Skip to main content
Dryad logo

Configurational crop heterogeneity increases within-field plant diversity

Citation

Alignier, Audrey et al. (2020), Configurational crop heterogeneity increases within-field plant diversity, Dryad, Dataset, https://doi.org/10.5061/dryad.t76hdr7xh

Abstract

1. Increasing landscape heterogeneity by restoring semi-natural elements to reverse farmland biodiversity declines is not always economically feasible or acceptable to farmers due to competition for land. We hypothesized that increasing the heterogeneity of the crop mosaic itself, hereafter referred to as crop heterogeneity, can have beneficial effects on within-field plant diversity.

2. Using a unique multi-country dataset from a cross-continent collaborative project covering 1451 agricultural fields within 432 landscapes in Europe and Canada, we assessed the relative effects of compositional and configurational crop heterogeneity on within-field plant diversity components. We also examined how these relationships were modulated by the position within the field.

3. We found strong positive effects of configurational crop heterogeneity on within-field plant alpha and gamma diversity in field interiors. These effects were as high as the effect of semi-natural cover. In field borders, effects of crop heterogeneity were limited to alpha diversity. We suggest that a heterogeneous crop mosaic may overcome the high negative impact of management practices on plant diversity in field interiors, whereas in field borders, where plant diversity is already high, landscape effects are more limited.

4. Synthesis and applications. Our study shows that increasing configurational crop heterogeneity is beneficial to within-field plant diversity. It opens up a new effective and complementary way to promote farmland biodiversity without taking land out of agricultural production. We therefore recommend adopting manipulation of crop heterogeneity as a specific, effective management option in future policy measures, perhaps adding to agri-environment schemes, to contribute to the conservation of farmland plant diversity.

Methods

Region and landscape selection

The study was conducted in eight agricultural regions comprising seven regions in Europe and one region in eastern Canada (near Ottawa). The European regions followed a south-to-north gradient, with four regions in France (near Arles, Niort, Rennes, Toulouse), one in England (centred on Ely, Cambridgeshire), one in Germany (near Goettingen) and one in Spain (near Lleida; Fig. 1). Within these agricultural regions, we selected a total of 432 1 km × 1 km landscapes, with 60 to 90 % of crop cover in each. These landscapes represented, by design, uncorrelated gradients of compositional crop heterogeneity, assessed by the Shannon diversity index of the crop cover types, and of configurational crop heterogeneity, assessed by the total length of crop field borders (see Pasher et al., 2003 and Sirami et al., 2019 for details). The landscape selection process used the most recent remotely sensed data or land cover map available for each agricultural region (see Table S1 in Supporting Information).

While land cover maps were adequate for landscape selection, their coarse spatial resolution did not allow for the accurate delineation of narrow strips of non-crop covers between fields. Thus, all landscapes were digitized from aerial photos to create detailed maps delineating all fields managed for agricultural production (including crops, and temporary and permanent grasslands), linear semi-natural boundaries between crop fields and non-crop patches. Non-crop cover types included woodland, open land, wetland and built-up areas. Linear semi-natural boundaries included hedgerows, grassy strips and watery boundaries such as ditches. These maps were visually validated by field crews within each agricultural region before the sampling of the vegetation in a given landscape.

Based on these more accurate and recent maps, several landscape variables were calculated. Compositional crop heterogeneity was assessed using the Shannon diversity index of agricultural cover types (Crop_SHDI). Configurational crop heterogeneity was measured as the total field border length (Crop_TBL). Crop_TBL was the sum of perimeters of all fields within the 1 x 1 km landscape minus the length of perimeters artificially created by intersection with the limits of the 1 km × 1 km landscape. The percentage of semi-natural cover types (Seminat_Cover) was calculated as the sum of the proportions of woodland, open land and wetland in the landscape. The length of semi-natural boundaries (Seminat_boundary) was calculated as half of the sum of the perimeter of woody, grassy and watery boundaries in the landscape.

Sampling site (field) selection

Within each landscape, we selected three to four sampling sites. Sampling sites were fields managed for agricultural production including crops, temporary and permanent grasslands. Fields were selected such that at least one contained the dominant crop type in the region, the other fields being representative of crops present within the focal landscape. Fields were at least 200 m apart, at least 50 m away from the border of the 1 km × 1 km landscape and at least 50 m away from large non-crop cover type patches such as woodland. We selected fields bordered by a similar boundary types within each region, i.e. only grassy strips or hedgerows, wherever possible. In total, 1451 agricultural fields were sampled.

