Looks like smell is probably the key, but the exact details aren’t well known:
Continental drift has been cited as to the egg-laying habits of sea turtles.
Theory is, back when the Atlantic was just a narrow strait, turtles lived in it and just hopped up on the local beaches to lay their eggs. Over the millenia, as the ocean grew, the turtles kept returning to the same beaches, even though they were getting further and further from where they normally lived.
No problem - although I should mention that my second cite has a silly name, but was written by Milton Love, one of the most respected ichthyologists on the US west coast, also author of a number of ichthyology texts, such as this one, which I still use from time to time. He’s got a good sense of humor, and it’s easier to find things in his silly book than in many of my reference books.
Sorry, I should have answered this in my last post: there’s not a lot of information out there on how far out to sea you can continue to find various types of salmon. Fish distribution is usually reported in guides as to how far north or south along the coast you will find the fish, and sometimes to what depth or below what depth it will be found. My guess is that salmon mostly stay above the continental shelf (varying between 40 to 200 nautical miles offshore depending on where you are), but I really couldn’t say more specifically than that. They’re also found off the coast of Japan, but I suspect the distribution is continuous along the coast north to Alaska rather than continuous across the width of the North Pacific at mid-latitudes.
I think Archie Carr may have speculated about this in one of his books, but it is very unlikely that homing to specific localities goes back anywhere near that long. Most beaches are quite transient things, and may not persist in the same location over geological time. Likewise, this scenario doesn’t explain anything about turtle nesting and homing in the Pacific, which is converging.
My point is that the ability to fly has to evolve through intermediate steps. But what are the intermediate steps to the ability to transmit specific locations to one’s offspring purely through genes? I’m not going to argue that because I can’t imagine them that they don’t exist, but I cannot find any animal that demonstrates the ability.
Developing an ability like that would be analogous to a computer evolving a zip drive without ever having had a floppy drive and without data files or software. An animal would have to receive a map transmission to benefit from it, but how would that trait (the ability to receive map transmissions) arise from the gene pool if it can’t be selected for; if it’s not beneficial yet?
Imagine a species that migrates using general cues, like the “fly north for 3 weeks in fall, fly south for 3 weeks in spring” example above. Such a species already migrates. It is in this kind of species that more elaborate methods of controlling migration can evolve.
And I don’t think it is established that there even is such a thing as “map transmission”, even for species that can migrate to a place they’ve never been. Even if the route chosen for migration is under genetic control, we don’t have any good reason to believe that the birds have a mental map of the migration, just that they are responding to cues that we don’t understand. Even if there is such a map, its probably more of a recipe than a map…turn left at Albequerque, follow the coast till you hit Miami Beach, and you’ll recognize Miami Beach by some list of cues.
And also note that migration doesn’t have to be perfect, nor is it perfect. Even though salmon usually find their home stream, lots of them (~10%) swim up the wrong stream, and without this looseness new streams would never be colonized. You always find birds that migrate too late or too early or to the wrong place. If those birds die because they follow the wrong cues then they are selected out. But if the climate is changing then birds that differ from the norm in a particular direction will be selected for.
I am not aware that any animal has a heritable map to a specific location. Instead, there are programmed directions to travel a certain distance in a certain direction.
Long-distance migration requires both a “map” (to tell you where you are, and where your destination is) and a “compass,” to tell you what direction to travel in. The “map” sense is understood far less than the “compass” sense.
Lemur866’s post gives a good description of the process by which migration routes may evolve and change over time.