estimating recruitment instead of survival with known fate

[sixtystrat]

We are conducting a radio telemetry study on elk in a National Park and want to estimate reproductive rates. We monitored radiocollared cows and then collared calves soon after birth. We also have a number of cows that are not radiocollared but are otherwise known to Park officials. We get frequent (weekly) visuals on these cows and have been able to monitor their reproductive performance as well. Some of the cows (collared and otherwise) are never seen with calves and it is often unknown if they were not pregnant or if the calf died soon after birth.

Rather than estimating pregnancy rate (unknown for some individuals) and then calf survival to estimate the per capita annual recruitment rate, we estimated the recruitment rate directly using a known fate analysis. To code the data, we created a capture history for each potential mother elk. If the cow was never seen with a calf, we coded the capture history as a mortality (i.e., calf mortality) on the first day of the calving season. For the rest of the cows, we created weekly observation histories and recorded a mortality when the calves went missing or if we found the calf dead. The known fate analysis then gave us, not the cow survival rate, but the weekly per capita recruitment rates for both collared and frequently seen uncollared cows.

Does this approach seem reasonable to bypass the assumption that the pregnancy rate is known? A reviewer has questioned this approach and I have not been able to find a literature citation that supports it. Thanks for any feedback!
Joe