Producing rigorous population trends

The SRMS sets out to produce population trend information that is rigorous and defensible for a wide range of conservation purposes. It does this on the understanding that many of the long-term studies contributing records to the SRMS annually were not set up with the original aim of producing long-term trend information. This is in no way a criticism of these studies, which were set up with other extremely laudable aims. However, it does mean that it is important to consider quite carefully how these data can be used to produce rigorous trends.

High quality and accurate population trends can be generated from:

  • a comprehensive and consistent survey of an entire area of interest (i.e. the whole of Scotland, a whole Scottish region or a single study area); or
  • a consistent sample of sub-areas (either based on grid cells or defined study areas) that are together representative of the area of interest as a whole.

When using SRMS data to generate such high quality trends, the territories and breeding attempts monitored must not be a biased sample, and the areas surveyed need to be representative of the whole area, or of their respective regions.

Within each study area this means:

  1. All breeding attempts should have an equal likelihood of detection. This is much less likely to hold true where search effort has changed either spatially, or in terms of intensity of searching. If detection-likelihood changes, the extent to which the sample represents the population as a whole can change, and the sample may not produce a reliable index of change.

Some examples:

  • If coverage within a long-term study has expanded to include an additional area (that wasn’t previously monitored) adjacent to the original core area, it would be inappropriate to include records from the adjacent area in producing trends in breeding numbers for the original study area, as by expanding the study area the probability of detecting more pairs is likely to have increased.
  • It may still be possible to include the whole data set when producing trends in productivity, as long as the area covered is still representative and the additional area does not differ markedly in terms of, for example, breeding habitat.
  • If a change in circumstance means that a raptor worker is out of action, gets out far less regularly or is unable to cover the same amount of ground as usual, it would be inappropriate to produce a trend in breeding numbers without taking this change in survey effort into account. At best, failure to take account of the change in survey effort could lead to extra between-year variation in the trend, with the potential to mask more important long-term changes. At worst, a more systematic reduction in survey effort through time could give the false impression of a decrease in breeding numbers that was not real.
  • Changes in effort may not be so important when monitoring changes in productivity but only as long as the particular breeding attempts that fall out of the sample through time show no systematic bias. For example, if a particular raptor worker increasingly focused monitoring effort on sites within the study area that were easily accessible without climbing, and these less accessible sites were also the less successful ones, then the change in effort would result in the false suggestion of a decline in productivity through time.
  1. A lack of record must correspond to a lack of breeding attempt. ‘Negative’ records are just as important as records of breeding. To only report occupied territories/nests does not allow absences to be distinguished from unchecked sites.

Within each region (or nationally) this means:

  1. Studied and unstudied areas must be subject to the same drivers of change (so that change observed in the study areas can be assumed to have occurred more widely). Without this, increases or decreases in breeding numbers or productivity from the sample monitored cannot be assumed to occur elsewhere. In this case the trend from the sample of sites could not be said to be representative of that of the regional or national population.

The challenge…

Data collection under the SRMS has ‘evolved’ from the field studies of individual volunteer raptor workers, some of which were established as part of national surveys of breeding raptors. Hence, unlike other national surveys that are specifically designed to produce rigorous unbiased trends in bird numbers (like the BTO/JNCC/RSPB Breeding Bird Survey), SRMS fieldwork was not designed with production of national and regional trends in breeding numbers or productivity as the objective at the outset.

The SRMS data thus come from a number of discrete study areas for each species (which have been selected by the volunteers themselves, rather than based on some form of rigorous randomised sampling design).

The solution…

There is considerable potential to be able to generate trends at the scale of individual study areas, whole Scottish regions or the whole of Scotland but to be able to produce definitive trends (particularly trends in numbers of breeding pairs) it is critical to understand any variations in coverage and survey effort between years (Roos et al., 2015).

There are three tasks that the SRMG is pursuing to help produce definitive trends for as many species as possible in future:

  • We are working with our existing volunteers to produce study-area specific trends for study areas with consistent coverage and exploring the extent to which these trends are representative for the species in the region(s) spanned by the study area. As study area trends are produced (with the volunteers’ permission) they will be published on the SRMS website.
  • We are strategically developing SRMS survey coverage towards more robust survey design/more representative monitoring for each species, including the setting up of a new initiative ‘Raptor Patch’ to complement existing study-area-based monitoring and encourage new raptor volunteers to get involved. This is being specifically designed to produce rigorous unbiased trends in some of the more common and widespread raptor species which currently receive poorer coverage.
  • We are ensuring that the new online data entry system which is being developed is readily able to capture current survey coverage and effort information, and changes between years, so that this is available in a useable form for producing high quality trends efficiently in future.

What can I do to help?

If you are an existing SRMS volunteer – please work with us to produce robust trends for your study area. When approached by the SRMC, please do your best to recall how your coverage and effort have changed over the course of your study. It may be some time before the SRMC gets around to contacting you as we will have to prioritise which species to focus on first, so please bear with us. Also, when you submit data to the SRMS, please remember that negative returns (i.e. where a site has been checked but it has not been occupied) are just as important as positive ones. It is also really helpful to know when sites genuinely haven’t been checked. Please complete a record for every site/area in your study area, even if it is either unoccupied or you cannot check it for any reason.

If you are interested in becoming a SRMS volunteer – please consider getting involved in our new Raptor Monitoring Plots to help us generate the important information we need to help monitor the health of raptor populations across Scotland and to provide data to benefit raptor conservation.