment schemes for con-
serving biodiversity in temperate Europe – a systematic map. Environ-
mental Evidence, 1,1–21.
Reganold, J.P. (2012) The fruits of organic farming. Nature, 485, 176.
Roschewitz, I., Gabriel, D., Tscharntke, T. & Thies, C. (2005) The effects
of landscape complexity on arable weed species diversity in organic and
conventional farming. Journal of Applied Ecology, 42, 873–882.
Rundl
€
of, M., Bengtsson, J. & Smith, H.G. (2008) Local and landscape
effects of organic farming on butterfly species richness and abundance.
Journal of Applied Ecology, 45, 813–820.
Rundl
€
of, M., Edlund, M. & Smith, H.G. (2010) Organic farming at local
and landscape scales benefits plant diversity. Ecography, 33, 514–522.
Rundl
€
of, M., Nilsson, H. & Smith, H.G. (2008) Interacting effects of
farming practice and landscape context on bumble bees. Biological Con-
servation, 141, 417–426.
Rundl
€
of, M. & Smith, H.G. (2006) The effect of organic farming on but-
terfly diversity depends on landscape context. Journal of Applied Ecol-
ogy, 43, 1121–1127.
Sandhu, H.S., Wratten, S.D. & Cullen, R. (2010) Organic agriculture and
ecosystem services. Environmental Science & Policy, 13,1–
7.
Schwarzer, G. (2010) Meta-analysis with R: package ‘meta’. R Package
Version 3.0-1 edn. http://cran.r-project.org/web/packages/meta/index.
html
Seufert, V., Ramankutty, N. & Foley, J.A. (2012) Comparing the yields of
organic and conventional agriculture. Nature, 485, 229–232.
Stevens, J.R. & Taylor, A.M. (2009) Hierarchical dependence in
meta-analysis. Journal of Educational and Behavioral Statistics, 34,
46–73.
Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I. & Thies, C.
(2005) Landscape perspectiv es on agricultural Intensification and biodi-
versity – ecosystem service management. Ecology Letters, 8, 857–874.
Tscharntke, T., Clough, Y., Wanger, T.C., Jackson, L., Motzke, I., Per-
fecto, I., Vandermeer, J. & Whitbread, A. (2012) Global food security,
biodiversity conservation and the future of agricultural intensification.
Biological Conservation, 151,53–59.
Tuck, S.L., Winqvist, C., Mota, F., Ahnstr
€
om, J., Turnbull, L.A. & Ben-
gtsson, J. (2014) Data from: land-use intensity and the effects of organic
farming on biodiversity: a hierarchical meta-analysis. Dryad Digi tal
Repository, doi:10.5061/dryad.609t7.
Tuomisto, H.L., Hodge, I.D., Riordan, P. & Macdonald, D.W. (2012)
Does organic farming reduce environmental impacts? – a meta-analysis
of European research. Journal of Environmental Management, 112,
309–320.
Vandermeer, J. & Perfecto, I. (2007) The agricultural matrix and a future
paradigm for conservation. Conservation Biology, 21, 274–277.
Winqvist, C., Ahnstr
€
om, J. & Bengtsson, J. (2012) Effects of organic farm-
ing on biodiversity and ecosystem services: taking landscape complexity
into account. Annals of the New York Academy of Sciences, 1249,
191–203.
Winqvist, C., Bengtsson, J., Aavik, T., Berendse, F., Clement, L.W.,
Eggers, S. et al. (2011) Mixed effects of organic farming and landscape
complexity on farmland biodiversity and biological control potential
across Europe. Jour nal of Applied Ecology, 48, 570–579.
Zhang, J., Zheng, X., Jian, H., Qin, X., Yuan, F. & Zhang, R. (2013)
Arthropod biodiversity and community structures of organic rice eco-
systems in Guangdong Province, China. Florida Entomologist, 96,1–9.
Received 19 November 2013; accepted 23 December 2013
Handling Editor: Ailsa McKenzie
Supporting Information
Additional Supporting Information may be found in the online version
of this article.
Appendix S1. List of studies included in the meta-analysis.
Appendix S2.
PRISMA flowchart showing the data collection deci-
sion process.
Table S1. Model estimates.
Table S2. Coefficient estimates for subgroups included in Fig. 1.
Table S3. List of studies rejected during data collection, with rea-
sons for rejection.
© 2013 The Authors. Journal of Applied Ecology published by John Wiley & Sons Ltd on behalf of British Ecological Society., Journal of
Applied Ecology, 51, 746–755
Hierarchical meta-analysis of organic farming 755