Home > Contents > Methodology
> The Alert System
The Alert System
The alerting system used
within this website is designed to draw attention to developing population declines
that may be of conservation concern. It also identifies situations where long-term
declines have been reversed leading to an improvement in the conservation status
of the species concerned. It must be stressed that the changes reported here are
advisory and do not represent a revision of agreed conservation listings (e.g.
JNCC's Birds of Conservation Importance list (JNCC
1996) or the NGO Birds of Conservation Concern List (Gibbons
et al. 1996)). However, they are based on similar criteria to the
formal lists so they provide a good indication of likely changes at future revisions.
The system is based on statistical
analyses of the population trend data for individual species. Alerts seek to identify
rapid declines (>50%) and moderate declines (25%-49%). These declines are measured
over the full length of the available time series, 25 years, 10 years and 5 years.
The conservation emphasis is particularly on the longer time periods but short
term changes help separate those species where the decline is continuing or even
accelerating from those which have declined previously but are now stable.
The alerts presented on
this website are calculated annually using standard, automated procedures. Where
species are at the margin of two categories (e.g. a decline of about 25%) they
may fire alerts in some years but not others.
Data on some species may
be potentially biased due to unrepresentative coverage by monitoring schemes or
imprecise due to small sample sizes. Because these data often provide the only
available information our general approach is to report all the trends that can
be calculated but to clearly flag up deficiencies in the data.
Smoothing population trends
Bird populations show long-term
changes that do not follow simple mathematical trajectories. In addition to such
long-term trends population indices also show annual fluctuations resulting from
a combination of natural population variability and statistical error. We use
smoothing techniques that aim to extract the long-term pattern of population change
without forcing it to follow any particular shape (such as a straight line or
a polynomial curve). These methods remove most of the effects of short-term fluctuations
so that the long-term trend is revealed more clearly.
details available here
Years used for analysis
Once a smoothed population
trend has been calculated change measures are calculated from the ratio of the
smoothed population indices for the two years of interest. Population indices
for the first and last years of a smoothed time series are less reliable than
the others. Therefore we always drop these years before calculating any alerts.
It may seem that that the alerts are therefore less up-to-date than they might
be but the advantage is that fewer false alarms will be generated. It is important
to stress that the final year of data do contribute to the smoothed curve and
that the final point is only dropped after the smoothing has taken place.
The time taken to collate
and analyse bird monitoring data is another factor affecting the years that can
be included in these analyses. Full analyses of all data sets are not usually
available until 12-15 months after the end of a particular breeding season. This
report was prepared in the first half of 2002 when we had analyses of monitoring
data up to 2000. As we drop the final year of the smoothed time series we are
using change measures up to 1999.
Long-term changes for most
of the species included in this report are calculated from Common Birds Census
(CBC) data. The CBC started on farmland in 1962 and on woodland in 1964. However,
the early years of the CBC population indices are strongly influenced by the effects
of the unusually severe winters if 1961/62 and 1962/63, as well as by developments
in methodology (Marchant et al.
1990). Therefore CBC indices have been calculated using the data from
1966 onwards and population changes are calculated back to 1968.
Data for other schemes generally
start as soon as the scheme had reached a sufficient size to produce reliable
results. The maximum time periods available from the main schemes contributing
to this website are set out in the following table.
The Breeding Bird Survey
started in 1994 and has not been running for long enough for it to be worthwhile
to apply formal alerts methodology. Six year changes based on annual indices are
reported here but we do not flag formal alerts.
Confidence limits and statistical testing
We show 90% confidence limits
for population change measures wherever possible. Any decline where the confidence
limits do not overlap zero (no change) is regarded as statistically significant
and will be used to trigger an alert if it is of sufficient magnitude. Note that
because we are only seeking to detect declines we are using a one-tailed test
with a P value of 0.05. Therefore these confidence limits should not be use
to determine whether increases are statistically significant.
The graphs of population
trends show 85% confidence limits because these allow an approximate visual test
of whether the difference between the indices for any two given years is statistically
significant. These 85% confidence intervals provide us with an heuristic test
for population change: if the indices for two given years are assumed independent
and normally distributed with standard errors of comparable size (standard errors
differing by a factor of up to about 2 are quite acceptable), then to a good approximation
the difference between the indices is significant at the 5% level if there is
no overlap in their 85% confidence intervals (Buckland
et al. 1992). This test is fairly robust, and the independence
assumption is reasonable if the years are some distance apart.
details available here
Data deficient species
There is uncertainty about
the reliability of the results for some species, either because data may be unrepresentative
or because that are based on a very small sample of plots. In these cases the
cause of the uncertainty is recorded in the comment column of the population change
This assessment is based
on the criteria developed by Gibbons
et al. (1993). Data from the New Breeding Atlas were used to compare
the average abundance of a given species in 10-km squares with and without CBC
plots. If average abundance is higher in squares without CBC plots it is likely
that much of the population is not well sampled by the CBC. CBC data for such
species are labeled as "unrepresentative". Where there are insufficient
data to undertake such calculations expert opinion is used.
Sample size is assessed
from the average number of plots contributing to the population indices for a
given species in each year. A plot with a zero count would be included provided
that the species had been recorded there in at least one year and that records
for that plot were available for at least two years. Plots where a species has
never been recorded do not enter the index calculations. These average sample
sizes are shown in column four (plots) of the population change tables. For CBC,
WBS and CES a mean of less than 20 plots is flagged as a small sample. For BBS
a mean of less than 50 plots is flagged as a small sample.
Alerts are flagged in two
categories, greater than 50% decline (>50%) and 25%-49% decline (>25%).
The change measures used are calculated from smoothed time series, wherever possible
based on generalized additive models. After smoothing the first and last years
are dropped from the time series as they may be unreliable. Alerts are only flagged
if the estimated population change is significantly different from zero (no change)
based on bootstrapped 90% confidence limits for the population change measure
(a one-tailed test). Where change measures may be unreliable due to unrepresentative
data or small sample sizes the alert is still flagged but the potential problem
Alerts are evaluated over
the maximum length of the time series, 25 years, 10 years and 5 years. The maximum
lengths of the time series used in this report are 31 years for the Common Birds
Census, 24 years for the Waterways Bird Survey, 15 years for the Constant Effort
Sites Scheme and 75 years for the Heronries Census.
Application to individual schemes
Currently the full methodology
outlined above is applied to results from the Common Birds Census and the Waterways
Bird Survey. For the Constant Effort Sites scheme and the Heronries census we
present annual indices with confidence limits and the fit a smoothed curve through
the annual index values. We do not currently have confidence limits for this smoothed
curve. Therefore all alert labels for CES are shown in square brackets. There
are no alerts for Grey Heron.
details available here
Breeding Bird Survey
The breeding survey started
in 1994 so only six years of data (1994-2000) were available for this report.
This is not a long enough time series to apply the smoothing methods and alerts
framework outlined above. Therefore we have simply calculated change measures
between the first and last years of the BBS time series based on the standard
sites x years model that is used to produce the BBS indices each year.
Next - Statistical methods used for alerts
to Methodology Index