Hummingbirds

[jjroper]

Hello all,

I have been catching hummingbirds since 2004 in southern Brazil. To date, we have around 2000 individuals captured, in a total of nearly 3000 captures, of 12 species. Captures are monthly, but not all species are captured in all months.

Seasonality is a fundamental issue here. Some individuals of 2-3 species may be resident during the breeding season. Most are all transients, and so the two migration times each year (altitudinal migration, apparently) are the times when most captures occur.

I tried two ways to organize the data in MARK. In one, I considered each species as GROUPS. The other way, I separated each species into its own file and analyzed each separately.

The best model in the first method was that species were different, but captures were constant (phi different, p the same). In this output, I get a survival rate for each species and its confidence interval. This is using 7 species with > 80 captures.

When I separate each species out into its own file, and then run it individually, p is not always different among years, and phi is not the same as it was in the joint analysis.

My question is which is the best way to go about doing this kind of analysis to estimate survival and compare species?

Cheers,

Jim

[darryl]

If you truly want to compare species then easiest to do so in single data file. It's not surprising you don't get consistent results because when you are doing everything at once, whenever you don't include a species effect you are essentially pooling data across species. You're data also sounds very sparse (average # of recaptures = 0.5), so I suspect some of your "best" models will end up being relatively simple due to small samples sizes; some people tend to forget that AIC also suffers from low 'power' (for wont of a better word) to identify effects with small samples.

[cooch]

Hello all,

I have been catching hummingbirds since 2004 in southern Brazil. To date, we have around 2000 individuals captured, in a total of nearly 3000 captures, of 12 species. Captures are monthly, but not all species are captured in all months.

Seasonality is a fundamental issue here. Some individuals of 2-3 species may be resident during the breeding season. Most are all transients, and so the two migration times each year (altitudinal migration, apparently) are the times when most captures occur.

I tried two ways to organize the data in MARK. In one, I considered each species as GROUPS. The other way, I separated each species into its own file and analyzed each separately.

The best model in the first method was that species were different, but captures were constant (phi different, p the same). In this output, I get a survival rate for each species and its confidence interval. This is using 7 species with > 80 captures.

When I separate each species out into its own file, and then run it individually, p is not always different among years, and phi is not the same as it was in the joint analysis.

My question is which is the best way to go about doing this kind of analysis to estimate survival and compare species?

Cheers,

Jim

I'll post a note that Ken Pollock sent to Jim (which he copied me on).

This is a very interesting question and one I have thought quite a bit about. One of my student explored it in a related context and I have attached the paper in case you are interested.

-> Alldredge, MW, KH Pollock, TR Simons and SA Shriner. (2007) Multiple-species analysis of point count data: a more parsimonious modelling framework. J Anim Ecol 44: 281-290

If you have species with lots of recaptures I would focus on the separate species analyses first as then you dont have to worry about bias induced by an inadequate model for the multiple species.

If you have rare species then joint species analyses will be essential but I would be looking at additive models . Not sure if you are familiar with those models which require use of the design matrix?

You may also have assumption violations used in the single species analyses. there might be a need to use different periods for different species?