Effects of Farmland Heterogeneity On Biodiversity
Effects of Farmland Heterogeneity On Biodiversity
Effects of Farmland Heterogeneity On Biodiversity
Keywords: Pressure to increase food production to meet the demands of a growing human population can make con-
Agricultural intensification servation-motivated recommendations to limit agricultural expansion impractical. Therefore, we need to iden-
Agroecosystem tify conservation actions that can support biodiversity without taking land out of production. Previous studies
Cropland suggest this can be accomplished by increasing “farmland heterogeneity”—i.e. heterogeneity of the cropped
Landscape context
portions of agricultural landscapes—by, for example, decreasing field sizes. However, it is not yet clear whether
Land management
Land sharing
policies/guidelines that promote farmland heterogeneity will be as effective as those targeting farming practices.
Here, we estimated the relative effects of six practices—annual/perennial crop, fertilizer use, herbicide use,
insecticide use, tile drainage, and tillage—versus two aspects of farmland heterogeneity—field size and crop
diversity—on the diversity of herbaceous plants, woody plants, butterflies, syrphid flies, bees, carabid beetles,
spiders, and birds in rural eastern Ontario, Canada. The strength of effect of farming practices and farmland
heterogeneity varied among taxonomic groups. Nevertheless, we found important effects of both farming
practices and farmland heterogeneity on the combined (multi) diversity across these groups. In particular, we
found greater multidiversity in untilled, perennial crop fields than tilled, annual crop fields, and greater mul-
tidiversity in agricultural landscapes with smaller crop fields and less diverse crops. The directions of effect of
these variables were generally consistent across individual taxonomic groups. For example, richness was lower
in landscapes with larger fields and more diverse crops than in landscapes with smaller fields and less diverse
crops for all taxa except spiders. The negative effect of crop diversity on multidiversity and the richness of most
of the studied taxa indicates that this aspect of farmland heterogeneity does not necessarily benefit wildlife
species. Nevertheless, a compelling implication of this study is that it suggests that policies/guidelines aimed at
reducing crop field sizes would be at least as effective for conservation of biodiversity within working agri-
cultural landscapes as those designed to promote a wildlife-friendly farming practice.
Corresponding author.
⁎
E-mail addresses: [email protected] (A.E. Martin), [email protected] (S.J. Collins), [email protected] (S. Crowe),
[email protected] (J. Girard), [email protected] (I. Naujokaitis-Lewis), [email protected] (A.C. Smith),
[email protected] (K. Lindsay), [email protected] (S. Mitchell), [email protected] (L. Fahrig).
https://doi.org/10.1016/j.agee.2019.106698
Received 29 April 2019; Received in revised form 20 September 2019; Accepted 27 September 2019
0167-8809/ © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
heterogeneity can be increased by decreasing crop field sizes. De- relative effects of six farming practices—annual/perennial crop, ferti-
creasing crop field sizes increases the length of edges between different lizer use, herbicide use, insecticide use, tile drainage, and tillage—-
crop types and between crop and natural/semi-natural land cover types. versus field size and crop diversity on the combined (multi) diversity
Farmland heterogeneity can also be increased by increasing crop di- and species richness of herbaceous plants, woody plants, butterflies,
versity, i.e. planting more types of crops and having more even re- syrphid flies, bees, carabid beetles, spiders, and birds in farmland.
presentation of the crop types in the landscape. Given the results of previous comparisons of the effects of individual
Empirical studies have found support for this hypothesis. More farming practices versus farmland heterogeneity on the richness of in-
species and/or higher abundances are found in agricultural landscapes dividual taxa, and the strong evidence of positive effects of wildlife-
with smaller fields, when controlling for total farmland area. Such ef- friendly practices on species richness, we expected farming practices to
fects have been found in a range of taxa, including amphibians, mam- have larger effects on multidiversity than farmland heterogeneity.
mals, birds, insects, spiders, and plants (Collins and Fahrig, 2017;
Ekroos et al., 2019; Fahrig et al., 2015; Hass et al., 2018; Kirk et al.,
2011; Monck-Whipp et al., 2018; Reynolds et al., 2018; Sirami et al., 2. Methods
2019; Zhou et al., 2018). Positive effects of crop diversity on wildlife
have also been reported (Collins and Fahrig, 2017; Ekroos et al., 2019; 2.1. Overview
Lee and Goodale, 2018; Monck-Whipp et al., 2018; Novotný et al.,
2015; Palmu et al., 2014; Redlich et al., 2018a; Reynolds et al., 2018); This study builds on a previous study of the effects of farmland
however, these effects have been less consistent (and often weaker) heterogeneity on biodiversity (Fahrig et al., 2015). That study sampled
than those of field size (Ekroos et al., 2019; Fahrig et al., 2015; Hass biodiversity in crop field edges and field interiors within 93, 1 × 1 km
et al., 2018; Kirk and Freemark Lindsay, 2017; Redlich et al., 2018b; farmland-dominated sampling areas (hereafter shortened to “sampling
Wilson et al., 2017; Zhou et al., 2018). Taken together, these studies areas”) in rural eastern Ontario, Canada (Appendix A). The dominant
suggest that policies/guidelines that encourage farmers to increase crop types in eastern Ontario are hay (30% of farmland), pasture (24%),
farmland heterogeneity should benefit biodiversity in agricultural corn (21%), and soybean (19%; OMAFRA, 2011). The sampling areas
landscapes. were primarily selected to represent the regional variation in two as-
However, a key knowledge gap is whether such policies/guidelines pects of farmland heterogeneity: crop field size (measured as the mean
will be as effective as those targeting other aspects of agricultural in- size of crop fields) and crop diversity (Shannon diversity of crop types).
tensification, namely conventional farming practices such as the use of Herbaceous plants, woody plants, butterflies, syrphid flies, bees, car-
pesticides. In particular: what are the relative effects of farming prac- abid beetles, spiders, and birds were surveyed in 4–8 sampling sites per
tices versus farmland heterogeneity on wildlife found in the cropped area (depending on the taxonomic group). Half the sampling sites were
portion of the landscape (i.e. farmland)? Many studies have docu- at field edges and half in field interiors, except for woody plants and
mented negative effects of conventional farming practices on wildlife, birds which were only sampled at field edges. The sampled taxa were
including those in farmland (Tuck et al., 2014 and references therein). selected to represent a range of potential responses to farmland het-
Some studies have evaluated the relative effects of practices versus erogeneity, a range of ecosystem services, and for ease of sampling. The
landscape heterogeneity/complexity on wildlife in farmland (typically data from the bird surveys in field edges and herbaceous plant, but-
indexed as crop cover, or the amount/diversity of semi-natural habitats; terfly, syrphid fly, bee, carabid beetle, and spider surveys in field in-
e.g. Gagic et al., 2017; Roschewitz et al., 2005; Tamburini et al., 2016). teriors were previously used to evaluate the relative effects of field size
However, few have evaluated the relative effects of farming practices and crop diversity on alpha, beta, and gamma diversity and on relative
versus farmland heterogeneity on wildlife in the cropped portion of the abundance per 1 × 1 km sampling area (Fahrig et al., 2015).
