Some months ago, I reported with excitement that loons that leave their first territory — usually because they are evicted from it by another loon — have a strong tendency to settle on a second territory with similar water clarity to the one they left. It was a striking finding and one that would have added a fascinating story to ecologists’ limited understanding of how animals choose where to settle once they reach adulthood. That pattern, which initially seemed a robust one, has turned out to be trivial.
Let me explain. When loons are forced to shift from one territory to another (or do so voluntarily), they typically settle on a second territory very close to their first one. Often they move to the lake next door or even settle on a different part of the same lake. What we thought we had found was that those “resettlers” carefully choose to settle on second lakes of similar clarity to their first lakes. But such a pattern is tricky to show conclusively. In order to do so, one has to determine how similar 2nd lakes would have been to 1st lakes if loons had settled at random and compare the actual pattern to random settlement.
The problem we ran into was spatial autocorrelation. Spatial autocorrelation simply means that within a large area, two points that are physically close to each other resemble each other more closely than two points that are farther apart. It turns out that lakes of similar clarity tend to be clustered. That is, if one lake is clear, then its neighboring lake is likely to be clear also. The same is true of murky lakes. So while we thought that loons were purposely choosing to settle on second lakes of similar water clarity to their first ones, they were merely settling on lakes near their first one which happened to be similar in clarity to their first because of proximity. In other words, we have no evidence that choice of lakes is non-random.
It took us a long time to discover the error. In fact, Mike Palmer and I had written up this finding and submitted it for publication before we realized — after running a new analysis requested by a reviewer — that spatial autocorrelation was solely responsible for our result.
Naturally, I am feeling a bit like the female on Muskellunge Lake, photographed recently by Linda Grenzer. A month or so ago, this bird and I were on top of the world — she because she and her mate had just fledged two chicks (a bumper crop for loons) and was in her showy black and white formal attire; I because I had an exciting finding that I could not wait to share with colleagues. Now the luster of the recent past is gone. The female’s chicks have dispersed to fend for themselves, she has lost her gaudy summer plumage, and she is about to resume survival mode during migration and winter. I have lost my thrilling discovery and will have to pick myself up, dust myself off, and attack some other scientific question.
Luckily, the Loon Project supplies an endless string of questions. One that tantalizes me at the moment: do chicks that fledge and leave their natal lakes to forage on other lakes nearby favor those that resemble their natal lakes? Such a result, if it occurs, would dovetail nicely with our finding that the first lake a loon ever settles on as an adult (at 4-7 years of age) tends to be strongly natal-like in size and pH. This result (he adds quickly) is not confounded by spatial autocorrelation, because first lakes tend to be far from natal lakes. At any rate, I shall let you know what I find.