New methods of crop production and farmland birds: effects of plastic mulches on species richness and abundance

Authors


Summary

  1. Modern methods of crop production are regarded as one of the major factors moderating ecosystem processes in agricultural landscapes and may negatively affect farmland biodiversity. One method which is on the increase is mulching: the use of plastic foil to cover crops at sowing in order to promote early growth by reducing the negative effects of variable temperatures and to limit pesticide use. However, almost nothing is known of the role of mulching on farmland biodiversity.
  2. In this study, carried out in southern Poland in 2011, we investigated the impact of mulching with plastic foil on the number of species and abundance of farmland birds at two spatial scales.
  3. At the local scale, bird species number and abundance were lower in areas where foil was used than those in the control areas, both during the period when the foil was used and after it was removed from the crops. At the landscape scale, we found a significant negative relationship between the proportion of crops covered by foil and bird species richness and abundance. Farmland specialists, nonfarmland birds, ground nesters and above-ground nesting species were all negatively affected by foil mulching. Foil had a negative effect on potential resources for birds including adult butterflies and their larvae and weed species.
  4. Synthesis and applications. Our results provide the first evidence that the use of foil for mulching has negative effects on farmland bird populations, probably through the trophic cascade and habitat disturbance. Therefore, foil mulches must be considered as another factor contributing to the decline of farmland biodiversity. We suggest limiting the use of this method of vegetable production at the farm level. Decreasing the field size and converting some arable fields into grassland patches are proposed as mitigation measures in landscapes with high foil cover.

Introduction

Modern agriculture in Europe is based on large-scale food production. To achieve this, farmers introduce new methods of crop management in order to increase yield, limit the effects of pests and unfavourable weather conditions and, ultimately, to maximize their own economic benefits. These methods include genetically modified crops that are herbicide-tolerant and insect-resistant or UV-blocking foils that disrupt insect pest infestation (Diaz & Fereres 2007). Plastic mulches are a relatively new method used in agriculture and date back to the 1950s (Lamont 1991). The use of plastic mulch has become a common practice for all vegetable farmers; benefits include reduced evaporation and soil compaction, weed control, reduction in herbicide and pesticide application and elevated soil temperatures and thus frost protection, which promotes earlier plant maturity compared with conventional vegetable farming (Lamont 1991; Gustavsson 1999; Tarara 2000; Waterer 2000; Franczuk, Kosterna & Zaniewicz-Bajkowska 2010). Though very effective and affordable, plastic mulch has become an environmental management concern owing to disposal issues (Durham 2003). At the beginning of the 21st century, almost 13 million hectares world-wide were covered with plastic mulch (Timsina & Connor 2001). In Europe, foil mulches covered about 430 thousand hectares (Scarascia-Mugnozza, Sica & Russo 2011).

New methods of crop production in agriculture often have detrimental effects on wildlife (Schifferli 2001; Tryjanowski et al. 2011). Many of the organisms inhabiting farmland are amongst the currently most endangered species, with rapid population declines (Thomas et al. 2004; Wretenberg et al. 2007). For example, common farmland birds, which are often regarded as indicators of ecosystem health (Gregory et al. 2005), have declined dramatically in Europe compared with birds in other habitats (Donald, Green & Heath 2001). Almost nothing is known of the effect of plastic mulching on farmland birds.

Biodiversity conservation on farmland encompasses a range of different measures that take into account within-field processes and spatial heterogeneity at both local and larger spatial scales (Tscharntke et al. 2005; Batáry et al. 2011; Kleijn et al. 2011). At the local (field) scale, biodiversity is mostly shaped by management intensity and disturbance rates (Kruess & Tscharntke 2002; Kleijn & Sutherland 2003). At larger spatial scales, habitat heterogeneity and species dispersal abilities are important for sustaining biodiversity (Tscharntke et al. 2005; Kleijn et al. 2011). Thus, local (field scale) alpha species diversity was shown to be poorly, if at all, related to the landscape-wide beta species diversity (Tscharntke et al. 2012).