landscape. Additionally, the management implications of some pre- Here, we use data from a subset of the biodiversity surveys, in
vious studies are limited because they compared only the aggregated combination with additional data on farm management practices, from
effect of many farming practices to the effects of individual farmland 112 sampling sites (57 field edge and 55 field interior sites) within 32 of
heterogeneity variables (Palmu et al., 2014; Sirami et al., 2019). Thus, the sampling areas (Appendix A). For each sampling site we calculated
one cannot use these study results to pinpoint which specific farming ‘multidiversity’ (Allan et al., 2014), i.e. the mean biodiversity across the
practices or aspects of farmland heterogeneity have the largest effects studied taxa. In field edges multidiversity was based on herbaceous
on biodiversity. Only a few studies have investigated the relative effects plants, woody plants, butterflies, syrphid flies, bees, carabid beetles,
of individual farming practices versus field size on wildlife in farmland spiders, and birds. In field interiors multidiversity was based on her-
(Geiger et al., 2010) or individual farming practices versus crop di- baceous plants, butterflies, syrphid flies, bees, carabid beetles, and
versity on wildlife in farmland (Chiron et al., 2014). These studies spiders. There are no publicly available data on farming practices in our
suggest that farmland heterogeneity is typically less important than study region at a resolution appropriate for our analyses. Instead, we
fertilizer or pesticide use. However, there may be variability in re- collected data on farm management practices at biodiversity sampling
sponses among taxa. For example, Billeter et al. (2008) found that plant sites using landowner interviews. We then estimated the relative effects
and bird richness in agricultural landscapes (including species found in of six farming practices—annual/perennial crop, fertilizer use, herbi-
crop fields, semi-natural patches, and linear features such as hedge- cide use, insecticide use, tile drainage, and tillage—and two measures
rows) were significantly affected by a farming practice (fertilizer use) of farmland heterogeneity—field size and crop diversity in the land-
but not crop diversity; in contrast, arthropod richness in agricultural scapes surrounding the sampling sites—on multidiversity using a mixed
landscapes was significantly affected by crop diversity but not farming effects model. We also included two additional fixed effects: crop cover
practices. To maximize biodiversity in working agricultural landscapes (the proportion of the landscape covered by crop fields), and the lo-
we want to identify the management actions that have the most benefit cation of the sampling site within the field (field edge or field interior;
for species across taxa. Thus, there is a need for additional study of the hereafter referred to as the “sampling location”), to control for their
relative effects of individual farming practice and farmland hetero- effects on biodiversity. Sampling area was included as a random effect,
geneity variables on farmland biodiversity, particularly to understand to account for repeated sampling within each area. We repeated this
the relative importance of effects across taxonomic groups. analysis using species richness of each sampled taxon. Standardized
Here we address this knowledge gap, estimating the relative effects model coefficients from the mixed effects models were used to estimate
of individual farming practices, field size, and crop diversity on biodi- the effect sizes.
versity in rural eastern Ontario, Canada, when controlling for the
amount of farmland (i.e. crop cover). Specifically, we estimate the
2
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
2.2. Selection of farmland-dominated sampling areas and sampling sites (d) ≥50 m from the edge of the 1 × 1 km area. Field edge sites were
within sampling areas also selected to minimize tree cover in the adjacent non-crop vegeta-
tion. These conditions were confirmed by visual surveys of each site
Fahrig et al. (2015) selected a set of 93, 1 × 1 km sampling areas in prior to the field season. If criteria were not met during visual surveys,
rural eastern Ontario, Canada (Appendix A), according to criteria de- or if landowner permission for surveys was denied, another random site
tailed in Pasher et al. (2013). The areas were selected to: (a) be re- was selected. Each of the four field edge sites was paired with a field
presentative of the variability in field size and crop diversity across interior sampling site in the same crop field, ∼25 m from the field edge
eastern Ontario; (b) be spatially independent, i.e. non-overlapping with site.
minimal spatial autocorrelation of field size and crop diversity across
areas; (c) be dominated by agricultural land use (44–95% of the area 2.3. Biodiversity surveys
used in crop production); and (d) have weak correlations between field
size, crop diversity, and crop cover across the sampling areas. The 32 Biodiversity surveys were conducted in each sampling area for each
sampling areas used in the present study were similarly dominated by of eight taxonomic groups: herbaceous plants, woody plants, butterflies,
agricultural land use (51–92% of the landscape used in crop produc- syrphid flies, bees, carabid beetles, spiders, and birds (as described
tion); however, the correlations between field size, crop diversity, and below). Each sampling area was sampled in one year (2011 or 2012), in
crop cover were generally stronger than in the complete dataset (field random order. However, for logistical reasons, all sampling sites within
size and crop diversity r = −0.68; field size and crop cover r = 0.55; a given sampling area were sampled on the same day. We identified
crop diversity and crop cover r = −0.35). individuals to the lowest taxonomic level possible (species, genus, or
For the full biodiversity dataset, eight sampling sites were selected family). The majority of samples were identified to species; thus, to
within each sampling area (Fig. 1a,b), four at field edges and four in simplify, we hereafter refer to this as “species richness”. Species rich-
field interiors. To select these sites, all field edges within each 1 × 1 km ness at a sampling site was the total number of species observed across
area were delineated using 2008 aerial photographs (Ontario Ministry all surveys at that site.
of Natural Resources, 2010) in ArcMap v10.0 (ESRI, Redlands, Cali-
fornia). Four sampling sites at field edges were then randomly selected, 2.3.1. Herbaceous plants
under the following constraints. Sites were only selected if they were: Herbaceous plants were surveyed along a 50 m transect at each of
(a) adjacent to another crop field; (b) ≥200 m from all other sites; (c) the eight sampling sites per sampling area. Surveys were conducted
≥50 m from non-agricultural land uses (e.g. houses, roads, forests); and twice per sampling site, first between May 24 and July 9 and again
3
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
Fig. 2. (a) Examples of the alternative layouts of field edge and field interior transects used in biodiversity surveys. Field edge transects were at the very edge of the
field and always aligned with the field edge. Where possible, the field interior transect was placed parallel to its field edge transect pair, at a distance of ∼25 m into
the field. Otherwise, we used a transect layout that preserved the approximate distance from the field edge transect and the total transect length. (b–i) Depiction of
the sampling design for each taxonomic group. Note that in b, d, and e, half of the surveyed area was in the crop field and the other half was in the field border
vegetation. Imagery taken from Google Earth Pro v. 7.3.0.3832.
between July 17 and August 30 in both sampling years. Field edge herbaceous plants (see 2.3.1. Herbaceous plants) at three of the four
transects were aligned with the field edge (Fig. 2a). Where possible, the paired sampling sites per area. Surveys were conducted twice per
field interior transect was placed parallel to its paired field edge sampling site, first between June 27 and August 3 and again between
transect. However, it was not possible to survey in this manner in fields August 4 and 31 in both sampling years. Surveys were conducted be-
where crop rows ran perpendicular to the field edge: in these cases, we tween 830 and 1730 on sunny days when winds were low (≤3 on the
used a transect layout that preserved the approximate distance from the Beaufort scale) and temperatures were > 15 °C. The surveyor walked a
field edge transect and total transect length (Fig. 2a). The surveyor transect once per survey at 5 m/min., and recorded all butterfly species
walked each transect once per survey and recorded all herbaceous plant seen within an imaginary 5 m3 box in front of the observer, i.e. within
species seen within 1 m of the transect. In field edge transects this 5 m from the ground, 2.5 m of either side of the transect, and 5 m in
meant that half the width of the transect was in the field and the other front of the observer. As for the herbaceous plant surveys, this meant
half was in the adjacent vegetation (Fig. 2b). that half the width of a field edge transect was in the field and the other
half was in the adjacent vegetation (Fig. 2d).
2.3.2. Woody plants
We surveyed for woody plants along the field edge transects used for 2.3.4. Syrphid flies
herbaceous plants (see 2.3.1. Herbaceous plants). Woody plants were Syrphid fly surveys were conducted immediately following the
surveyed once per sampling site between July 17 and August 30 in both butterfly surveys, along the same transects (see 2.3.3. Butterflies). The
sampling years. The surveyor recorded the species of all woody in- surveyor walked a transect once per survey at 5 m/min., collecting all
dividuals > 1 m tall found across the width of the border between the observed syrphids within an imaginary 2 m3 box in front of the observer
two crop fields (Fig. 2c). using a 38-cm diameter butterfly net (Fig. 2e). Collected syrphids were
euthanized with ethyl acetate and frozen until they could be pinned and
2.3.3. Butterflies identified using Shannon (1923), Curran and Fluke (1926), Curran
We surveyed for butterflies along the same transects used for (1951), Telford (1970), Vockeroth and Thompson (1987), and
4
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
5
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
Table 1 (see above). We used a generalized linear mixed effects model with a
Total number of species across all sampling sites, and the mean, minimum, and negative binomial distribution and log link, including crop cover as a
maximum species richness per site for each taxonomic group. Multidiversity fixed effect and sampling area as a random effect in each model. We
was calculated as the mean standardized species richness, averaged across all also included the sampling location as a fixed effect for all taxa except
taxa sampled at a given sampling site.
for woody plants and birds, because these taxa were sampled in field
Species richness edges only. We then used these statistical models to predict how much
species richness changed in response to a change in each farming
Taxa Total Mean Minimum Maximum
practice and measure of farmland heterogeneity. However, to avoid
herbaceous plants 165 20.97 2 51 implying that our predictions are of actual rather than relative species
woody plants 41 3.12 0 12 richness, we opted to express each change in richness as a percentage of
butterflies 25 2.79 0 7 the maximum species richness observed across all sampling sites
syrphid flies 29 2.17 0 10
(Table 1). For example, we calculated the increase in species richness
bees 87 7.10 1 20
carabid beetles 45 3.76 0 11
when moving from an annual to perennial field as the (predicted
spiders 60 3.83 0 12 richness in a perennial field – predicted richness in an annual field) /
birds 34 2.28 0 6 maximum observed richness at any sampling site × 100%.
multidiversity – −0.05 −1.08 1.40 We also evaluated whether the effects of farming practices and
farmland heterogeneity on multidiversity and the richness of individual
taxa differed between field edges and interiors. To determine whether
then averaging across the taxa for each site. Thus a positive value in-
effects of annual/perennial crop on multidiversity differed between
dicates higher than average multidiversity, and a negative value in-
field edges and interiors, we used linear mixed effects to model multi-
dicates lower than average multidiversity. Note that we did not simply
diversity as a function of 10 fixed effects—six farming practices, two
add up the total number of species per site because this would result in
farmland heterogeneity variables, crop cover, and sampling loca-
a measure that was highly dependent on the most diverse taxon (her-
tion—plus an interaction between annual/perennial crop × location.
baceous plants; see Table 1).
