Sparrowhawks responsible for House Sparrow decline says scientist (1 Viewer)

[Wildmoreway]

The old "Life" program of ZX81 days.

 

[Jane Turner]

I never had the programming skills, but I always wanted a model with a meso-predator too.

 

[Nightranger]

The old "Life" program of ZX81 days.

There was a better version with angelfish and sharks that operated slower and was easy to experiment with but I can't find it online now.

 

[Wildmoreway]

Very simplistic but it still showed the mechanism whereby the shark and fish actually depended upon each other for survival, both prey and victim controlling each others numbers and making sure that only the best victim and prey survived and bred on.

In general the only species stupid enough to completely eliminate another species is Homo sapiens.

 

[Wildmoreway]

In addition, the vet drugs given to cattle mean that cowpats support far fewer invertebrates these days.

Interesting to me in that I read a comment from a fruitarian that she believes that the reason that humans are taller and bigger these days is because of the growth hormones fed and injected into farm animals.

 

[MJB]

Interesting to me in that I read a comment from a fruitarian that she believes that the reason that humans are taller and bigger these days is because of the growth hormones fed and injected into farm animals.

Interesting to me in that I read a comment from a fruitarian that she believes that the reason that humans are taller and bigger these days is because of the growth hormones fed and injected into farm animals.

I may be reading too much into your brief post, but do you know if your fruitarian has also taken into account the masses of existing research evidence? I always like to know the basis f any expressed belief...

For example, an enormous amount of research has been done in, for example, the USA (prior to mass growth hormone feeding of cattle), consistently showing very strong correlation between improved diet and increasing height and body mass (of fit, not obese people).

I think it's reasonable to have a concern about the amounts of growth hormone used in farming - do the hormones break down into harmless, over the long term, byproducts? My guess is that if they don't then some correlation between obesity and their presence in humans is perhaps more likely than a simple size increase. I base that guess on hormones used in cattle being less likely to have identical effects on other mammals than to have a modified effect, which is the kind of findings that drug companies have when testing protoype drugs that may have value for humans on animals, morally repugnant though that might be to many.
MJB

 

[Wildmoreway]

It was only a comment that I read somewhere. The basic problem with Homo sapiens is that there are simply far too many of us.

 

[CPBell]

Another tempo-spatial data set available here

If we’re going to go through all the counties one by one, your first going to have to explain why the rebuttal I posted here doesn’t apply to your latest example – and please also, can we have the information needed to interpret the maps properly?

So let's give it a try...

- How many results are included in in each zone/site type classification? (and what is the probability of misclassification)

- How precisely did you determine the classification with each site?

- Which are the results you excluded in terms of zone and type.

The answers will stay the same, no matter how many times you ask the questions.

(i) You have the maps showing distribution of sample sites in relation to zone/rurality, and this confirms some contingencies, notably urban/zone 1 & 2, have sample sites in single figures. Your implication that I’m trying to hide something because I can’t provide the exact figures is groundless. The fact is that I did my data sorting with a free trial download of MS Access which has since run out, so I can’t check the exact figures. We amateurs have to make do and mend, you know. What is the probability of misclassification? I haven’t the faintest idea.

(ii) The urban/rural classification was done by centring the old OS ‘Get-a-map’ facility on the site grid reference and counting the proportion of the surrounding 16 grid squares that were more than half built up. The proposition that we should have measured the proportions using a romer dot grid or a planimeter, and repeated this for ordnance survey maps published at intervals during the study period to take account of development, is preposterous and revealingly anal, since it involves increasing the labour involved by several orders of magnitude for an uncertain gain in relevant accuracy.

(iii) This presumably refers to the 5-6 data points out of 797 used in the repeat of the Thomson method in the Auk paper – again, I haven’t the faintest idea.

So now explain why this invalidates the analysis.

It’s a shame this thread has largely descended into a pi**ing match, since the underlying subject matter is very interesting.

I live on Jersey. Sparrowhawks became extinct as a breeding species in the 1950s, and for the next few decades were essentially a vagrant – I might see 1-2 per year in the early 1980s. During this period of local extinction, House Sparrow populations declined markedly.

In the last decade or so, Sparrowhawks have firmly re-established themselves as a breeding species, and are probably at carrying capacity. For the last 3-4 years, House Sparrow numbers do seem to be on the rise.

