Abstract from BTO Research Report No 367:

S.E. Newson & D.G. Noble (2005)

The production of population trends for UK mammals using BBS mammal data, 1995-2002

1. The BTO/RSPB/JNCC Breeding Bird Survey (BBS) was expanded in 1995 to record mammals as well as birds. This was the first multi-species, annual mammal survey to be carried out in the UK. It focuses on large-sized easily identifiable species, although observers record any mammal species seen or known to be present. In this report we build upon the findings and recommendations of Newson & Noble (2003) by using BBS mammal data for 1995-2002 to generate estimates of population change.

2. Annual indices of relative abundance are produced at a national scale for nine mammal species for 1995-2002 - Brown Hare, Mountain/Irish Hare, Rabbit, Grey Squirrel, Red Fox, Red Deer, Fallow Deer, Roe Deer and Reeves’s Muntjac. Comparing the abundance of these species in 2002 relative to 1995, Grey Squirrel, Roe Deer and Reeves’s Muntjac were significantly higher in 2002, whilst Rabbit, Mountain Hare, Red Fox, Red Deer and Fallow Deer were significantly lower. Several of these species show significant fluctuations in abundance between years, whereas Roe Deer and Reeves’s Muntjac have increased progressively during this time.

3. Newson & Noble (2003) examined the potential for producing regional indices of relative abundance for three broad regions of Britain, the north, the south-east and the south-west. These regions do not correspond with any political jurisdictions and for this reason, we explore here the production of population trends for the nine English Government Office Regions (GOR) and the four countries that constitute the UK. Indices of relative abundance could be produced for five mammal species (Brown Hare, Rabbit, Grey Squirrel, Red Fox and Roe Deer) for two or more regions. Additionally, data were sufficient to produce trends for Red Deer in Scotland and for Fallow Deer and Reeves’s Muntjac in England. It is recommended that in the future population trends be produced at the GOR and country level where data permit.

4. Population trends are produced for government Environmental Zones for the most commonly sighted species. Environmental Zones are categories of landscapes found in the UK from the lowlands of the south and east, to the uplands and mountains of the north and west. The resolution of these analyses is at the 1 km square level, and hence this approach is comparable with other mammal surveys associated with the Tracking Mammals Partnership, such as the BTO/MS Winter Mammal Monitoring (Noble et al. 2002).

5. There are six mammal species (Badger, Mole, Hedgehog, Brown Rat, Stoat and Weasel) for which there were insufficient count data to produce indices of abundance, but for which observers collected a large amount of information on presence/absence from field signs, dead animals or local knowledge. These data were used to examine their change in presence/absence on BBS squares over time. As discussed in Newson & Noble (2003), interpreting the data from the first few years may be difficult because they may reflect increasing awareness by the observer of the presence of a particular species. With existing data, it is not possible to assess the significance of this potential bias. However, since 2002 observers have recorded the criteria that they used for reporting presence (live animals, field signs, dead animals, local knowledge of presence from that season or live animals seen on additional visits), which should aid interpretation in the future. We present information on the change in presence on BBS squares of these six species from 1996 to 2002 and discuss reasons why caution is needed in interpreting these trends.

6. We explore the potential of geostatistics for examining finer scale spatial patterns in relative abundance than is possible through the production of regional indices or visually through the production of distribution maps of species presence. Geostatistical methods are based on statistical models that model autocorrelation (statistical relationship among measured points). Using Brown Hare as an example and the geostatistical method of co-kriging, we explore the extent to which CEH landcover data improves the model fit and hence prediction of relative abundance for 1995 and 2002. In this example it was found that including arable habitat as a predictor greatly improved the model fit, which was improved slightly further by including moorland and heath in the model. In a similar way, statistically valid maps of this type could be used to produce maps of presence/absence using indicator co-kriging. Analyses of this type are at present time-consuming for the analyst as well as computationally, so it is not suggested that interpolated maps of this type are produced routinely. However, the results for this species are encouraging and demonstrate the potential of this methodology for the future.

7. Data for a large proportion of mammal species recorded by the BBS are insufficient to calculate robust indices of relative abundance or occurrence. However, these data still provide important information on the distribution of species, which in many cases are not properly monitored by any existing scheme. For most of these species, it would not be useful to produce annual maps of distribution, but distribution maps of species presence over intervals of perhaps five or ten-year blocks might be considered as more data are collected. There is also the potential for combining these data with those from other surveys and perhaps with incidental records through the National Biodiversity Network to provide a better understanding of species distribution and perhaps if temporal data were available, identify changes in distribution over time. Using the geostatistical methods trialed here, one could predict species presence at unsurveyed/unrecorded sites and by controlling for survey/recorder coverage using declustering there is potential for producing unbiased maps of species distribution.

Back to Research Reports 345 onwards