We also included a random effect of sampling area in the model. We
We included each of our three landscape context variables (field
then ran similar models to test for interaction effects between each of
size, crop diversity, and crop cover) at its estimated scale of effect, i.e.
the remaining farming practice/farmland heterogeneity variables and
the spatial extent within which the landscape variable has the strongest
sampling location. We did not include all interaction terms within a
influence on multidiversity. We estimated the scale of effect for each
single model because of sample size limitations. We repeated the above
landscape context variable while controlling for the effects of farming
analyses for species richness of all taxa sampled in field edges and in-
practices and the other landscape context variables, and allowed for
teriors, with the exception that here we used generalized linear mixed
selection of different spatial extents for different variables. To do so, we
effects with a negative binomial distribution and log link.
used linear mixed effects to model multidiversity as a function of the six
We tested for relationships among our 10 predictor variables. We
farming practices and three landscape context variables 64 times. That
used generalized linear mixed effects with a binomial distribution and
is, we ran one model for each unique combination of the four spatial
logit link for pairwise comparisons of the binary farming practices
extents (250, 500, 750, and 1000 m) and the three landscape context
variables, including sampling area as a random effect to account for
variables (as, for example, in Boscolo and Metzger, 2009), i.e. all
non-independence and spatial autocorrelation of measurements within
variables at 250 m, field size and crop diversity at 250 m and crop cover
a sampling area. We used linear mixed effects for pairwise comparisons
at 500 m, field size at 250 m and crop diversity and cover at 500 m, etc.
of continuous landscape context variables and for pairwise comparisons
(see Appendix E for the full list of models). We also included sampling
of farming practice to landscape variables, including sampling area as a
location (field edge or field interior) as a fixed effect and sampling area
random effect. We indexed the variance shared by two predictors as the
as a random effect in each model. Each predictor variable was stan-
r2marginal, i.e. the variance explained by fixed effects in a mixed effects
dardized to a mean = 0 and SD = 1 prior to analysis. We calculated the
model (calculated according to Nakagawa and Schielzeth, 2013).
small-samples Akaike Information Criterion (AICc) for each model and
We also tested for spatial autocorrelation of model residuals. We
retained the model with the smallest AICc for further analysis. Thus the
used a one-tailed Global Moran’s I to test for positive spatial auto-
scale of effect for each landscape context variable was the scale in-
correlation in model residuals, i.e. whether similarity in residual values
cluded in the most supported model.
declined with distance between sampling sites. We used a permutation
We evaluated the relative effects of the six farming practices and
approach (with 1000 permutations) to calculate the significance-level
two farmland heterogeneity variables (measured at their scales of ef-
for each test and residuals were considered spatially autocorrelated at
fect) on multidiversity, using the following methods to minimize the
p < 0.05.
potential impact of collinearity on our effect size estimates. First, we
Analyses were conducted in R (R Core Team, 2017), using the lme4
did this by including all predictors within the statistical model, because
(Bates et al., 2015), glmmADMB (Fournier et al., 2012; Skaug et al.,
effect size estimates can be biased when an influential predictor is
2015), MuMIn (Barton, 2016), and ape (Paradis et al., 2004) packages.
omitted from the model (Smith et al., 2009). Second, we used the
standardized model coefficients from a regression-type model to esti-
3. Results
mate the relative effects of the variables. Although even low–moderate
collinearity among predictors within a regression-type model increases
The total number of species identified per taxonomic group in the
uncertainty in the coefficient estimates for the correlated predictors
subset of 112 sampling sites used in our analyses ranged from 25 but-
(i.e. standard error of the estimate; Graham, 2003), it does not gen-
terfly species to 165 herbaceous plant species (Table 1, Appendix F).
erally bias the size or direction of the model coefficient (Smith et al.,
Herbaceous plants had, on average, the most species per sampling site
2009). An effect was considered strongly supported if it was significant
(mean = 20.97, range = 2–51 species) and syrphid flies the least
at α = 0.05 (i.e. the 95% confidence interval [CI] did not cross zero).
(mean = 2.17, range = 0–10). Multidiversity ranged from −1.08 to
To determine which taxonomic groups drove effects of farming
1.40 (mean = −0.05). See Appendix C for a summary of farming
practices and farmland heterogeneity on multidiversity, we estimated
practice and landscape context variable values across our sampling
the relative effects of farming practices and farmland heterogeneity on
sites.
species richness for each individual taxonomic group. Each landscape
The strength of effect of the landscape context variables varied with
context variable was measured at its scale of effect on multidiversity
the spatial extent (Appendix E). We identified the scale of effect for
6
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
Table 2
Pairwise relationships among our 10 predictor variables. We used generalized linear mixed effects with a binomial distribution and logit link for pairwise com-
parisons of the binary farming practices variables and linear mixed effects for pairwise comparisons of continuous landscape context variables (at their scales of
effect) and for pairwise comparisons of farming practice to landscape variables. Sampling area was included as a random effect to account for non-independence and
spatial autocorrelation of measurements within a sampling area. We indexed the variance shared by two predictors as the r2marginal, i.e. the variance explained by the
fixed effect in a mixed effects model.
fertilizer herbicide insecticide tile drainage tillage field size crop diversity crop cover sampling location
annual/perennial crop < 0.01 0.01 0.07 0.04 0.00 < 0.01 0.12 0.01 < 0.01
fertilizer < 0.01 0.12 –a 0.06 0.01 0.06 0.01 < 0.01
herbicide 0.07 0.01 0.16 < 0.01 0.01 0.02 < 0.01
insecticide 0.03 0.17 0.05 0.02 0.14 < 0.01
tile drainage 0.01 0.06 0.03 0.07 < 0.01
tillage 0.01 < 0.01 0.04 < 0.01
field size 0.01 0.24 < 0.01
crop diversity 0.03 < 0.01
crop cover < 0.01
a
r2marginal for the relationship between tile drainage and fertilizer use was omitted, because this model failed to converge.
each landscape context variable as the scale included in the most sup- variable on multidiversity—also had negative effects on species rich-
ported model (i.e. the model with ΔAICc = 0), with effects of field size ness for 6 / 8 taxonomic groups (Fig. 4a). The strong negative effect of
and crop cover on multidiversity within 500 m of the sampling site, and annual crops (relative to perennial ones) on multidiversity was largely
the effect of crop diversity within 250 m. driven by its negative effects on herbaceous plant, carabid beetle, and
Although correlations between field size, crop diversity, and crop spider richness: annual/perennial crop had the largest effect of any
cover across our sampling areas were moderately strong (see 2.2. Se- farming practice/heterogeneity variable on richness of each of these
lection of farmland-dominated sampling areas and sampling sites taxa (Appendix H). It also had negative effects on syrphid fly, butterfly,
within sampling areas), the relationships between predictor variables and bird species richness. Our models predict that we can expect
across the sampling sites were generally weak (Table 2). The strongest richness for these taxa to increase by 2–35% (relative to the maximum
relationship occurred between field size and crop cover observed richness; Table 1) when moving from an annual to perennial
(r2marginal = 0.24), with more crop cover in landscapes with larger field field (Fig. 4a).
sizes. Field size and crop diversity also had consistently negative effects
Inclusion of the sampling location (field edge or interior) and on the species richness of individual taxonomic groups. Richness was
sampling area was sufficient to control for spatial autocorrelation in our lower in landscapes with larger fields than in landscapes with smaller
model residuals. All Global Moran’s I tests for spatial autocorrelation of fields, and lower in landscapes with more diverse crops than in land-
model residuals were non-significant (p ≥ 0.08; Appendix G). scapes with less diverse crops, for all taxa except for spiders (Fig. 5).
