Dödlighet hos vuxna älgar: en analys baserat på GPS-märkta älgar från Kronobergs län i söder till Norrbottens län i norr

In this work, we analyzed cause-specific mortality of adult cow and bull moose (Alces alces) using a comprehensive dataset of 1021 GPS-marked moose (796 cows, 225 bulls) geographically distributed between Kronoberg and Norrbotten counties during the period 2003 - 2024. This is the first time different causes of mortality have been estimated with such a large individual-based empirical dataset covering different areas along an environmental gradient in Sweden. Our work shows four main results: (1) Hunting accounted by far for the largest proportion of mortality over time even for GPS-marked moose and the risk of being shot was relatively higher for bulls than for cows. (2) Cows and bulls differed in their risk of dying due to hunting, but also for other causes where cows tended to have a higher risk of being killed in traffic or dying due to unknown causes than bulls. (3) The proportion of non-human-related mortality (i.e. not hunting or traffic) varied between areas. Natural mortality tended to be more prevalent in southern study areas with the greatest variation in hemiboreal landscapes south of the Limes norrlandicus. In addition, the unknown mortality tended to constitute a larger proportion on Öland (Kalmar County) and at Östermalma (Södermanland County), and for some of the study areas in Norrbotten. (4) Our analysis of mortality other than hunting, traffic and predation during the first year we could follow a given GPS-marked moose resulted in a mortality of 3.2% (confidence interval: 2.1-4.4%). Our data contained a larger sample of cows than bulls and some causes of mortality occurred rarely. This skewed distribution led to that the cause-specific mortality of cows was given more weight in the results of competing causes of death. The risk of a given cause of death varied greatly for both cows and bulls between areas, especially for rare causes. This resulted in large and overlapping confidence intervals in many cases which in turn restricted the possibility to find distinct statistical differences.