Accurately Counting Grackle-Starling Mixed Flocks (1 Viewer)

[UnknownSpecies56]

Hey folks,

I know that common grackles and starlings frequently intermingle is large flocks, so are there any good guidelines to follow in attaining accurate counts of each bird in large flocks or do you just have to resort to visual estimations? For example, is there any literature that shows that an average percentage of grackle flocks are European Starlings and vice versa? For example, if you take a long distance photograph of a flock that can reasonably be determined to be a mostly grackles based on shapes and the vocalizations heard during the observation, can you assume that a percentage could be starlings?

I can demonstrate a more real world example in the photos below. Pic1 shows a wide angle view of about 178 birds perched in some trees about 250 meters from the camera. I think this is mostly a grackle flock based first on the shapes, and second on the sounds heard during the observation. However, when I zoomed in on a branch with tree birds, I saw 2 grackles and 1 starling. I didn't have time to zoom in on every area of the flock before they flew off, so assuming the birds are uniformly distributed, maybe I can zoom in on a sample area and estimate the percentage of each bird in the flock. In the case of pic2, the ratio might be 2:1, grackles to starlings.

Or, just maybe some wishful thinking here, the literature shows common percentages to go by. This would be useful when you can't zoom in on a sample size, for example when a wide angle photo is taken of flock in the air maybe .5 miles away, which is the maximum discernable range of my camera.

Does anyone have some helpful advice for this problem?

Thanks

 

[UnknownSpecies56]

Here are the pictures.

 

[tom baxter]

Unfortunately the honest answer is that it is scientifically ill advised to ever assume something along the lines of what you are suggesting. Generally, you should not ever record a species that you "think" was present without solid confirmation. Starlings and Grackles are not the most touchy species to estimate counts on or to assume presence, but it is never a good field practice to get into the habit of for any reason. Just my two cents.

 

[D Halas]

And if you do know for sure that grackles and starlings are both present, a sample of three birds is of course not nearly large enough to allow you to extrapolate numbers to the rest of the flock. You'd need more birds in each sample (10, say) and also take several samples from different parts of the tree, since you really can't assume that each species is uniformly distributed throughout the flock.

 

[dantheman]

Agree with the others posting. If you have no real way of telling then you would just have to call it a 'mixed flock of x and y'.

But if you suspect it is almost all one species then call it as that species, and add up the odd individuals of the other species as extras.

If you can actually identify the various species involved then you can start to go on to roughly 50:50 or 80:20 or similar.


For eg mixed gull or duck flocks where I have time and all are id-able but one species predominates I personally tend to count the whole flock as quickly as possible, then go back counting the less frequent species, subtracting from the total which then gives me a number for the commoner one.

 

[UnknownSpecies56]

Thanks for the good advice.

 

[UnknownSpecies56]

I went back and manually picked out and circled (in yellow) the individuals in the 178 bird flock (i.e. pic1) that looked like European Starlings to me. It came out to about 80 birds, which is almost half of the flock. Does this make sense that half of a large flock like this would be a different bird? It seems to me that a half-and-half flock this size would separate, as I have seen 100 bird flocks of starlings isolated. I based the starling silhouette ID mainly on short tail length, smaller body size, and a slightly steeper forehead compared to grackles.

 

[tom baxter]

I went back and manually picked out and circled (in yellow) the individuals in the 178 bird flock (i.e. pic1) that looked like European Starlings to me. It came out to about 80 birds, which is almost half of the flock. Does this make sense that half of a large flock like this would be a different bird? It seems to me that a half-and-half flock this size would separate, as I have seen 100 bird flocks of starlings isolated. I based the starling silhouette ID mainly on short tail length, smaller body size, and a slightly steeper forehead compared to grackles.

When it comes to the behavior and distribution of birds, it is too complicated to approach things with this mentality. To answer your question, I would not expect to see a large flock of starlings and grackles at a 50:50 or 60:40 ratio. This however does not mean it could not happen. If i interpret you correctly, you are suggesting that you have noticed that large flocks of starlings tend to exclude other species, I would tend to agree with that. Also I do not believe that there are anywhere near 80 starlings in your photo, but again it is best to leave birds or any wildlife for that matter as unidentified when evidence is vague. These photos do not show enough detail to clearly identify all of the birds individually without cropping or otherwise enhancing the photo, and even then many individuals would surely be suspect ID's due to the distance and resolution of the original photo.