Farming practices and farmland heterogeneity had similarly strong The negative effects of field size and crop diversity on species richness
effects on farmland multidiversity (Fig. 3). For example, the effect of were particularly important for syrphid flies: field size and crop di-
the most influential farming practice variable—annual/perennial versity had the largest and second-largest effects on fly richness, re-
crop—on multidiversity was only 1.4 times the effect of field size and spectively (Appendix H). For taxa negatively affected by field size, our
1.4 times the effect of crop diversity. Two farming practices varia- model predicts that we can expect richness to increase by 3–46% (re-
bles—annual/perennial crop and tillage—and both farmland hetero- lative to the maximum observed richness) when mean field sizes within
geneity variables—field size and crop diversity—had significant effects a 500 m radius are reduced by ∼12 ha (from 13 to 1 ha; Fig. 5a). For
on multidiversity (i.e. the 95% CI of each effect did not cross zero; taxa negatively affected by crop diversity, our model predicts that we
Fig. 3). Thus this analysis indicates that farmland multidiversity is can expect richness to increase by 1–51% when the Shannon diversity
lower in farmlands with tilled, annual crops than untilled, perennial of crop types within a 250 m radius is reduced from 1.58 to 0.00
fields, and in landscapes with larger fields and more diverse crops. (Fig. 5b).
The directions of effect of these strongly-supported farming practice Although tillage had a significant negative effect on multidiversity,
and farmland heterogeneity variables on multidiversity were generally its effect on the species richness of individual taxa was only negative for
consistent across individual taxonomic groups. For example, annual/ half of the taxonomic groups (Fig. 4b). However, when an effect of
perennial crop—which had the strongest (negative) effect of any tillage on the richness of an individual taxonomic group was in the
7
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
Fig. 4. (a) Predicted change in species richness (relative to the maximum observed richness) when moving from an annual to perennial crop field, calculated as the
(predicted richness in a perennial field – predicted richness in an annual field) / maximum observed richness at any sampling site (see Table 1) × 100%. (b) Predicted
change in species richness (relative to the maximum observed richness) when moving from a tilled to untilled field, calculated as the (predicted richness in an untilled
field – predicted richness in a tilled field) / maximum observed richness at any sampling site × 100%. For each taxon, we modeled species richness as a function of six
farming practices (annual/perennial crop, fertilizer use, herbicide use, insecticide use, tile drainage, tillage) and two measures of farmland heterogeneity (field size,
crop diversity) using generalized linear mixed effects with a negative binomial distribution and log link. We included a fixed effect of crop cover and a random effect
of sampling area in all models. Sampling location (field edge or field interior) was included in all models except for models of woody plant and bird richness, because
these taxa were sampled in field edges only. Silhouettes were taken from http://phylopic.org/ (licensed for use under Public Domain licenses [https://
creativecommons.org/publicdomain/mark/1.0/], except for the bee and beetle silhouettes which were made available for use by M. Broussard under CC BY 3.0
licenses [https://creativecommons.org/licenses/by/3.0/]).
opposite direction (positive) it was weak. Tillage had the largest (ne- Multidiversity was higher in field edges than field interiors, as was the
gative) effect of any farming practice/heterogeneity variable on but- richness of all individual taxa (except woody plants and birds, which
terflies and birds, and the direction of effect of tillage was also negative were sampled in field edges only). Multidiversity was higher in land-
for woody plants and spiders (Appendix H). Our model predicts that we scapes with low crop cover than in landscapes with high crop cover.
can expect richness of these taxa to increase by 12–21% (relative to the The richness of individual taxa was also typically higher in landscapes
maximum observed richness) when moving from a tilled to untilled with less crop cover than in landscapes with more crop cover, except for
field (Fig. 4b). In contrast, tillage had slightly positive effects on the butterflies and syrphid flies.
richness of syrphid flies, bees, herbaceous plants, and beetles. For these
taxa, we can expect richness to decline by 1–3% when moving from a
tilled to untilled field (Fig. 4b). 4. Discussion
Farming practices and farmland heterogeneity generally had si-
milar effects on multidiversity and species richness in field interiors We found that farmland heterogeneity can have similar—and
and edges. There were significantly different effects of annual/per- sometimes even larger—effects on farmland biodiversity than the farm
ennial crop (X2annual/perennial crop × sampling location = 9.83, p = 0.002), management practices used in individual crop fields. Although annual/
crop diversity (X2crop diversity × sampling location = 5.99, p = 0.01), and perennial crop had the largest effect on biodiversity in our study, its
tillage (X2tillage × sampling location = 5.37, p = 0.02) on multidiversity in effect was only 1.4 times the effect of field size and 1.4 times the effect
field edges and interiors. However, the direction of effect for each of of crop diversity. Additionally, we estimated that the effects of field size
these variables was the same—i.e. negative—in field edges and in- on multidiversity were larger than effects of using fertilizers (2.5×),
teriors (Appendix I). Similarly, the directions of effect of these vari- herbicides (3.8×), insecticides (11.4×), and tile drainage (34.1×).
ables on the species richness of individual taxa were the same in field The effects of crop diversity on multidiversity were 2.5×, 3.9×,
edges and interiors, with one exception: herbaceous plant richness 11.9×, and 35.3× larger than effects of using fertilizers, herbicides,
decreased as crop diversity increased in field edges, but increased insecticides, and tile drainage, respectively.
(slightly) as crop diversity increased in field interiors (X2crop diversity × This was surprising, given that previous comparisons of the effects
of individual farming practices versus farmland heterogeneity on the
sampling location = 9.78, p = 0.002; Appendix I). Tile drainage had a
negative effect on multidiversity in field interiors and a (slightly) richness of individual taxa have generally found the opposite (Billeter
positive effect on multidiversity in field edges (X2tile drainage × sampling et al., 2008; Chiron et al., 2014; Geiger et al., 2010). However, these
studies could have underestimated the effects of farmland hetero-
location = 6.16, p = 0.01). However, the effect of tile drainage on
multidiversity was non-significant and smaller than the effects of an- geneity relative to farming practices because they did not measure
nual/perennial crop, crop diversity, field size, and tillage whether we farmland heterogeneity at the spatial extent of its effects on the studied
distinguished between field edges and interiors or not (Appendix I). taxa. Previous studies have shown that the spatial extent at which
Finally, although the effects of sampling location and crop cover on landscape context variables are measured can strongly affect conclu-
multidiversity were not the focus of this study, we note that both had sions about the direction, relative importance, and estimated biolo-
significant effects on farmland multidiversity (Appendix J). gical/statistical significance of an effect (e.g. Holland et al., 2004;
Smith et al., 2011). To our knowledge, none of the previous studies of
8
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
9
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
2018; Novotný et al., 2015; Palmu et al., 2014; Redlich et al., 2018a; diversity for landscape complementation (Dunning et al., 1992) at the
Reynolds et al., 2018), negative (Hass et al., 2018), and mixed effects scale of movement among complementary resources (such as breeding
(Hiron et al., 2015; Kirk and Freemark Lindsay, 2017; Santana et al., and foraging habitats) are greater than its negative effect at the smaller
2017; Sirami et al., 2019; Wilson et al., 2017) of crop diversity on spatial extent.
species richness/abundance. We note that these inconsistent findings
may occur, at least in some cases, because the effect of crop diversity on 4.1. Management implications
biodiversity depends on the farmland amount: both Wilson et al. (2017)
and Sirami et al. (2019) found evidence that crop diversity can have The most compelling implication of this study is that it suggests that
positive effects on biodiversity in landscapes with low crop cover, but policies/guidelines aimed at reducing crop field sizes would be at least
negative effects on biodiversity in landscapes with high crop cover. as effective for conservation of biodiversity within working agricultural
However, we found no support for an interacting effect of crop di- landscapes as policies/guidelines designed to promote a wildlife-
versity × crop cover on multidiversity in our study (Appendix M). friendly farming practice, such as adoption of no-till or low-input (e.g.