Foil mulches may be regarded as a high-level disturbance at the local scale as far as birds are concerned. Plastic foils are used mostly in early spring (April–May) when many farmland bird species establish breeding territories and start nesting. Fields covered with plastic foil prevent access to basic requirements such as nesting sites and food resources. However, foils are removed in the middle of May, so the effect of foil mulching is temporary and vacant sites could be quickly filled by birds from neighbouring areas (spillover effect; Tscharntke et al. 2012). However, several studies have shown that foil mulches permanently reduce species richness of weeds and density of several invertebrates, which may be food for farmland birds (Ricotta & Masiunas 1991; Radics & Szne Bognar 2004; Frank & Liburd 2005; Coolong 2012). Thus, according to food web theory (Cohen, Briand & Newman 1990; Williams & Martinez 2000), the impact of foil mulching on taxa from lower trophic levels should also be manifested in a negative impact on populations of taxa from higher trophic levels, for example, birds, even after removal of the foil from crops. Hence, one may hypothesize that at the local scale, foil mulches have a strong negative effect on farmland birds both during mulching and after its removal.

The conservation of biodiversity and ecosystem services in agricultural systems requires a landscape perspective (Tscharntke et al. 2005, 2012). Therefore, the effect of foil mulching on birds should also be considered at a larger, landscape scale. Farmland birds with high dispersal abilities may be less susceptible to small-scale changes in resource availability. Thus, at a landscape scale, the negative effect of foil mulches on farmland birds may be mitigated by spatial heterogeneity, for example, the presence of marginal habitats such as grassland patches providing nest sites and food resources, or by crop diversity. Therefore, when studying the effects of foil mulches on farmland birds, the impact of other, potentially important, factors should be taken into the account. Obviously, if a level of habitat disturbance is spatially extensive equating to a high cover of foil mulch in a landscape, then this should act in a negative manner on farmland birds. Thus, a negative relationship between foil cover in a landscape and bird species richness and abundance is expected.

When considering farmland as a habitat of birds, it is important to realize that species inhabiting agricultural landscapes are not a homogenous group. Some management schemes may affect particular species or functional groups of species in different ways than others, leading to a species group filtering effect (landscape-moderated functional trait selection hypothesis; Tscharntke et al. 2012). For example, some birds are farmland specialists but many other species are nonfarmland birds that occur in several other habitats, for example, forests and human settlements (Tryjanowski et al. 2009, 2011). Bird species also differ in their breeding biology, especially nesting site choice. Many farmland birds are ground nesters, while others build nests in shrubs, trees or buildings. It is clear that foil mulches should negatively affect farmland specialists and species breeding on the ground, but nonfarmland birds and above-ground breeding species may remain unaffected by the presence of foil mulches.

Use of plastic foils in agriculture increases every year (Scarascia-Mugnozza, Sica & Russo 2011); thus, it is important to understand how foil mulches affect the farmland bird community. We hypothesized that both the number of species and abundance of birds would be lower in sites covered by plastic foil than those in control sites without foil at the local scale and that at the landscape scale, higher cover of foils in a landscape will decrease the number of species and their abundance (Kruess & Tscharntke 2002; Kleijn & Sutherland 2003). However, the decrease may be counteracted by the cover of semi-natural habitats and crop diversity (Kleijn et al. 2011; Tscharntke et al. 2005). We expected that foil mulches cause profound changes to the food resources of farmland birds (weed and invertebrate populations) that extend beyond the period of foil cover, and therefore, the negative effect of foil mulches on farmland birds would also be sustained during the period after foil is removed from crops (Cohen, Briand & Newman 1990; Williams & Martinez 2000). We also predict that foils affect mostly farmland specialists and ground nesting birds but not the above-ground nesters and nonfarmland species (Tscharntke et al. 2005, 2012).

Materials and methods

Study area

The study area was located in southern Poland, in agricultural landscapes situated to the north of the city of Kraków (Fig. 1). The elevation of the study area varies between 180 and 250 m a.s.l. The average annual temperature is 7 °C, and the annual precipitation is 600 mm (Kondracki 2002). The agriculture in this area is focused on the production of vegetables. Vegetable crops constitute over 50% of all crops, and cabbage fields predominate especially in the parishes of Igołomia, Nowe Brzesko and Charsznica. The other most common crops are cereals (25%). Many of the cabbage fields are mulched with plastic foil typically for a period of around 6 weeks from the beginning of April to the middle of May (Appendix S1, Supporting information). The mulched fields are entirely covered with white foil during this period.

Figure 1.

Map of Poland showing the study area and the 25 1×1 km landscape plots (grey squares).

