Thanks Martin for the extra ideas.
The way I designed the GBS database, it output instantly all the summary statistics for each or all species through history (in any combination or selection
desired, on pressing the one (or few) button/s). Thus, outputting the correct and useful values (A, F, R & W) for the relevant graphs is dead easy.
I also suspect that the number of sites surveyed and observer weeks (fully documented in Figure 1 & Table 1 of The GBS Report) has since then increased significantly.
I have not found an easy source of data to show this. Maybe each COG ABR tells it somewhere but not highlighted in a way that the casual reader would even notice that this is important. I think any reference to interpreting GBS data on raw numbers, without
putting the data fully in context is a huge disservice to the data. That is why I gave years to writing it up. On that basis I suspect that the same style graph
“charted the total number of birds reported”
as nice as a graph might look,
of most species would suggest that most species have increased. That would of course be false. Sure, some species have increased.
I am also unconvinced about inserting a calculated polynomial curve. Not that it matters a lot. What is does is smooth out variations between years. The data
are all we have but they are a sample. The issue is in thoughts about how well the sample represents the real. Variation between years may be real or not. The real population may have actually varied more between adjacent years than less. We have no way of
knowing. Fitting the curve just assumes less. There is zero basis for making that value judgement.
Another thing. I strongly suspect is that in recent years there have been proportionately many more sites on outer fringes of the city than there were in the
early years (with none or very few such sites). This surely would also impact on data suggesting change in this and several other similar habitat-use species, that like the fringes. Even with these aspects, that inflate the extent of change, I am confident
that the increase in this species in suburban habitat is real.
Philip
From: Martin Butterfield [
Sent: Saturday, 31 August, 2019 6:25 AM
To: Philip Veerman
Cc: Don Fletcher; COG List
Subject: Re: [canberrabirds] Grey Butcherbird
I have confirmed that using A values gives a very similar pattern to simply counting summing the number of birds.
However I think Philip, is correct in his second paragraph in saying that a high proportion of the increase is due to the same birds being reported multiple times. I came to that conclusion after
looking at the % of sites at which the species was reported each year (ie F, rather than A),
This shows a much more gradual curve rather than a sharp increase around year 24. Repeating the approach from playing around with A I projected a trend in this series up to year 24 forward for
6 years which suggested a value of F of close to 50% - almost exactly the calculated value.
This it would seem that, rather than the fires as a push factor, some pull factor - possibly flying mincemeat - was making it more attractive for the Butcherbirds to hang about.
As one of my bosses used to say "I'm not often right, but I was surely wrong that time".
On Fri, 30 Aug 2019 at 20:36, Philip Veerman <> wrote:
The method that Martin has used “charted the total
number of birds reported”. That does not allow for that the sample size of observer weeks (which should be the denominator in any such calculation, is variable (although
Martin has chosen to bypass that and it may well be sort of excusably right that that
using A values would give a similar pattern). Earlier COG ABR and The GBS report based everything on data relative to the
appropriate denominator. Actually the A values are given in the COG ABR but only in a way that it is very difficult to find, because they are not given in the text! As was the case years ago! These data can only be found by scrolling through that long summary
table, with no way of indexing it! How dumb…. Too tedious even for the most dedicated. I have had this discussion before.
However note that there is a huge component in the data Martin has used, of repeat observation of the
same birds at the same site on subsequent weeks. In early years the species did not stay around much so there was not much repeat observation of the same birds at the same site. Now they are becoming more resident, so a large part of the increase shown in
the graph of data that Martin has used relates that change. As in nowhere near most of the increase suggested is more birds, I suggest although sadly I don’t have the figures, it is mostly or overwhelmingly due more of repeat weekly counts of the same individuals,
due to increased residency. As I said The GBS Report needs to be redone so these things can be properly explained in context and alternate interpretations needs to be done in a manner that comprehends the data.
The suggestion of role of fire is interesting.
Philip