Another possible explanation for our observed negative effect of fertilizer, pesticide) farming practices. However, we caution against
crop diversity on farmland biodiversity is that this relationship is in fact direct extrapolation of this conclusion to other agricultural regions.
driven by another, unmeasured landscape context variable that is cor- Such extrapolation would require that the range and distribution of
related with crop diversity. We considered the possibility that the ap- predictor values in other regions is reasonably similar to those in our
parent negative effect of crop diversity on biodiversity was driven by data set and region. Additionally, Dormann et al. (2013) demonstrated
the proportion of farmland in annual crops. In our study region there that collinearity among predictor variables can be problematic when
are more different annual crop types than perennial ones, and thus crop extrapolating results to other sites with different relationships among
diversity may be higher in landscapes dominated by annual crops than those same variables (e.g. different degree of correlation between field
in landscapes dominated by perennial crops. And, one would expect size and crop cover). However, we note that Dormann et al. (2013) also
lower biodiversity in landscapes with more annual crops. However, the found that regression-type models—like the ones used in this study—-
relationship between crop diversity and the proportion of farmland in tended to produce reasonably accurate predictions so long as the cor-
annual crops was weak (r2marginal = 0.06), and inclusion of the propor- relations among predictors were not extreme (i.e. ǀrǀ should be below
tion of farmland in annual crops (at its scale of effect) in our statistical ∼0.70). Nevertheless, we recommend further research to determine
model did not change the direction of—or support for—a negative ef- whether our results apply in other agricultural regions, as well as to
fect of crop diversity (Appendix M). We also considered the possibility taxa not studied here. We also caution against extrapolation of the ef-
that exclusion of semi-natural cover from our analyses could have led to fects of farming practices versus farmland heterogeneity on biodiversity
an inaccurate estimate of the effect size of crop diversity. However, we within the cropped portion of the landscape to wildlife found in other
found a very weak relationship between crop diversity and semi-natural (e.g. semi-natural) land cover types. In this study we only sampled for
cover (r2marginal < 0.01), and our conclusions did not change when we biodiversity in crop fields and along crop field edges, and species found
included semi-natural cover in our statistical model (Appendix M). in other land cover types may respond differently to farming practices
Therefore the negative effect of crop diversity on farmland biodiversity and the farmland landscape context.
was not driven by its relationship with either the proportion of farm- We also stress that our findings confirm the conservation benefits of
land in annual crops or semi-natural cover. reducing, where possible, the amount of crop cover. In our study we
Our observed negative effect of crop diversity on biodiversity ap- found a significant negative effect of crop cover on farmland multi-
pears to contradict Fahrig et al. (2015), who found positive (though diversity even though we specifically selected sampling areas to reduce
weak) effects of crop diversity on the richness/abundance of most taxa. the range of variability in crop cover which, all else being equal, should
This contradiction was particularly surprising because our biodiversity reduce its estimated effect on farmland biodiversity. Thus we likely
data included some of the data used by Fahrig et al. (2015). However, underestimated the effect of crop cover on biodiversity.
there are a number of differences between our study and Fahrig et al. Nevertheless, recommendations to reduce crop cover in agricultural
(2015) that could potentially explain the opposite effects of crop di- landscapes may not be practical, given the pressure to increase food
versity on farmland biodiversity observed in the two studies (as detailed production to meet the demands of a growing human population
in Appendix N). The most likely explanation is that crop diversity was (United Nations, 2017). Indeed, projections based on current popula-
measured within different spatial extents in the two studies. We mea- tion and food consumption trends suggest land clearing for agriculture
sured crop diversity within a radius of 250 m around the sampling sites will continue (e.g. Alexandratos and Bruinsma, 2012; Wirsenius et al.,
—selected empirically as this variable’s scale of effect—whereas Fahrig 2010). Thus there is value in identifying conservation actions that can
et al. (2015) measured crop diversity within an arbitrary 1 x 1 km maintain or increase biodiversity in working agricultural landscapes
landscape (approximately the same extent as our measurements within without taking land out of production.
a 500-m radius). The effect of crop diversity on multidiversity in our In addition to allowing for both biodiversity conservation and food
study varied with the spatial extent, with strong negative effects when production, biodiverse agricultural landscapes benefit people through
we measured crop diversity within a radius of 250 m and weak (non- provisioning of ecosystem services. For example, meta-analysis has
significant) positive effects when we measured crop diversity at larger shown that increases in plant diversity result in increases in provi-
spatial extents (Appendix N). Thus our findings do not in fact contradict sioning of plant products, erosion control, invasion resistance, soil
those of Fahrig et al. (2015). fertility, and regulation of pests and pathogens (Quijas et al., 2010).
Thus a key question to answer is: why does the negative effect of Pollination services may also be positively related to pollinator di-
crop diversity on farmland biodiversity at a small spatial extent dis- versity (Klein et al., 2007).
sipate—and even become slightly positive—at larger spatial extents? For decision-makers, key next steps are to understand the potential
We speculate that, if land cover within a 250 m extent indexes core socioeconomic impacts of reducing field size on yield and the likelihood
habitat availability for species at the sampling site, then there may be that farmers would adopt policies/guidelines aimed at keeping field
fewer species when there is high versus low crop diversity because there sizes small. Evidence suggests a preference for larger crop fields, with
is less area per crop type when crop diversity is high. Thus the area of a observations of increasing field sizes over time in some (but not all)
given crop type may be too small to support species that use it as ha- crop types and regions (Daystar et al., 2017; Robinson and Sutherland,
bitat in landscapes with high crop diversity within 250 m. However, at 2002; White and Roy, 2015). There are a number of factors that may
larger extents the observed effect of crop diversity on multidiversity/ motivate preference for larger fields, including increasing the amount
species richness may become weakly positive if the benefits of crop of land in crops within a farm holding (by replacing field border
10
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
vegetation with crops), increasing efficiency of farm machinery, and the variation in land-use intensity enhances grassland multidiversity. Proc. Natl. Acad.
perception that plant productivity is lower in small than large fields, Sci. U. S. A. 111, 308–313. https://doi.org/10.1073/pnas.1312213111.
Barton, K., 2016. MuMIn: Multi-model Inference. R Package Version 1.15.6.
because small fields have higher edge:interior ratios, and productivity is Bates, D., Mäechler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models
lower at field edges. Both theoretical and empirical studies suggest that, using lme4. J. Stat. Softw. 67, 1–48. https://doi.org/10.18637/jss.v067.i01.
although there are increases in machine efficiency with increasing field Billeter, R., Liira, J., Bailey, D., Bugter, R., Arens, P., Augenstein, I., Aviron, S., Baudry, J.,
Bukacek, R., Burel, F., Cerny, M., De Blust, G., De Cock, R., Diekötter, T., Dietz, H.,
size, the marginal returns are declining (Deverell et al., 2009; Dirksen, J., Dormann, C., Durka, W., Frenzel, M., Hamersky, R., Hendrickx, F.,
Rodríguez and Wiegand, 2009; Xangsayasane et al., 2019). If there are Herzog, F., Klotz, S., Koolstra, B., Lausch, A., Le Coeur, D., Maelfait, J.P., Opdam, P.,
no real gains in machine efficiency above a threshold field size, then Roubalova, M., Schermann, A., Schermann, N., Schmidt, T., Schweiger, O., Smulders,
M.J.M., Speelmans, M., Simova, P., Verboom, J., Van Wingerden, W.K.R.E., Zobel,
reducing the size of fields to this threshold will have minimal impacts M., Edwards, P.J., 2008. Indicators for biodiversity in agricultural landscapes: a pan-
on efficiency. It is not yet clear whether there is a similar pattern for the European study. J. Appl. Ecol. 45, 141–150. https://doi.org/10.1111/j.1365-2664.
relationship between field size and yield. Interestingly, there is a body 2007.01393.x.
Boscolo, D., Metzger, J.P., 2009. Is bird incidence in Atlantic forest fragments influenced
of literature suggesting that productivity is actually greater in smaller
by landscape patterns at multiple scales? Landsc. Ecol. 24, 907–918. https://doi.org/
fields than larger ones, although the validity of these findings is un- 10.1007/s10980-009-9370-8.
certain (Gourlay et al., 2017). Thus future work is needed to understand Chiron, F., Chargé, R., Julliard, R., Jiguet, F., Muratet, A., 2014. Pesticide doses, land-
the effects of field size on yield. scape structure and their relative effects on farmland birds. Agric. Ecosyst. Environ.
