Ardea
Official journal of the Netherlands Ornithologists' Union

login


[close window] [previous abstract] [next abstract]

Piersma T. & Van Brederode N.E. (1990) The estimation of fat reserves in coastal waders before their departure from northwest Africa in spring. ARDEA 78 (1-2): 221-236
To derive realistic equations for evaluating the fat loads of waders before their departure from NW. Africa in spring, we have analysed samples of waders inadvertedly killed during catching operations in Morocco, Tunisia and Mauritania. We studied the relationship between body and fat mass and structural size variables. Part of the original variation in body mass was attributable to a constant relative water loss between capture and first weighing, and body mass values used in the subsequent analyses were corrected accordingly. In the four species for which large samples were available (Knot, Little Stint, Dunlin and Redshank), linear regressions of fat mass on body mass indicated that 50-60% of the differences in individual body mass is due to differences in the total fat mass. In all four species wing length correlated well with lean mass, suggesting that this dimension is generally a better structural size variable than total head, bill and tarsus plus toe length. Only in Dunlins, bill length correlated best with lean mass. Multiple regressions with fat mass as the dependent variable and body mass and structural size variables as independent variables, showed that only those dimensions which correlated with lean mass, contributed significantly to the explained variance in fat mass in addition to body mass. An alternative regression model in which body mass was estimated from fat mass and structural size variables and then reformulated, did not lead to improved predictive equations. The suggested equations to estimate fat mass in individual Knots, Little Stints, Dunlins and Redshanks allows the estimation of fat mass of the heaviest birds with a accuracy of ca. 30%, but with a much lower accuracy in light birds (all estimated values within 100% from the true value). The question whether the presented predictive equations are more widely applicable is discussed.


[close window] [previous abstract] [next abstract]