Selection of plots and bird censuses

The effect of mulching on birds was studied in two ways, which corresponded with two spatial scales, namely small, circular local plots (3·14 ha) and large landscape plots of 1 × 1 km (100 ha).

Local scale survey

We selected areas with a high cover (over 80%) of foil mulch and compared these with areas without mulching. We selected 50 points in areas with a high cover of foil mulch and an equal number of points in the foil-free control areas. At each point, birds were counted within a 100 m radius; thus, the sample unit, hereafter referred to as ‘the point’, was 3·14 ha. Both foil and control points were chosen in accordance with several criteria in order to keep landscape structure and other potentially confounding variables as constant as possible (Appendix S2, Supporting information).

Between the beginning of April and the end of June 2011, we made four bird counts. We divided the breeding season into two periods: during mulching, from the beginning of April to the middle of May and once the foil was removed, from the middle of May to the end of June. Two counts were carried out in each period. Counting dates were homogeneous between the points with and without foil mulches. At each survey point, birds were counted for a period of 10 min. The surveys were conducted between sunrise and 10 a.m. under favourable weather conditions, with no rain and wind rating below 3 on the Beaufort scale (Bibby, Burgess & Hill 1992).

Landscape scale survey

We selected 25 100 ha landscape plots (1×1 km) differing in the percentage cover of foil mulch (Table 1). We chose the plots in such a way as to ensure that all possibly confounding variables were not correlated either with each other or with the foil mulch cover, and we included them as covariates in the subsequent analysis (Table 1).

Table 1. Characteristics of the 25 landscape plots (100 ha each)
Variable codeDescriptionMeanSEMinMax
Foil coverFoil mulch cover (%)172·7043
Field sizeMean field size (ha)1·30·10·32·5
SettlementHuman settlement cover (%)40·508
ForestForest cover (%)40·509
CabbageCabbage crop cover (%)591·84070
GrasslandGrassland cover (%)81·3018
Crop diverDiversity of crops (Simpson reciprocal diversity index)2·00·131·123·53

Between the beginning of April and the end of June 2011, we carried out four surveys. Two surveys were included during the mulching period and two when the foil was removed, as detailed in the section on local scale survey methods above. The observers walked through the plot (both along field borders and inside crop fields) for three hours, taking an arbitrary path (about 5 km long); however, every plot was covered, visually and aurally, in its entirety. The exact location and number of all the birds seen or heard within the sampling area were recorded on a map, with the exception of those which simply passed over in flight. The surveys were conducted between one hour after dawn and 11 a.m. and under favourable weather conditions (see above).

Lepidoptera and weed counts

We estimated two food resources of farmland birds: species richness and density of butterflies and species richness of weeds.

We counted all adult butterflies, and we also estimated abundance of larvae of two of the most common species: large white Pieris brassicae and small white P. rapae that are regarded as pests of cabbage crops.

To count adult butterflies, we established forty 100 m transects (20 in mulched fields and 20 in foil-free cabbage fields). We counted butterflies in a 5-m-wide belt (2·5 m on both sides of the transect path). We made six counts, three during mulching and three after foil was removed (at the beginning of June and July and in late July).

Butterfly larvae were counted in late July in the same 40 fields (20 covered by foil and 20 without foil) where adult butterflies were counted. In each field, we inspected 30 cabbage heads to search for larvae of the two Pieris butterflies. We noted presence and number of larvae of each butterfly species. In total, 1200 cabbage heads were inspected.

Weed species were counted in the same fields as butterfly larvae. In each field, two square 100 m2 plots were set. Counts were performed twice, at the end of May and in late July. Forty fields were investigated (20 covered by foil and 20 without foil) with 80 plots in total.

Data handling and statistical analysis

The larger of the two bird counts for both the points and the landscape plots during each period was used in the subsequent summaries and analyses; the analysis of mean counts gave very similar results. Each species was classified as: a farmland specialist or nonfarmland species, and ground nesting or above-ground nesting species, based on Tryjanowski et al. (2009) (Appendix S2, Supporting information). Statistical analyses were performed on total species number and abundance as well as on data split into the aforementioned groups.

To compare the bird species richness and abundance between the foil points and the control points during the mulching period and after the removal of the foil, we used a General Linear Mixed Model (GLMM). Habitat type (two levels: mulched sites and foil-free controls sites) and period (two levels: before 15 May and after 15 May) were introduced as fixed effects, and point ID nested in the habitat type was a random factor.