185, 153–160. https://doi.org/10.1016/j.agee.2013.12.013.
Additionally, understanding the relative socioeconomic impacts of Collins, S.J., Fahrig, L., 2017. Responses of anurans to composition and configuration of
reducing the intensity of farming practices versus reducing field sizes agricultural landscapes. Agric. Ecosyst. Environ. 239, 399–409. https://doi.org/10.
can aid decision-making, allowing policymakers and farmers to identify 1016/j.agee.2016.12.038.
Curran, C.H., 1951. Synopsis of the North American species of Spilomyia (Syrphidae,
the conservation action(s) with the lowest socioeconomic cost. If re- Diptera). Am. Museum Novit. 1492, 1–11.
ducing field sizes has smaller effects on yield than adoption of wildlife- Curran, C.H., Fluke, C.L., 1926. Revision of the nearctic species of Helophilus and allied
friendly practices, then reducing crop field sizes could be a promising genera. Trans. Wisconsin Acad. Sci. Arts Lett. 22, 207–281.
Daystar, J.S., Barnes, E., Hake, K., Kurtz, R., 2017. Sustainability trends and natural re-
strategy for conservation, allowing for conservation of biodiversity in source use in U.S. cotton production. BioResources 12, 362–392. https://doi.org/10.
working agricultural landscapes with minimal impact on agricultural 15376/biores.12.1.362-392.
production. Demjanová, E., 2004. Effects of crop rotation and tillage systems on weed populations,
density and diversity in maize (Zea mays L.). Acta Fytotech. Zootech. 7, 61–63.
Deverell, R., McDonnell, K., Devlin, G., 2009. The impact of field size on the environment
Funding and energy crop production efficiency for a sustainable indigenous bioenergy supply
chain in the Republic of Ireland. Sustainability 1, 994–1011. https://doi.org/10.
3390/su1040994.
This research was funded by a Natural Sciences and Engineering
Diefenbach, D.R., Brauning, D.W., Mattice, J.A., 2003. Variability in grassland bird counts
Research Council of Canada (NSERC) Strategic Project grant to LF. related to observer differences and species detection rates. Auk 120, 1168–1179.
https://doi.org/10.1642/0004-8038(2003)120[1168:VIGBCR]2.0.CO;2.
Declaration of Competing Interest Domagalski, J.L., Ator, S., Coupe, R., McCarthy, K., Lampe, D., Sandstrom, M., Baker, N.,
2008. Comparative study of transport processes of nitrogen, phosphorous, and her-
bicides to streams in five agricultural basins, USA. J. Environ. Qual. 37, 1158–1169.
The authors declare that they have no known competing financial https://doi.org/10.2134/jeq2007.0408.
interests or personal relationships that could have appeared to influ- Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G.,
Gruber, B., Lafourcade, B., Leitão, P.J., Münkemüller, T., McClean, C., Osborne, P.E.,
ence the work reported in this paper. Reineking, B., Schröder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2013.
Collinearity: a review of methods to deal with it and a simulation study evaluating
Acknowledgements their performance. Ecography 36, 27–46. https://doi.org/10.1111/j.1600-0587.
2012.07348.x.
Dunning, J.B., Danielson, B.J., Pulliam, H.R., 1992. Ecological processes that affect po-
We thank Pasher et al. (2013) for their work to select the farmland- pulations in complex landscapes. Oikos 65, 169–175.
dominated sampling areas used in this study. We also thank the land- Duro, D.C., Girard, J., King, D.J., Fahrig, L., Mitchell, S., Lindsay, K., Tischendorf, L.,
2014. Predicting species diversity in agricultural environments using Landsat TM
owners and farmers, who permitted access to their property for biodi-
imagery. Remote Sens. Environ. 144, 214–225. https://doi.org/10.1016/j.rse.2014.
versity surveys and provided us with information on their farming 01.001.
practices, and the team of graduate students and research assistants, Egan, J.F., Mortensen, D.A., 2012. A comparison of land‐sharing and land‐sparing stra-
tegies for plant richness conservation in agricultural landscapes. Ecol. Appl. 22,
who assisted with data collection and species identification. We would
459–471. https://doi.org/10.1890/11-0206.1.
like to acknowledge the contributions of specialists at Agriculture and Ekroos, J., Tiainen, J., Seimola, T., Herzon, I., 2019. Weak effects of farming practices
Agri-Food Canada, the American Museum of Natural History, and the corresponding to agricultural greening measures on farmland bird diversity in boreal
Royal Saskatchewan Museum who identified our bee specimens. Fi- landscapes. Landsc. Ecol. 34, 389–402. https://doi.org/10.1007/s10980-019-
00779-x.
nally, we thank the anonymous reviewers for their helpful comments on Fahrig, L., Baudry, J., Brotons, L., Burel, F.G., Crist, T.O., Fuller, R.J., Sirami, C.,
an earlier version of this manuscript. Siriwardena, G.M., Martin, J.-L., 2011. Functional landscape heterogeneity and an-
imal biodiversity in agricultural landscapes. Ecol. Lett. 14, 101–112. https://doi.org/
10.1111/j.1461-0248.2010.01559.x.
Appendix A. Supplementary data Fahrig, L., Girard, J., Duro, D., Pasher, J., Smith, A., Javorek, S., King, D., Lindsay, K.F.,
Mitchell, S., Tischendorf, L., 2015. Farmlands with smaller crop fields have higher
Supplementary material related to this article can be found, in the within-field biodiversity. Agric. Ecosyst. Environ. 200, 219–234. https://doi.org/10.
1016/j.agee.2014.11.018.
online version, at doi:https://doi.org/10.1016/j.agee.2019.106698. Fournier, D.A., Skaug, H.J., Ancheta, J., Ianelli, J., Magnusson, A., Maunder, M.N.,
Nielsen, A., Sibert, J., 2012. AD Model Builder: using automatic differentiation for
References statistical inference of highly parameterized complex nonlinear models. Optim.
Method. Softw. 27, 233–249. https://doi.org/10.1080/10556788.2011.597854.
Freemark, K.E., Kirk, D.A., 2001. Birds on organic and conventional farms in Ontario:
Alexandratos, N., Bruinsma, J., 2012. World Agriculture towards 2030/2050: the 2012 partitioning effects of habitat and practices on species composition and abundance.
Revision, ESA Working Paper No. 12-03. FAO, Rome. Biol. Conserv. 101, 337–350. https://doi.org/10.1016/S0006-3207(01)00079-9.
Allan, E., Bossdorf, O., Dormann, C.F., Prati, D., Gossner, M.M., Tscharntke, T., Blüthgen, Gagic, V., Kleijn, D., Báldi, A., Boros, G., Jørgensen, H.B., Elek, Z., Garratt, M.P.D., de
N., Bellach, M., Birkhofer, K., Boch, S., Böhm, S., Börschig, C., Chatzinotas, A., Christ, Groot, G.A., Hedlund, K., Kovács-Hostyánszki, A., Marini, L., Martin, E., Pevere, I.,
S., Daniel, R., Diekötter, T., Fischer, C., Friedl, T., Glaser, K., Hallmann, C., Hodac, L., Potts, S.G., Redlich, S., Senapathi, D., Steffan-Dewenter, I., Świtek, S., Smith, H.G.,
Holzel, N., Jung, K., Klein, A.M., Klaus, V.H., Kleinebecker, T., Krauss, J., Lange, M., Takács, V., Tryjanowski, P., van der Putten, W.H., van Gils, S., Bommarco, R., 2017.
Morris, E.K., Müller, J., Nacke, H., Pašalić, E., Rillig, M.C., Rothenwöhrer, C., Schall, Combined effects of agrochemicals and ecosystem services on crop yield across
P., Scherber, C., Schulze, W., Socher, S.A., Steckel, J., Steffan-Dewenter, I., Türke, M., Europe. Ecol. Lett. 20, 1427–1436. https://doi.org/10.1111/ele.12850.