If I was a mischievous sort, I might suggest that there may be a causal link in this positive correlation!

For the sake of balance, I should also say that Cirl Buntings, Reed Buntings, Yellowhammers and Stonechats all died-out (or virtually so) as breeding species during the last decade, and I am open-minded about the possibility that Sparrowhawks were at least partly responsible.

And this is the most interesting post for about 18 months. From the beginning I’ve encouraged people to report examples where there is an apparent discrepancy between sparrow decline and sparrowhawk colonization at various locations. A number have been posted (e.g. Finland, Spain), but on closer examination they prove to be no such thing, so you have to be careful.

A quick internet search throws up this, which indicates that sparrow decline on Jersey was late – on a similar timescale to that in central London – i.e. in the late 1990s when the general trajectory for Britain as a whole was levelling out. Is there any reliable information on the precise timescale for Sparrowhawk re-colonisation of Jersey? It should be available in the Jersey Bird Report – can you check?

http://www.cpbell.co.uk
http://www.youtube.com/CultoftheAmateur

 

[AlfArbuthnot]

(i) You have the maps showing distribution of sample sites in relation to zone/rurality, and this confirms some contingencies, notably urban/zone 1 & 2, have sample sites in single figures.

And this, folks, is the strength of the evidence on which he wants us to reject every other theory regarding sparrow declines, and also reject the work of the BTO and RSPB, and have us believe they're lying to us for the sake of money.

( Your implication that I’m trying to hide something because I can’t provide the exact figures is groundless. The fact is that I did my data sorting with a free trial download of MS Access which has since run out, so I can’t check the exact figures.

Didn't his co-authors at Cambridge University and BTO have a copy of MS Office? Or didn't they have much to do with this paper? Either way, he doesn't even know his own sample sizes ...

What is the probability of misclassification? I haven’t the faintest idea.

..and he doesn't know how reliable they are, or what the effect on his results would be if some of the data were wrong (and he has no idea if that's likely).

Sounding like a solid and rigorous study so far, isn't it? But there's more:

(ii) The urban/rural classification was done by centring the old OS ‘Get-a-map’ facility on the site grid reference and counting the proportion of the surrounding 16 grid squares that were more than half built up.

So the only thing he counted was the squares, BUT the thing he put into his model was the amount of urbanisation - which he didn't count. HE JUST GUESSED!

He says it would be too difficult to count:

The proposition that we should have measured the proportions using a romer dot grid or a planimeter, and repeated this for ordnance survey maps published at intervals during the study period to take account of development, is preposterous and revealingly anal, since it involves increasing the labour involved by several orders of magnitude for an uncertain gain in relevant accuracy.

But any GCSE student could have told himm to take a screen grab, put it into a free graphics/photo package, increase the contrast to 100% and then it will tell you the exact percentage. But notice that his defence is not that it wasn't necessary, BUT THAT IT WAS TOO HARD FOR HIM (even though it's simple)! I imagine it was quite hard to find the Higgs Boson this week, but that doesn't mean that they just guessed it was there instead, and left it at that!

(iii) This presumably refers to the 5-6 data points out of 797 used in the repeat of the Thomson method in the Auk paper – again, I haven’t the faintest idea.

So now explain why this invalidates the analysis.

Maybe because you need a good reason to leave out data, as the data points may be having a significant effect on your study, which you are then doctoring? But this is more reassurance that this was a seat-of-the-pants study - the man who did the maths doesn't have the faintest idea why he left out some of the data!

Just a reminder, folks, that he wants us to reject everything we think we know about why farmland birds have declined on the strength of this study, and he wants us to believe the BTO are part of a conspiracy to hoodwink us.

After reading CPBell's staggering post (which largely confirmed my suspicions about how sloppy his methods were), I wouldn't trust his study to line the bottom of a budgie cage, never mind confirm anything about Sparrowhawks.

Is there any reliable information on the precise timescale for Sparrowhawk re-colonisation of Jersey? It should be available in the Jersey Bird Report – can you check?

Do you mean check peoperly, or follow your lead and have a bit of a guess instead? Checking might be a bit too difficult, you see, perhaps involving some counting, and Bargain Hunt will be on the telly soon...