Most of species recorded at a point were single individuals; therefore, we compared species occurrences between the two habitat types, rather than comparing species abundance. To compare the occupancy rate of a given species between mulched and control points, we used a GLMM with logit link function and binomial error variance. The fixed and random effects were the same as in models described above.

A GLMM with identity link function was also used to compare the butterfly species richness and abundance between the transects on mulched and control fields during the mulching period and after removal of the foil. The structure of the model was the same as for the bird models. To compare the proportion of cabbage heads infested by Pieris brassicae and P. rapae, a GLMM with logit link function and binomial error variance was used. Habitat type was introduced as a fixed effect, and field ID nested in habitat type was a random factor. To compare the number of larvae of both pest species on cabbage heads, we used a GLMM with identity link function. Only cabbage heads with at least one larvae were considered in this analysis.

To compare weed species richness between mulched and control fields in the second half of May and at the end of July, we used a GLMM with the identity link function. Habitat type and period were introduced as fixed effects, and plot ID nested in field ID and the latter nested in habitat type were random factors.

In all GLMMs, the interaction between habitat type and period was introduced in initial parameterization. The interaction was removed from models if it was nonsignificant.

To identify the factors affecting bird species richness and abundance in the 1 × 1 km landscape plots, we used model selection procedures based on information theory (Burnham & Anderson 2002). The Akaike Information Criterion, corrected for small sample size (AICc), was used to identify the most parsimonious models from each candidate set, with 127 models tested for each dependent variable (Appendix S2, Supporting information). Finally, we ranked all the models built in accordance with their ∆AICc values and used those with the lowest AICc, together with the associated weight values, the probability that a given model is the best, as those best describing the data. We considered models with ∆AICc lower than two as equally good (Burnham & Anderson 2002). We used model averaging for estimates of the function slopes of the parameters of interest (Burnham & Anderson 2002). Finally, the model weights were used to define the relative importance of each explanatory variable across the full set of models evaluated by summing up the weight values of all models that contained the explanatory variable of interest (Burnham & Anderson 2002).

All GLMMs were performed in SPSS version 19 for Windows (SPSS Inc. 2010, Armonk, NY, USA). Model selection and averaging were run in the SAM 4.0 statistical software (Rangel, Diniz-Filho & Bini 2010). We did not find evidence for statistically significant spatial autocorrelation in our data by examining Moran's statistics. When necessary, we used logarithmic transformation to reduce the effects of outlier observations (Quinn & Keough 2002). We considered that the function slopes, the betas, were significant if their 95% confidence intervals (95% CI) did not overlap with zero. All the statistical parameters, betas and means, are quoted ± standard error (SE).

Results

The effect of foil mulches on bird communities at the local scale

We noted 31 bird species during the point counts (Appendix S2, Supporting information). Both the number of species (GLMM F1, 98 = 205·034, < 0·001) and the abundance of birds (GLMM F1, 98 = 130·174, < 0·001) were significantly lower at the mulched points than at the control ones (Fig. 2a). The number of species (GLMM F1, 98 = 49·707, < 0·001) and abundance (GLMM F1, 98 = 33·324, < 0·001) were lower during mulching than after foil removal (Fig. 2a). These analyses gave similar results when performed separately for farmland specialists, nonfarmland birds, ground nesters and above-ground nesters (Fig. 2b–e; Appendix S2, Supporting information). Analysis of the occurrence of species at the points revealed that 13 species had a significantly lower and none had significantly higher occupancy in the mulched points compared with the control points in either of the periods (Appendix S2, Supporting information).

Figure 2.

Species richness (white bars) and number of individuals (grey bars) per point of (a) all species, (b) farmland specialists, (c) nonfarmland birds, (d) ground nesting species and (e) above-ground nesting species during the mulching period and the period after the foil was removed at the points (3·14 ha each) covered with foil and the foil-free control sites. Whiskers are 95% confidence intervals.

The effect of mulching on bird communities at the landscape scale

In total, we noted 64 species in 25 landscape plots (Appendix S2, Supporting information).

There were three equally good models explaining 67% of the variation in the number of species during the mulching period (Table 2). Species richness was negatively affected by foil cover in the landscape (Table 3, Fig. 3a), crop diversity and by average field size, but positively by forest cover in a landscape (Table 3).