Weiner, C.N., Werner, M., Westphal, C., Wolters, V., Wubet, T., Gockel, S., Gorke, M., Geiger, F., Bengtsson, J., Berendse, F., Weisser, W.W., Emmerson, M., Morales, M.B.,
Hemp, A., Renner, S.C., Schöning, I., Pfeiffer, S., König-Ries, B., Buscot, F., Ceryngier, P., Liira, J., Tscharntke, T., Wingvist, C., Eggers, S., Bommarco, R., Pärt,
Linsenmair, K.E., Schulze, E.-D., Weisser, W.W., Fischer, M., 2014. Interannual T., Bretagnolle, V., Plantegenest, M., Clement, L.W., Dennis, C., Palmer, C., Oñate, C.,
11
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
Guerrero, J.J., Hawro, I., Aavik, V., Thies, T., Flohre, C., Hänke, A., Fischer, S., 3.4.0. R Foundation for Statistical Computing, Vienna.
Goedhart, C., Goedhart, P.W., Inchausti, P., 2010. Persistent negative effects of Redlich, S., Martin, E.A., Steffan-Dewenter, I., 2018a. Landscape-level crop diversity
pesticides on biodiversity and biological control potential on European farmland. benefits biological pest control. J. Appl. Ecol. 55, 2419–2428. https://doi.org/10.
Basic Appl. Ecol. 11, 97–105. 1111/1365-2664.13126.
Geroff, R.K., Gibbs, J., McCravy, K.W., 2014. Assessing bee (Hymenoptera: Apoidea) Redlich, S., Martin, E.A., Wende, B., Steffan-Dewenter, I., 2018b. Landscape hetero-
diversity of an Illinois restored tallgrass prairie: methodology and conservation geneity rather than crop diversity mediates bird diversity in agricultural landscapes.
considerations. J. Insect Conserv. 18, 951–964. https://doi.org/10.1007/s10841- PLoS One 13, e0200438. https://doi.org/10.1371/journal.pone.0200438.
014-9703-z. Reynolds, C., Fletcher Jr., R.J., Carneiro, C.M., Jennings, N., Ke, A., LaScaleia, M.C.,
Glover, J.D., Culman, S.W., DuPont, S.T., Broussard, W., Young, L., Mangan, M.E., Mai, Lukhele, M.B., Mamba, M.L., Sibiya, M.D., Austin, J.D., Magagula, C.N., Mahlaba, T.,
J.G., Crews, T.E., DeHaan, L.R., Buckley, D.H., Ferris, H., Turner, R.E., Reynolds, Monadjem, A., Wisely, S.M., McCleery, R.A., 2018. Inconsistent effects of landscape
H.L., Wyse, D.L., 2010. Harvested perennial grasslands provide ecological bench- heterogeneity and land-use on animal diversity in an agricultural mosaic: a multi-
marks for agricultural sustainability. Agric. Ecosyst. Environ. 137, 3–12. https://doi. scale and multi-taxon investigation. Landsc. Ecol. 33, 241–255. https://doi.org/10.
org/10.1016/j.agee.2009.11.001. 1007/s10980-017-0595-7.
Gourlay, S., Kilic, T., Lobell, D., 2017. Could the Debate Be Over? Errors in Farmer- Robertson, B.A., Doran, P.J., Loomis, L.R., Robertson, J.R., Schemske, D.W., 2011.
reported Production and Their Implications for the Inverse Scale-productivity Perennial biomass feedstocks enhance avian diversity. GCB Bioenergy 3, 235–246.
Relationship in Uganda. Policy Research Working Paper No. 8192. World Bank https://doi.org/10.1111/j.1757-1707.2010.01080.x.
Development Economics Development Data Group, Washington. Robinson, R.A., Sutherland, W.J., 2002. Post-war changes in arable farming and biodi-
Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regression. versity in Great Britain. J. Appl. Ecol. 39, 157–176. https://doi.org/10.1046/j.1365-
Ecology 84, 2809–2815. https://doi.org/10.1890/02-3114. 2664.2002.00695.x.
Hass, A.L., Kormann, U.G., Tscharntke, T., Clough, Y., Baillod, A.B., Sirami, C., Fahrig, L., Rodríguez, C., Wiegand, K., 2009. Evaluating the trade-off between machinery efficiency
Martin, J.-L., Baudry, J., Bertrand, C., Bosch, J., Brotons, L., Burel, F., Georges, R., and loss of biodiversity-friendly habitats in arable landscapes: the role of field size.
Giralt, D., Marcos-García, M.Á., Ricarte, A., Siriwardena, G., Batáry, P., 2018. Agric. Ecosyst. Environ. 129, 361–366. https://doi.org/10.1016/j.agee.2008.10.010.
Landscape configurational heterogeneity by small-scale agriculture, not crop di- Roschewitz, I., Gabriel, D., Tscharntke, T., Thies, C., 2005. The effects of landscape
versity, maintains pollinators and plant reproduction in western Europe. Proc. R. Soc. complexity on arable weed species diversity in organic and conventional farming. J.
B Biol. Sci. 285, 20172242. https://doi.org/10.1098/rspb.2017.2242. Appl. Ecol. 42, 873–882. https://doi.org/10.1111/j.1365-2664.2005.01072.x.
Hiron, M., Berg, Å., Eggers, S., Berggren, Å., Josefsson, J., Pärt, T., 2015. The relationship Santana, J., Reino, L., Stoate, C., Moreira, F., Ribeiro, P.F., Santos, J.L., Rotenberry, J.T.,
of bird diversity to crop and non-crop heterogeneity in agricultural landscapes. Beja, P., 2017. Combined effects of landscape composition and heterogeneity on
Landsc. Ecol. 30, 2001–2013. https://doi.org/10.1007/s10980-015-0226-0. farmland avian diversity. Ecol. Evol. 7, 1212–1223. https://doi.org/10.1002/ece3.
Holland, J.D., Bert, D.G., Fahrig, L., 2004. Determining the spatial scale of species’ re- 2693.
sponse to habitat. Bioscience 54, 227–233. https://doi.org/10.1641/0006- Sapkota, T.B., Mazzoncini, M., Bàrberi, P., Antichi, D., Silvestri, N., 2012. Fifteen years of
3568(2004)054[0227:DTSSOS]2.0.CO;2. no till increase soil organic matter, microbial biomass and arthropod diversity in
Kirk, D.A., Freemark Lindsay, K.E., 2017. Subtle differences in birds detected between cover crop-based arable cropping systems. Agron. Sustain. Dev. 32, 853–863. https://
organic and nonorganic farms in Saskatchewan Prairie Parklands by farm pair and doi.org/10.1007/s13593-011-0079-0.
bird functional group. Agric. Ecosyst. Environ. 246, 184–201. https://doi.org/10. Scharlemann, J.P.W., Balmford, A., Green, R.E., 2005. The level of threat to restricted-
1016/j.agee.2017.04.009. range bird species can be predicted from mapped data on land use and human po-
Kirk, D.A., Lindsay, K.E., Brook, R.W., 2011. Risk of agricultural practices and habitat pulation. Biol. Conserv. 123, 317–326. https://doi.org/10.1016/j.biocon.2004.11.
change to farmland birds. Avian Conserv. Ecol. 6, 5. https://doi.org/10.5751/ACE- 019.
00446-060105. Shannon, R.C., 1923. Review of the American Xylotine syrphid-flies. Proc. U.S. Natl.
Klein, A.-M., Vaissière, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, Museum 69, 9.
C., Tscharntke, T., 2007. Importance of pollinators in changing landscapes for world Sirami, C., Gross, N., Baillod, A.B., Bertrand, C., Carrié, R., Hass, A., Henckel, L., Miguet,
crops. Proc. Biol. Sci. B. 274, 303–313. https://doi.org/10.1098/rspb.2006.3721. P., Vuillot, C., Alignier, A., Girard, J., Batáry, P., Clough, Y., Violle, C., Giralt, D.,
Lee, M.-B., Goodale, E., 2018. Crop heterogeneity and non-crop vegetation can enhance Bota, G., Badenhauser, I., Lefebvre, G., Gauffre, B., Vialatte, A., Calatayud, F., Gil-
avian diversity in a tropical agricultural landscape in southern China. Agric. Ecosyst. Tena, A., Tischendorf, L., Mitchell, S., Lindsay, K., Georges, R., Hilaire, S., Recasens,
Environ. 265, 254–263. https://doi.org/10.1016/j.agee.2018.06.016. J., Solé-Senan, X.O., Robleño, I., Bosch, J., Barrientos, J.A., Ricarte, A., Marcos-
Lindroth, C., 1961. The ground-beetles (Carabidae excl. Cicindelinae) of Canada and Garcia, M.Á., Minaño, J., Mathevet, R., Gibon, A., Baudry, J., Balent, G., Poulin, B.,
Alaska. Part 2. Opusc. Entomol. Suppl. 20, 1–200. Burel, F., Tscharntke, T., Bretagnolle, V., Siriwardena, G., Ouin, A., Brotons, L.,
Lindroth, C., 1963. The ground-beetles (Carabidae excl. Cicindelinae) of Canada and Martin, J.-L., Fahrig, L., 2019. Increasing crop heterogeneity enhances multitrophic
Alaska. Part 3. Opusc. Entomol. Suppl. 24, 201–408. diversity across agricultural regions. PNAS 116, 16442–16447. https://doi.org/10.