 

[Jane Turner]

(i) You have the maps showing distribution of sample sites in relation to zone/rurality, and this confirms some contingencies, notably urban/zone 1 & 2, have sample sites in single figures. Your implication that I’m trying to hide something because I can’t provide the exact figures is groundless. The fact is that I did my data sorting with a free trial download of MS Access which has since run out, so I can’t check the exact figures. We amateurs have to make do and mend, you know. What is the probability of misclassification? I haven’t the faintest idea.

(ii) The urban/rural classification was done by centring the old OS ‘Get-a-map’ facility on the site grid reference and counting the proportion of the surrounding 16 grid squares that were more than half built up. The proposition that we should have measured the proportions using a romer dot grid or a planimeter, and repeated this for ordnance survey maps published at intervals during the study period to take account of development, is preposterous and revealingly anal, since it involves increasing the labour involved by several orders of magnitude for an uncertain gain in relevant accuracy.

(iii) This presumably refers to the 5-6 data points out of 797 used in the repeat of the Thomson method in the Auk paper – again, I haven’t the faintest idea.

So now explain why this invalidates the analysis.

Hmm well lets see, tiny data sets, guessed classifications no clue how accurate your guesswork was, no ability to model to identify or attempts to model your (plainly obvious) sources of error and no rationale for deleting the points you didn't like the look of, other than you didn't like the look of them. Were any of them in the "single figure" sets?

 

[CPBell]

Hmm well lets see, tiny data sets, guessed classifications no clue how accurate your guesswork was, no ability to model to identify or attempts to model your (plainly obvious) sources of error and no rationale for deleting the points you didn't like the look of, other than you didn't like the look of them. Were any of them in the "single figure" sets?

This is not an argument - it's just rhetoric. Take an analysis - any analysis in the paper - and explain why the critique implied by your three questions renders invalid the statistical inference leading the conclusions. Specifics please.

http://www.cpbell.co.uk
http://www.youtube.com/CultoftheAmateur

 

[Nightranger]

A quick internet search throws up this, which indicates that sparrow decline on Jersey was late – on a similar timescale to that in central London – i.e. in the late 1990s when the general trajectory for Britain as a whole was levelling out. Is there any reliable information on the precise timescale for Sparrowhawk re-colonisation of Jersey? It should be available in the Jersey Bird Report – can you check?

DMW already overviewed this point in the first post and the thread moved on. Having said that, what you ask for should be available from Jersey Environmental Section:

Contact name Environment Section
Phone number +44 (0) 1534 441600
Fax +44 (0) 1534 441601
Email address [email protected]
Opening hours Monday to Friday 9am - 5pm
Address Department of the Environment
Howard Davis Farm
La Route de la Trinite
Trinity
Jersey
JE3 5JP

In addition, your local library should be able to order Jersey Bird Reports from a central source for loan although I suspect going directly to the Environment Section woiuld be better for getting data as opposed to bird reports, which may or may not have actual figures. I doubt it will prove much because it is a look back in hindsight but go for your life.

 

[AlfArbuthnot]

This is not an argument - it's just rhetoric. Take an analysis - any analysis in the paper - and explain why the critique implied by your three questions renders invalid the statistical inference leading the conclusions. Specifics please.

http://www.cpbell.co.uk
http://www.youtube.com/CultoftheAmateur

For an open-goal start, how about sample sizes in the single figures (e.g. about 6 gardens representing thousands of km of landscape) are inherently unreliable and undermine ANY statistical basis you can dream of. The schoolboy yardstick is a sample size of 30 minimum, if you want stats to mean anything.

Of course, you can use much smaller sample sizes and get a result, but you would then cover that result in plenty of caveats - what you would NOT do is say that it's conclusive.

For seconds, you stated earlier in this thread that it would take "very few indeed" misclassifications to make your statistically significant result vanish. Jane Turner could try and nail down the greased eel once more and ask how many is "very few indeed" ["specifics please!"], but that's pointless because you seem to have thrown away all of your data and don't have the faintest idea. But we can presume that in the context of single figure sample sizes, it means literally 'a couple'. So even if your by-eye guesstimate of classifying gardens was correct 99% of the time (which is a very big 'if'), that still leaves about half a dozen times when it wasn't - which is enough to be "very few indeed".