Table 2. Best models describing the species richness and abundance of birds in the farmland landscape plots during the mulching period and the period after removal of the foil. The number of parameters in a model (k), the variance explained by the model (r2), the Akaike Information Criterion score (AICc), the difference between the given model and the most parsimonious model (Δ) and the Akaike weight (w) are listed. Field size – mean field size, foil cover – foil mulch cover in a landscape, crop diver – diversity of crops (Simpson diversity index), forest – forest cover, grassland – grassland cover, cabbage – cabbage crop cover
Model k r 2 AICcΔ AICc w
Species richness during mulching
 Field size + foil cover30·72138·16400·215
 Field size + foil cover + crop diver40·74139·8231·6590·094
 Field size + forest + foil cover40·73140·1081·9440·071
Abundance during mulching
 Foil cover20·85229·08900·128
 Field size + foil cover30·87229·6690·5800·096
 Cabbage + foil cover30·86230·8461·7560·053
 Foil cover + crop diver30·86230·9241·8340·051
 Grassland + foil cover30·86231·0851·9960·047
Species richness after the removal of the foil
 Foil cover + crop diver30·622142·02900·176
 Foil cover20·554143·3271·2980·092
 Grassland + foil cover + crop diver40·638144·0091·9800·063
Abundance after the removal of the foil
 Foil cover20·57266·26100·244
Table 3. Estimates of the function slopes of variables present in the most parsimonious models describing the species richness and abundance of birds in the farmland landscape plots during the mulching period and the period after the removal of the foil. Importance of each explanatory variable was calculated by summing up the weight values of all models that contained the variable. Standard errors (SE) and 95% confidence limits (CL) are also presented. A variable which confidence limits of the estimate of function slope overlapped with zero is in italic. Foil cover – foil mulch cover in a landscape, field size – mean field size, crop diver – diversity of crops (Simpson diversity index), forest – forest cover, cabbage – cabbage crop cover, grassland – grassland cover
VariableImportanceEstimateSELower 95% CLUpper 95% CL
Species richness of birds during mulching
 Foil cover1·000−0·3060·062−0·427−0·185
 Field size0·789−3·1341·061−5·214−1·054
 Crop diver0·280−1·9810·587−3·131−0·831
 Forest0·2480·2820·0710·1430·421
Abundance of birds during mulching
 Foil cover1·000−3·6410·352−4·332−2·951
 Field size0·352−11·0312·918−16·750−5·311
 Grassland0·3211·3470·4090·5452·149
 Crop diver0·31022·4928·1266·56538·419
 Cabbage 0·280 0·798 0·556 0·291 1·887
Species richness of birds after the removal of the foil
 Foil cover0·999−0·2850·060−0·403−0·167
 Crop diver0·544−2·8050·978−4·723−0·887
 Grassland0·2590·1150·0370·0410·188
Abundance of birds after the removal of the foil
 Foil cover1·000−3·7680·719−5·177−2·360
Figure 3.

The effect of a foil mulch cover in the landscape on species richness and abundance of all birds during the mulching period (a, b) and the period after the foil was removed (c, d).

There were five best models explaining the number of individuals during the mulching period (Table 2). These models explained 86% of the variation in the number of individuals. The abundance was negatively affected by foil cover in the landscape (Fig. 3b) and average field size (Table 3) but positively by grassland cover and crop diversity (Table 3).

There were three equally good models explaining 60% of the variation in species richness in the period after the removal of the foil (Table 2). Species richness in this period was negatively affected by previous foil cover in the landscape (Fig. 3c) and crop diversity but positively by grassland cover (Table 3).

There was only one best model explaining 57% of the variation in the number of individuals in the period after the removal of foil (Table 2). The only variable in this model was the previous cover of foil in the landscape, negatively affecting bird abundance (Fig. 3d, Table 3).

When the above analyses were performed separately on farmland specialist, nonfarmland birds, ground nesting species and above-ground nesting species, the effect of foil mulches was still negative and statistically significant in two periods after controlling for other variables (Appendix S2, Supporting information). The effect of other factors was highly variable depending on the species group. Only grassland cover always had a positive impact on all species groups when it was in the best models (Appendix S2, Supporting information).