 

[UnknownSpecies56]

When it comes to the behavior and distribution of birds, it is too complicated to approach things with this mentality. To answer your question, I would not expect to see a large flock of starlings and grackles at a 50:50 or 60:40 ratio. This however does not mean it could not happen. If i interpret you correctly, you are suggesting that you have noticed that large flocks of starlings tend to exclude other species, I would tend to agree with that. Also I do not believe that there are anywhere near 80 starlings in your photo, but again it is best to leave birds or any wildlife for that matter as unidentified when evidence is vague. These photos do not show enough detail to clearly identify all of the birds individually without cropping or otherwise enhancing the photo, and even then many individuals would surely be suspect ID's due to the distance and resolution of the original photo.

That's exactly why I wanted to get another opinion. It is also hard for me to expect a ratio like this. So either I need to go back and recount or just accept that this flock of birds may not be identifiable. I always like to exhaust every effort before I come to that point, but yes when evidence is this vague it may be best to drop it. I don't like this but that's the way it is. Can you tell if any of the circled birds may not be starlings?

 

[UnknownSpecies56]

Let me ask this question please. Would you or anyone else confidently agree that there are at least 98, or rounding up about 100 grackles based on this evidence, vague as it may be? I'm trying to report findings to eBird, but I know reporting nothing is better than reporting bade results. I'm fairly confident that there are at least 100 grackles in this photo, based on the long tails, size, head shapes, and the sound I heard coming from the flock.

 

[nartreb]

Looking at the birds you've circled, I agree that almost all of them are likely starlings rather than grackles (based primarily on tail length). And I agree that most of the un-circled birds are more likely grackles than starlings (on the same basis). There are a few exceptions towards the bottom of the photo - I think are some uncircled starlings there. In round numbers, you have something like 80 likely starlings out of a flock of nearly 200. Unfortunately I have no way of testing how accurate my guesses are. The fact that my guesses agree with yours is somewhat comforting, but could just mean that we're both suffering from the same lack of detail in the photo. On the plus side, after I did my count I went back and looked at your close-up photo; I think I can ID that branch in the big photo, and all three birds match my guesses.

Personally I'd be comfortable reporting this as "large mixed flock, at least 100 grackles with estimated 80 starlings, total ~180 birds."

But I wouldn't ever try to apply this ratio to a future flock or even a region within a flock.

 

[dantheman]

I'm thinking your starting point is inherently flawed - counting and id-ing birds from photos can be either difficult or impossible!

Birds seen in real life give off a lot more clues as to their identity - as they move slightly you get a better idea of shape and form and hence identity. Your own eyes (using optics) may also be better than the camera lens.

I'd concentrate on going through flocks if you encounter them and trying to work out what is within them in real time. Practise counting in eg tens, looking for whether the 'other' species tend to congregate on the edge of the flock etc. Ebird won't suffer if you don't manage to get a result.

 

[UnknownSpecies56]

Thanks for the advice. I'm going to make a mental note about turning the video-camera on in the future and capturing a slow panning clip of large flocks if possible. That way I can go through the video, pause it, take screen-shots, or whatever is needed. Like you said dan, if the quality poor, at least I might get a better sense of the 3-D shape of the animal.

 

[UnknownSpecies56]

I went back an added the enhanced frame showing more clearly the grackles versus the starlings. In this frame there are 30 distinguishable birds, 9 starlings and 21 grackles. That's a 3:7 ratio or 30% starlings in the sample. I guess to get scientific about this, I would need a few more enhanced sample areas like this inset, get a percentage of starlings in those, and average the results. But since I don't have anymore, I'm just going to have to go with what's in the wide angle picture. If the wide angle tallies are close to correct, the percentage of starlings on the whole is about 42%, or 75/178 birds; or slightly higher with earlier counts 80/178 (about 45%). Of course a margin of error could be calculated.

I think it would be safe to say that there are at least 50 starlings in this flock, or 30% on the whole.

Does anyone agree with this?

 

[UnknownSpecies56]

I went back and looked at the original counts from the wide angle shot in the sample area. I counted 18 birds total (12 less) in the wide angle compared to the enhanced sample; that's a 40% error, but I guess to be expected given the difference in resolutions of the birds. Of the 18 birds originally counted, 5 were tagged as starlings so that gives a percentage of 5/18, or 27.78%. This is close to the 30% starling figure in the enhanced sample. However, if I take off the misidentified grackle I get 4/18 or 22.22% starlings. So to be conservative, I'm thinking the percentage of starling on the whole is somewhere between 22-30%. Of course this is based on just one sample and margins of error are significant.

 

[Birdbrain22]

Hmm... I wonder how many BH Cowbirds might be in there?

 

[UnknownSpecies56]

Good question. Do cowbirds makes vocalizations similar to grackles, because judging from this based on my limited knowledge, the flock seemed to be mostly grackles. Can you tell from the enhanced sample whether there are any cowbirds?