Abstract
1. In those countries where lead shot is still in use, a secondary effect of waterfowl hunting is lead poisoning from shot ingested by birds during bottom feeding. Moreover, waterfowl injured during hunting can die undetected as a direct or indirect consequence of wounds. The occurrence and influence of these types of lead exposure have often been estimated by the inspection of dead bodies, but this method will yield biased estimates if dead birds are not a random sample of the population. 2. We analysed a historical set of recoveries of adult mallard Anas platyrynchos ringed in the Camargue, southern France, over the period 1960-71, for which the amount and type of lead exposure had been determined by X-ray inspection before release. It was therefore possible to investigate the prevalence and effect of lead shot, avoiding the problem of post-stratification that may arise when only dead individuals are considered. 3. Among the captured birds, the proportion of gizzard-contaminated birds was constant (0·11) during the study period. In contrast, the proportion of birds carrying pellets in muscles increased linearly from 0·19 to 0·29. Males and females were similarly exposed to shot from both sources. 4. The relative survival of lead-affected mallards was 19% lower than unaffected birds for both types of lead exposure. The two sources of mortality were additive on a logarithmic scale and unaffected by sex. 5. Attempts to estimate the consequence of lead poisoning on population dynamics were not conclusive because of large confidence intervals in survival estimates. Moreover , it was still not clear how much mortality due to lead exposure should be considered as additive to other causes. We nevertheless advocate measures to control the long-term impact of lead exposure on waterfowl populations. 6. This work presents a new approach to the analysis of survival, in which the standard recovery models were reformulated in terms of monthly survival. This allows for the effect of a long ringing period during which mortality cannot be ignored. It also allows the estimation of parameters for newly marked birds without additional information. Since any within-year time interval can be considered, this approach can be used to investigate seasonal changes in mortality.