The effect of mulching on butterflies and weeds

We noted 207 individuals of 18 butterfly species (Appendix S3, Supporting information). Both the number of species (GLMM F1, 38 = 13·530, = 0·007) and the abundance of butterflies (GLMM F1, 38 = 14·211, P = 0·004) were lower in mulched transects (Fig. 4a, Appendix S3, Supporting information). The number of butterfly species (GLMM F1, 39 = 41·294, P < 0·001) and their abundance (GLMM F1, 39 = 41·294, P < 0·001) were lower during mulching than after foil removal (Fig. 4a).

Figure 4.

The effects of mulching on (a) species richness (white bars) and mean number of individuals per count (grey bars) of butterflies, (b) proportion of infested cabbage heads (white bars) and number of larvae on the cabbage head (grey bars) of Pieris brassicae and P. rapae and (c) number of weed species. Whiskers are 95% confidence intervals.

The proportion of cabbage heads infested by larvae was smaller in mulched fields than that in foil-free fields for large white (GLMM F1, 30 = 10·639, P < 0·001) but not for small white (GLMM F1, 39 = 1·529, = 0·224; Fig. 4b). The mean number of larvae per cabbage head was smaller in mulched fields than that in foil-free fields for large white (GLMM F1, 36 = 17·875, P < 0·001 Fig. 4b) but not for small white (GLMM F1, 27 = 0·411, P = 0·527; Fig. 4b).

In total, we noted 32 weed species (Appendix S4, Supporting information). We found a statistically significant interaction between habitat type and period for weed data (GLMM F1, 78 = 4·120, P = 0·046). Tukey's post hoc tests showed that mean number of weed species per 100 m2 plot was lower in mulched fields than that in control fields during the mulching period but not after removal of the foil in late July (Fig. 4c; Appendix S4, Supporting information).

Discussion

Our study has revealed for the first time that the use of plastic foil for mulching had a negative effect on the species richness and abundance of farmland birds at both the small scale of individual fields and the landscape scale. Interestingly, the negative effect of foil mulch was also still present after foil removal in the middle of May. Contrary to expectations, foil did not act as a group-specific environmental filter and all groups of birds (farmland specialists, nonfarmland birds, ground nesting species and above-ground nesting ones) were negatively affected by mulching. Potential food resources for birds, adult and larval butterflies and weed species, also had lower density and species richness in mulched fields. This suggests that in accordance with our hypotheses, widespread foil use for crop mulching acted as a habitat disturbance preventing birds from access to food resources and breeding sites. Mulches may also have altered resources that, probably via the trophic cascade, negatively affected birds even after the foils were removed. These may add to the general negative trends in farmland biodiversity and the decline in the suitability of agricultural landscapes for birds.

The use of plastic foil for mulching is now widespread (Scarascia-Mugnozza, Sica & Russo 2011). Compared with other mulch types, such as living mulches or straw mulches, plastic foil is very much cheaper, enables large areas to be covered and is easier to handle and to remove (Durham 2003). Covering a field with foil means that all the habitat resources for birds are unavailable for breeding and foraging. We also cannot exclude the possibility that the reflection from the foil repels some species, preventing them from settling in these fields. However, this does not explain why the negative effect of foil use extends into the period after it is removed. After removal of foils, the number of birds did not increase to the level noted in control mulch-free sites and the landscapes without mulched fields.

Overall, species richness and abundance of species increased during the period after the foil was removed. This is probably linked in part with phenology and the arrival of late, long-distance migrants such as whinchat Saxicola rubetra, red-backed shrike Lanius collurio or marsh warbler Acrocephalus palustris. However, even they did not entirely fill the gaps brought about by the plastic foil. As we demonstrated, mulches altered habitat and food resources (weed species and adult butterflies and their larvae) even after foil removal, probably making those parts of the farmland less suitable for all group of birds. This result is in agreement with other studies on the effects of foils on invertebrates (Ricotta & Masiunas 1991; Hooks & Johnson 2003; Mahajan et al. 2007). It is also possible that some nonterritorial individuals known as floaters (Newton 1998) and breeding dispersers may have settled in fields after the removal of the foil, since occupancy at the mulched points increased slightly after foil removal for some species such as grey partridge Perdix perdix, Eurasian skylark Alauda arvensis and corn bunting Miliaria calandra, while occupancy remained stable in the control areas. Both the mulched and control sites were treated with pesticides and herbicides; however, the mulched sites were usually sprayed once, whereas pesticides were applied two or three times at the control sites. However, it is unlikely that this could explain why the number and abundance of bird species were higher at control points than mulched points. Numerous works have shown that pesticide and herbicide use has both direct and indirect detrimental effects on farmland bird species richness, abundance, reproduction and food resources of birds, such as weeds and insects (Moreby & Southway 1999; Holland et al. 2006; Geiger et al. 2010). Thus, it is possible that if we were able to control for the rate of pesticide use, the negative effect of foil mulches would be much more pronounced than those recorded in our study. The lower pest density and weed species number, together with lower pesticide and herbicide use in mulched sites, indicate that this method of crop production is somehow more environmentally friendly than conventional crop production. As we demonstrated, however, this decreases the number of farmland birds and ecosystem services potentially provided by these organisms (e.g. natural pest control). This raises a serious conservation dilemma of how to minimize the negative effects of pests and weeds on crops while sustaining high species richness and abundance of birds.

