BOLETÍN OFICIAL DEL ESTADO
UNIDAD FORMATIVA 2
3. Control de la elaboración y puesta en obra de las armaduras pasivas
Many of the conclusions from my thesis were arrived at through analysis of large existing datasets comprising records collected by volunteers. Patterns of speckled wood’s habitat associations and brown argus’ host plant associations were elucidated using butterfly records collated by Butterfly Conservation and the Centre for Ecology and Hydrology. For the brown argus analyses, plant records from a database of records collected by volunteers and collated by the Botanical Society of the British Isles were also used as well as butterfly abundance data from the UK Butterfly Monitoring Scheme. These datasets have been used widely in scientific research, particularly in detecting changes in species’ responses to climate change, such as changes in
distribution (Parmesan et al. 1999), abundance (Roy et al. 2001) and phenology (Roy and Sparks 2000). These datasets are, however, not without their problems which can limit their use.
6.2.1.1 Distribution datasets
Distribution data are collected non-systematically and volunteers are encouraged to submit records of any butterfly species they have observed, although effort is made to achieve complete geographic coverage at a 10 km × 10 km grid square resolution for the
production of published Atlases. Problems arise, therefore, when trying to detect trends in the data because of variation in recorder effort. Spatially, recorder effort is much greater in areas with high human population densities, and recorder effort has increased substantially through time (Fox et al. 2006). The first butterfly atlas (Heath et al. 1984) was based on 185,649 records submitted in the period 1970-82, whereas the second atlas (Asher et al. 2001) was based on 1,710,586 records from the period 1995-99 and the third (Fox et al. 2006) 1,616,620 records submitted in the period 2000-04.
Furthermore, the proportion of records submitted at a fine resolution (100 m grid
resolution), which are particularly useful when considering species’ habitat associations, has also increased over time.
Thus analyses may be biased by variation in recorder effort. My estimation of leading-edge range margin expansion by the brown argus butterfly, for example, could have been exaggerated by an increase in recorder effort. However, the estimation of range change was based on data pooled for each study period (1970-1987 and 1988- 2009) which should improve coverage and hence the estimate of the location of the northern range boundary for each period. In addition, because the butterfly was
localised in the first period its 10 km grid square resolution distribution is likely to have been well described. Furthermore, I only used records south of the butterfly’s zone of hybridisation with the northern brown argus, Aricia artaxerxes (Mallet et al. 2011) and thus my estimated rate of expansion will have been an underestimate if ‘pure’ southern brown argus individuals have expanded into this region.
The plant database also has problems with variation in recorder effort. Fine resolution (100 m) records vary considerably in their abundance between counties, presumably due to variation in the enthusiasm of local recorders. Furthermore, recorder effort also varies between host plant species which presented a problem in Chapter 5 when quantifying the availability of alternative host plants in the landscape for brown
argus. Rockrose is a species of unimproved or semi-improved calcareous grasslands which tend to be relatively species-rich and hence attract a lot of recording effort. Dove’s-foot cranesbill, by contrast, is a common species which is less likely to compel people to submit records; and it occurs in habitats that are less likely to be visited by recorders. Thus achieving a robust comparison of the distribution of the two host plant species in the landscape is difficult. In Chapter 5, I tried to account for this issue by analysing data only from two counties that were well recorded, but the coverage of dove’s-foot cranesbill in these counties is still likely to be an underestimate.
6.2.1.2 Transect data
Data from the UK Butterfly Monitoring Scheme are standardised: transects are only walked in certain weather conditions, the same route is repeated each year and the same recording methodology used so that measures of butterfly abundance can be compared between sites and years (Pollard and Yates 1993). However, these data also have certain biases because many of the transects are on “high-quality” sites such as nature reserves and so butterfly trends in these sites may not be representative of the wider landscape (Fox et al. 2006). Transects are also much more numerous in the south of Britain than the north, and those in the north tend to have been added to the scheme more recently. However, this did not cause a problem for the analysis of brown argus data because the butterfly has a southern distribution in Britain. When investigating speckled wood habitat associations I used transect data to test whether the species’ strength of
association with woodland was related to butterfly density. Changes in butterfly habitat associations can be quantified by calculating the proportion of individuals seen in different sections (habitats) on the transect (Oliver et al. 2009). However, this analysis did not reveal any clear trends in habitat associations. This may have been due to the low number of transects in northern Britain, which from my analysis in Chapter 3,
appears to be where much of the variation in habitat associations occurs. Alternative methods are, therefore, required to investigate the effect of density on the butterfly’s habitat associations.