ANTES DEL SILENCIO 2.1 Sobre el concepto de silencio
79 Ignacio Siles vincula el concepto de sociedad de la información‐ que otros teóricos
3.2. Formas del silencio Instalar el silencio en el espacio narrativo
Data on the occurrence of pollinator species (solitary and social bees) recorded per 10km x 10km square was provided by the Bees, Wasps and Ants Recording Society (BWARS, http://www.bwars.com). Subsequently our analysis was restricted to those bee species known to be associated with pollination of crops. The area of insect pollinated crops (hectares per 2km grid) was obtained from agricultural census statistics from UK government departments (e.g. DEFRA) downloaded from EDINA (http://edina.ac.uk/). We focussed on key insect-pollinated crops namely: oilseed rape (Brassica napus), field bean (Vicia fabia) and broad crop groups: soft fruits (strawberry, blackcurrant, redcurrant, gooseberry, blackberry and raspberry) and top fruits (apple, pear, plum and cherry). The pollinator species occurrence data were then standardised for variable recording effort (Hill, 2011) and the number of pollinator species adjusted based on neighbourhood predictions modelled from agricultural data (EDINA 2010 England and Wales 5km resolution, soft fruit data 2004 data; 2010 Scotland 2km resolution).
The degree of correlation between pollinator and crop abundance was investigated to identify places within the UK that grew high proportions of a given crop but had an apparent deficit in that crops respective pollinators. Potential mismatches between the distributions of crops and their pollinators were visualised by plotting the standardised residuals from linear regression of each crop’s area (hectares of oilseed rape, field bean, soft fruits, top fruits, respectively) against the adjusted richness of its associated pollinators. Autocovariate models were used to address spatial autocorrelation in the data by estimating how much the response variable at any given site reflected response values at surrounding locations. This is achieved by fitting a distance-weighted function of neighbouring response values to the model’s
explanatory variables to capture autocorrelation arising from processes such as limited dispersal and movement of censused individual pollinators between sampling sites (Dormann et al., 2007).
10.5.3. Results
10.5.3.1 Objective 1: Model the density of pollinator (bumblebee) wild flower resources at a fine
resolution across the British landscape
Overall, the density of wild flower resources across the British landscape for pollinators was greater in lowland areas and in the south and west of Britain (Figure 10.18). That the south-west of England was relatively high in floral resources probably reflected the wetter, milder climate and lower land-use intensity compared with the more intensively farmed east of England (e.g. Lincolnshire and east Anglia). The uptake of AES was similarly heterogeneous across the British landscape and tended to follow the same southwesterly distribution pattern as wild plant resources (Figure 10.19).
The best fitting model (GAMM) comprised mean annual temperature, mean monthly accumulated rainfall, a first order interaction between mean annual temperature and mean monthly accumulated rainfall, nitrogen deposition, broad habitat type, altitude, woody cover in each plot and geometry (habitat patch size and total length of linear features in each 1km square). Models also included a spatial trend surface in the form of an interaction term for the spatial location of each 1km square and the random effect of survey square on between plot variance.
To map the response variable of interest across Great Britain, we require the covariate information to be available at the same resolution and at the same locations as we wish to map. Some of the variables chosen in our best fitting model were, however, not available over the whole of Great Britain. Therefore, in order to demonstrate the mapping capabilities of the method we modeled the data once again, but only using covariates that we have information on over the whole of Great Britain. This shows that mapping is possible with the GAMM-based approach even when data for more finely resolved predictors is unavailable. The model was then used to predict the nectar producing plant species richness for bees over Great Britain at a 1km square resolution (Figure 10.18). This model is therefore capable of up- scaling to national estimates and demonstrates the usefulness of such an approach. As the model used
was based only on covariates that were available at a national scale, estimates will not be as accurate as would be if the best fitting model could be applied. Nevertheless, though the results from these maps may not be considered as being truly reflective or entirely accurate because of issues with model fitting, these results show the method works well and can produce informative national coverage maps. Any shortfall here does not come from the modeling approach but simply through a lack of data.