Our surveys on the large landscape plots showed that other variables also had significant effects on bird species richness and abundance. Thus, according to theory (Kleijn et al. 2011; Tscharntke et al. 2012), spatial heterogeneity may, at least to some degree, mitigate the bird habitat disturbance caused by the use of foils on crops. However, most of the analysed factors (crop diversity, forest and human settlement covers, cabbage crop cover) had different (positive or negative) effects depending on the group of species (Appendix S2, Supporting information); thus, they are of little practical application when conservation and landscape planning for birds are concerned. Only grassland cover in the landscape always had a positive effect on birds when it was in best models. Grassland patches spread through a landscape dominated by intensively managed crops play a key role in sustaining farmland biodiversity (Söderström & Pärt 2000; Batáry, Báldi & Erdős 2007). For birds, they provide suitable breeding and foraging habitats (Concepción et al. 2012). In our study area, the grasslands were usually small patches that were either intensively mown or grazed by cows, which is particularly good for birds foraging on the ground because prey are easy to catch in those conditions (Morris & Thompson 1998). Thus, the possibility of mitigating the negative effect of foil mulches on birds in a landscape is the conversion of some fields into grassland patches.

Amongst other factors that potentially could be of use when mitigating the effects of foils, field size is particularly interesting. More bird species were present in landscapes with smaller fields. This was also true for farmland specialists. We counted birds in different parts of large fields; thus, this effect could not be attributed to greater species detectability in smaller fields. Agricultural landscapes with smaller fields are inhabited by more species (e.g. Söderström & Pärt 2000; Herzon & O'Hara 2007). Fields in Poland are separated from each other by narrow grass strips (Appendix S1, Supporting information); thus, the smaller fields increase the density of these potentially important microhabitats for birds (Concepción et al. 2012). Hence, another initiative that could help to maintain species richness of farmland birds in landscapes with mulched fields is to divide large fields into smaller ones.

Synthesis and application

Plastic foil mulches are an efficient measure to control invertebrate pests and weeds; however, they are detrimental to farmland birds. This raises a serious conservation dilemma. If conservation of farmland birds is a priority in a given landscape, the use of foil mulches should be carefully considered. This would require a monitoring of species richness and abundance in a target area before the use of foils to estimate their potential impact. We suggest maintaining a low cover of plastic foil mulches in the landscape; on the basis of our regression equations, it seems probable that a cover of less than 20% is sufficient to support c. 80% of all the species that occur in the areas without the foil mulches. Small field sizes and a high proportion of grassland cover in the landscape may also alleviate the negative effect of foil mulches on birds. This would require actions involving many farmers because average farm size in our study area is small (<10 ha). The problem of foil mulches should be regulated in the framework of the common agricultural policy of the European Union. For example, farmers should be encouraged to limit the area covered with plastic mulches and select late-growing vegetable varieties. However, to state specific recommendations, more detailed studies are required and they should focus on the estimation of potential profits resulting from lower pest densities and weed species in mulched areas which may counterbalance potential costs linked with the loss of the ecosystem services provided by birds. A comparison of different types of mulch with mechanical and chemical methods of weed and pest control (including at a landscape scale) is also required in future studies.

Acknowledgements

We thank David Kleijn and three anonymous referees for helpful comments on the manuscript. We are grateful to Tim H. Sparks for linguistic improvements.

Ancillary