Giacomo Tavecchia, Eduardo Minguez, Ana de León, Maite Louzao, and Daniel Oro. 2008. Living close, doing differently: small-scale asynchrony in demography of two species of seabirds. Ecology 89:77–85.

Appendix A. Verifying capture–recapture assumptions.

Capture–recapture analyses typically began by testing the fit of a general model (Burnham et al. 1987). The goodness-of-fit test is based on a -statistic which is calculated from four contingency tables that compare, at each recapture occasion, i, the fate of individuals according to their past and future history (Pollock et al. 1990, Choquet et al. 2005). Specifically, two sets of contingency tables, named TEST3.SR and TEST3.SM, test for homogeneity in survival processes and two, named TEST2.CT and TEST2.CL, test for homogeneity in recapture processes. These contingency tables help to identify the source of heterogeneity. For each recapture occasion, i, they are built as follow:

TEST3R_i: number of animals "not seen before" (individuals newly marked at i) vs. "seen before" (individuals already marked at i) against number if individuals "seen again" vs. "not seen again".

TEST3SM_i: number of animals "not seen before" (individuals newly marked at i) vs. "seen before" (individuals already marked at i) occasion i, against "when next seen again by capture occasions". Note that this test is conditional on animals captured at least once after marking.

TEST2CT_i: number of animals "seen at i" vs. "seen before i but not seen at i" against "seen at i+1" vs. "seen later".

TEST2CL: number of animals "seen at i" vs. "not seen at i but seen before" against "when next seen again" after i+1. Note that this test is conditional on animals known to be alive at both i and i+1.

The sum of all values is the global goodness-of fit statistic. Choquet et al. (2000) developed additional Z tests designed to detect systematic departures from the homogeneity assumption. The Z statistic is calculated as the signed-square roots of the values from the TEST3.SR and TEST2.CL tables (Choquet et al. 2005). The statistic follows asymptotically a standardised normal distribution (Choquet et al. 2005). A negative and significantly large value of Z of TEST3.SR values indicates a systematic excess of newly marked birds among individuals never seen again. This test is generally referred to as a test for the presence of "transients" (i.e., an excess of individuals seen only at marking). A significantly large value of Z for the TEST2.CT indicates the presence of individuals that have been captured very often.