Vegetation sampling

Within each sampling site, we surveyed within-field plant species along two parallel, 1 m wide and 50 m long transects, one located on the field border, the other within the field interior resulting in 2788 transects surveyed. Transects were about 25 m distant from each other. We sampled five plots (4 m × 1 m) along each transect, i.e. 20 m² per transect. Note that in Ottawa, transects were 2 m wide and the field border transect encompassed part of the boundary vegetation. We verified that this slight difference in sampling protocol did not affect our conclusions. Percentage cover of all vascular plant species was recorded. We conducted these plant surveys over two years between 2011 and 2014, each sampling site being sampled only within a single year. Surveys were conducted once before crop harvesting, except in Ely, Goettingen and Ottawa where surveys were conducted twice. In those regions, we pooled within-field plant data from the two visits per year and retained the total number of plant species for each sampled plot. Plant nomenclature followed TaxRef (Gargominy et al., 2014).

Data processing

Following Whittaker (1972), we used the multiplicative diversity partitioning method to assess plant species diversity components where β = γ/α. Gamma diversity (γ) was the total number of species across all plots sampled in a given transect and alpha (α) diversity was the number of within-field plant species present in each plot averaged across the five plots per transect. This measure of beta diversity (β) describes variation in plant species composition in the whole transect by comparison with an average plot.

Usage Notes

Read-me:

Field (column) name

Description

Year      

Year of point count : 2011, 2012, 2013 or 2014

Region 

Region of point count

Study site          

Name of the nearest big city

Landscape         

Landscape ID sampled (1km²)

Field     

Sampling site ID sampled (field)

Transect             

Sampling transect (border “B”/center “C” of the field)

Gamma

Total number of species across all plots sampled in a given transect

Alpha   

Number of within-field plant species present in each plot averaged across the five plots per transect

Beta     

Its was calculated as Gamma/Alpha. It describes variation in plant species composition in the whole transect by comparison with an average plot

Crop     

Crop sampled

Boundary           

Type of boundary

Crop_SHDI        

Shannon diversity index of agricultural cover types within the 1 x 1 km landscape

Crop_TBL           

Sum of perimeters of all fields within the 1 x 1 km landscape minus the length of perimeters artificially created by intersection with the limits of the 1 km × 1 km landscape

Woodland_Cover          

Percentage of woodland cover within the 1 x 1 km landscape

Open_Cover    

Percentage of open cover within the 1 x 1 km landscape

Water_Cover   

Percentage of water cover within the 1 x 1 km landscape

Seminat_Cover

Percentage of semi-natural cover within the 1 x 1 km landscape. It was calculated as the sum of the proportions of woodland, open land and wetland in the landscape.

Built_area_Cover           

Percentage of built area cover within the 1 x 1 km landscape

Crop_Cover      

Percentage of crop cover within the 1 x 1 km landscape

Non_Crop_Cover          

Percentage of non crop cover within the 1 x 1 km landscape

Bare_ground_boundary_m               

Total length of bare ground boundaries within the 1 x 1 km landscape

Grassy_boundary_m   

Total length of grassy boundaries within the 1 x 1 km landscape

Track_boundary_m      

Total length of track boundaries within the 1 x 1 km landscape

Woody_boundary_m  

Total length of woody boundaries within the 1 x 1 km landscape

Watery_boundary_m  

Total length of water boundaries within the 1 x 1 km landscape

Seminat_boundary       

Total length of semi-natural boundaries. It was calculated as half of the sum of the perimeter of woody, grassy and watery boundaries in the landscape

Total_boundary_m       

Total length of boundaries within the 1 x 1 km landscape

Funding

French National Research Agency, Award: ANR-11-EBID-0004

German Ministry of Research and Education

German Research Foundation and Spanish Ministry of Economy and Competitiveness

UK Government Department of the Environment, Food and Rural Affairs (Defra), Award: WC1034

Natural Sciences and Engineering Research Council of Canada (NSERC) Strategic Project grant

Canada Foundation for Innovation

Canada School of Energy and Environment

Environment Canada (EC)

Agriculture and Agri-Foods Canada (AAFC)