Figure 5.18. Predicted mean bee nectar plant richness per 4m2 in 2007 averaged over the Broad Habitats in each 1km square. Predictions generated from a Generalised Additive Mixed Model that included Broad Habitat class, climate, patch size, length of linear features in each 1km square and agri-environment scheme (ELS) status in 2007 as explanatory variables.
Figure 5.19. Provision of pollen and nectar resources from the agri-environment schemes (based on analysis of AES uptake data for England)
10.5.3.2 Objective 2: Model the distribution of insect-pollinated crops and crop pollinator
richness across the British landscape and the extent of any spatial mismatch.
The modeling of the spatial overlap between crop area and crop pollinator richness showed that in some areas of Britain there is a potential for a spatial mismatch between insect-pollinated crops and the species richness of wild pollinators. For example, there is a high potential mismatch (shades of light blue) in the east of the country for field bean and its pollinators (Figure 10.21), where there is a high crop cover but lower pollinator resource. Similarly, Scotland supports an economically important soft-fruits industry in the south-east, and area that supports relatively low wild pollinator richness (Figure 10.23). Similar patterns for the other crops suggest that while in certain areas of the country there are sufficient pollinator resources to meet crop pollination needs, elsewhere the same crop may suffer from a deficit of wild pollination services (Figures 10.20-10.23). However, the available data only allowed an assessment of pollinator species richness. The abundance of individual pollinator species may be more important than species richness in the successful delivery of pollination services to crops. Such data on the abundance of wild pollinators are simply not available due to a lack of systematic monitoring in the UK and worldwide.
This approach of modeling crop areas and bee richness (from species occurrence records) produces maps of spatial pattern that are easy to interpret visually. These maps must, however, be treated with appropriate caution. The probability of each grid cell in the map having a certain value is strongly influenced by the values of surrounding cells (spatial autocorrelation). To quantify accurately the relationship between crop and associated pollinator distributions this spatial autocorrelation in the data was quantified and accounted for in a regression analysis (Dormann et al., 2007). For all four crops, spatial autocorrelation proved a highly significant variable in the regression analysis of crop area with pollinator species richness. Nonetheless, once this autocorrelation had been accounted for a highly significant relationship remained between crop area and crop pollinator richness (Table 10.6).
Table 10.6. Accounting for spatial autocorrelation, the results of linear regression of crop pollinator richness with four crop types.
Crop type AIC Crop ha Spatial autocorrelation
F P F P
Vicia faba 7853.93 11062.77 <0.001 42676.72 <0.001
Brassica napus 9137.66 13545.45 <0.001 60237.54 <0.001
Top Fruit 8702.69 2435.799 <0.001 65552.04 <0.001 Soft Fruit 8966.97 793.5909 <0.001 54337.39 <0.001
Figure 5.20. Oilseed rape (Brassica napus).
Distribution of oilseed rape (Brassica napus) crops, the species richness of its wild pollinators, and the potential for pollination mismatches. Pink shades indicate high overlap between crops and pollinator richness and light blue shades show low overlap. Dark blue indicates no crop grown.
Figure 5.21. Field bean (Vicia fabia) Distribution of Field bean (Vicia fabia) crop, the species richness of its wild pollinators, and the potential for pollination mismatches. Pink shades indicate high overlap between crops and pollinator richness and light blue shades low overlap. Dark blue indicates no crop grown.
Figure 5.22. Top fruit (orchard crops e.g. apples)
Distribution of top fruit (apples, pears, plums, cherries) crops, the species richness of its wild pollinators, and the potential for pollination mismatches. Pink shades indicate high overlap between crops and pollinator richness and light blue shades show low overlap. Dark blue indicates no crop grown.
Figure 5.23. Soft fruit (strawberries, raspberries etc.)
Distribution of Top fruit (strawberry, blackcurrant, redcurrant, gooseberry, blackberry and raspberry) crops, the species richness of their wild pollinators, and the potential for pollination mismatches. Pink shades indicate high overlap between crops and pollinator richness and light blue shades show low overlap. Dark blue indicates no crop grown.