But all that is rather academic, and so is talking about specifics. Modelling studies are full of doubt and unknowns at the best of times (you don't know that all of the important variables are in your model). This is bad enough, without introducing more doubt of your own because you don't know how to handle a digital map and don't have a spreadsheet package (though why you couldn't have stored your data in notepad as a text file is beyond fathoming, seeing as that would be the format you used in the R modelling program).

Quite how none of this was picked up in review is astonishing.

 

[Nightranger]

And this is the most interesting post for about 18 months. From the beginning I’ve encouraged people to report examples where there is an apparent discrepancy between sparrow decline and sparrowhawk colonization at various locations. A number have been posted (e.g. Finland, Spain), but on closer examination they prove to be no such thing, so you have to be careful.

No Dr Bell, you have encouraged people to post observations that support your conclusions. I and quite a few others have posted observations that clearly show a very active sparrowhawk population and a correspondingly healthy house sparrow population. I cannot be 100% certain until I observe something to the contrary but I doubt the sparrowhawks round here bother going after sparrows because there are so many bigger and easier targets. The only time I have seen a sparrowhawk going after house sparrows was when a hawk surprised the sparrows on the roof of the building where I was staying in Bangor. The sparrows dropped three storeys into a convenient bush and left one angry hawk trying to figure out if it could reach into the bush. Round here, the hawks just drop in on the local feral pigeon population and the occasional young starling just after the latter have fledged.

 

[Robin Edwards]

Nightranger. I said i wouldn't post on this thread any more but here goes.

I have a friend currently allowing me access to his urban garden where a pair of Sparrowhawks are feeding five healthy young. Monitoring so far has shown regular prey being brought in but smallest prey I've yet seen is blackbird/thrush size and mainly pigeon/dove. When the young were newly hatched it is likely that smaller finches and sparrows would have found themselves as prey but certainly not now the young are a good size.

As I've indicated before, if like me, you only have life experiance to offer rather than hard scientific research then this thread is not for you and posts will have little value.

 

[rob lee]

I`ve just worked out what this thread reminds me of now.

http://www.youtube.com/watch?v=zKhEw7nD9C4

 

[Jane Turner]

This is not an argument - it's just rhetoric. Take an analysis - any analysis in the paper - and explain why the critique implied by your three questions renders invalid the statistical inference leading the conclusions. Specifics please.

/URL]

Let's go for a practical demonstration.


The attached example are real results - for fun relating to peak sparrow counts at two different types of feeding station. By your own standards these are headily large double figure data sets. The 1st 11 points from type 1 appear to have a substantial outlier, of the type you excluded. If you add just two more data points,you get a rather different answer.

Now since you were dealing with averaged peak counts, from single figure sites, the relevance of noise and deletion becomes suddenly rather relevant.
You have by your own admission single figure data sets. You have deleted points from the results you used because you treated them as outliers without any analysis as to any they were outliers. With the inherent noise and miniscule sample sizes, that's ridiculously flawed thing to do.


Open office is freely available and should be able open your original files - should youwish to spill the beans on what your sample sizes and deletions really were btw

 

[Allen S. Moore]

I`ve just worked out what this thread reminds me of now.

http://www.youtube.com/watch?v=zKhEw7nD9C4

Good one, Rob!

 

[Nightranger]

I have a friend currently allowing me access to his urban garden where a pair of Sparrowhawks are feeding five healthy young. Monitoring so far has shown regular prey being brought in but smallest prey I've yet seen is blackbird/thrush size and mainly pigeon/dove. When the young were newly hatched it is likely that smaller finches and sparrows would have found themselves as prey but certainly not now the young are a good size.

Thanks Robin. I perhaps should have qualified my local observations by saying that we have sparrowhawks active throughout the year. The gardens around here are more accurately, suburban but would probably classify as urban under CP's study. Interestingly, the main prey taken by the sparrowhawks in Ian Newton's study was the chaffinch but given this was largely in Scottish forests, it is perhaps not so surprising. The hawks rarely took prey from more than half a mile from the nest, which is a fair reflection on how abundant chaffinches must have been. Put all this together and if we are to believe anything CP has said, it is a wonder there was no crash in the Scottish chaffinch population if they were so selectively being taken out.

 

[Jane Turner]

Its worth adding, that if I use the kind of sample sizes that my friendly neighbourhood statistician would recommend, I find that there is no significant difference for the two site types.