Lindroth, C., 1966. The ground-beetles (Carabidae excl. Cicindelinae) of Canada and 1073/pnas.1906419116.
Alaska. Part 4. Opusc. Entomol. Suppl. 29, 409–648. Skaug, H., Fournier, D., Bolker, B., Magnusson, A., Nielsen, A., 2015. Generalized Linear
Lindroth, C., 1968. The ground-beetles (Carabidae excl. Cicindelinae) of Canada and Mixed Models Using “AD Model Builder.” R Package Version 0.8.3.3.
Alaska. Part 5. Opusc. Entomol. Suppl. 33, 649–944. Smith, A.C., Fahrig, L., Francis, C.M., 2011. Landscape size affects the relative importance
Lindroth, C., 1969. The ground-beetles (Carabidae excl. Cicindelinae) of Canada and of habitat amount, habitat fragmentation, and matrix quality on forest birds.
Alaska. Part 6. Opusc. Entomol. Suppl. 34, 945–1192. Ecography 34, 103–113. https://doi.org/10.1111/j.1600-0587.2010.06201.x.
Monck-Whipp, L., Martin, A.E., Francis, C.M., Fahrig, L., 2018. Farmland heterogeneity Smith, A.C., Koper, N., Francis, C.M., Fahrig, L., 2009. Confronting collinearity: com-
benefits bats in agricultural landscapes. Agric. Ecosyst. Environ. 253, 131–139. paring methods for disentangling the effects of habitat loss and fragmentation.
https://doi.org/10.1016/j.agee.2017.11.001. Landsc. Ecol. 24, 1271–1285. https://doi.org/10.1007/s10980-009-9383-3.
Murphy, S.D., Clements, D.R., Belaoussoff, S., Kevan, P.G., Swanton, C.J., 2006. Sosnoskie, L.M., Herms, C.P., Cardina, J., 2006. Weed seedbank community composition
Promotion of weed species diversity and reduction of weed seedbanks with con- in a 35-yr-old tillage and rotation experiment. Weed Sci. 54, 263–273.
servation tillage and crop rotation. Weed Sci. 54, 69–77. https://doi.org/10.1614/ Tamburini, G., De Simone, S., Sigura, M., Boscutti, F., Marini, L., 2016. Conservation
WS-04-125R1.1. tillage mitigates the negative effect of landscape simplification on biological control.
Nakagawa, S., Schielzeth, H., 2013. A general and simple method for obtaining R2 from J. Appl. Ecol. 53, 233–241. https://doi.org/10.1111/1365-2664.12544.
generalized linear mixed-effects models. Methods Ecol. Evol. 4, 133–142. Tang, X., Zhu, B., Katou, H., 2012. A review of rapid transport of pesticides from sloping
Norris, K., 2008. Agriculture and biodiversity conservation: opportunity knocks. Conserv. farmland to surface waters: processes and mitigation strategies. J. Environ. Sci. 24,
Lett. 1, 2–11. https://doi.org/10.1111/j.1755-263X.2008.00007.x. 351–361.
Novotný, D., Zapletal, M., Kepka, P., Beneš, J., Konvička, M., 2015. Large moths captures Telford, H.S., 1970. Eristalis (Diptera: Syrphidae) from America north of Mexico. Ann.
by a pest monitoring system depend on farmland heterogeneity. J. Appl. Entomol. Entomol. Soc. Am. 63, 1201–1210.
139, 390–400. https://doi.org/10.1111/jen.12185. Toler, T.R., Evans, E.W., Tepedino, V.J., 2005. Pan-trapping for bees (Hymenoptera :
OMAFRA, 2011. Eastern Ontario Region at a Glance. (Accessed 28 June 2017). http:// Apiformes) in Utah’s West Desert: the importance of color diversity. PanPac.
www.omafra.gov.on.ca/english/stats/county/eastern_ontario.htm. Entomol. 81, 103–113.
Ontario Ministry of Natural Resources, 2010. DRAPE (Digital Raster Aquisition Project for Tuck, S.L., Winqvist, C., Mota, F., Ahnström, J., Turnbull, L.A., Bengtsson, J., 2014. Land-
the East) Orthoimagery. [dataset]. . use intensity and the effects of organic farming on biodiversity: a hierarchical meta-
Palmu, E., Ekroos, J., Hanson, H.I., Smith, H.G., Hedlund, K., 2014. Landscape-scale crop analysis. J. Appl. Ecol. 51, 746–755. https://doi.org/10.1111/1365-2664.12219.
diversity interacts with local management to determine ground beetle diversity. Basic Ubick, D., Paquin, P., Cushing, P.E., Roth, V., 2005. Spiders of North America: an
Appl. Ecol. 15, 241–249. https://doi.org/10.1016/j.baae.2014.03.001. Identification Manual. American Arachnological Society, Keene.
Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics and evolution United Nations, 2017. World Population Prospects: the 2017 Revision, Key Findings and
in R language. Bioinformatics 20, 289–290. https://doi.org/10.1093/bioinformatics/ Advance Tables. Working Paper No. ESA/P/WP/248. UN Department of Economic
btg412. and Social Affairs, Population Division, New York.
Pasher, J., Mitchell, S.W., King, D.J., Fahrig, L., Smith, A.C., Lindsay, K.E., 2013. Vockeroth, J.R., 1992. The Flower Flies of the Subfamily Syrphinae of Canada, Alaska,
Optimizing landscape selection for estimating relative effects of landscape variables and Greenland: Diptera, Syrphidae. The Insects and Arachnids of Canada Part 18.
on ecological responses. Landsc. Ecol. 28, 371–383. https://doi.org/10.1007/ Canada Communications Group Publishing, Ottawa.
s10980-013-9852-6. Vockeroth, J.R., Thompson, F.C., 1987. Syrphidae. In: McAlpine, J.F. (Ed.), Manual of
Quijas, S., Schmid, B., Balvanera, P., 2010. Plant diversity enhances provision of eco- Nearactic Diptera. Canada Communications Group Publishing, Ottawa, pp. 713–743.
system services: a new synthesis. Basic Appl. Ecol. 11, 582–593. https://doi.org/10. White, E.V., Roy, D.P., 2015. A contemporary decennial examination of changing agri-
1016/j.baae.2010.06.009. cultural field sizes using Landsat time series data. Geo Geogr. Environ. 2, 33–54.
R Core Team, 2017. R: a Language and Environment for Statistical Computing. Version https://doi.org/10.1002/geo2.4.
12
A.E. Martin, et al. Agriculture, Ecosystems and Environment 288 (2020) 106698
Wilson, S., Mitchell, G.W., Pasher, J., McGovern, M., Hudson, M.-A.R., Fahrig, L., 2017. harvesting efficiency as affected by rice field size and other factors and its implication
Influence of crop type, heterogeneity and woody structure on avian biodiversity in for adoption of combine contracting service. Plant Prod. Sci. 22, 68–76. https://doi.
agricultural landscapes. Ecol. Indic. 83, 218–226. https://doi.org/10.1016/j.ecolind. org/10.1080/1343943X.2018.1561196.
2017.07.059. Zhou, W., Lee, M.-B., Goodale, E., 2018. The relationship between the diversity of her-
Wirsenius, S., Azar, C., Berndes, G., 2010. How much land is needed for global food baceous plants and the extent and heterogeneity of croplands in noncrop vegetation
production under scenarios of dietary changes and livestock productivity increases in in an agricultural landscape of south China. Glob. Ecol. Conserv. 14, e00399. https://
2030. Agric. Syst. 103, 621–638. https://doi.org/10.1016/j.agsy.2010.07.005. doi.org/10.1016/j.gecco.2018.e00399.
Xangsayasane, P., Phongchanmisai, S., Bounphanousai, C., Fukai, S., 2019. Combine
13