Lars Hillström

Biodiversity in different ecosystems in the Mediterranean region

This is a collaborative research project with the University of Cordoba, Spain, which is one of our partners in the Erasmus exchange of teachers and students. Part of this project is to study the factors that limit the distribution and abundance of wild rabbits in the Sierra Morena mountain range, Andalusia. The wild rabbit is a very important species in the Mediterranean ecosystem, but it has declined sharply due to two different viral diseases in recent decades. The project aims to study what is limiting different populations of the wild rabbit in Spain. An interesting question is: how can some of the populations become immune to these viral diseases, while other populations are very severely affected by these viral diseases?

Heavy metal exposure in some endangered ducks

This is a collaborative research project with the University of Aberdeen, Scotland and the Doñana Research Institute, Seville, Spain. Many species of ducks have been exposed to a variety of heavy metals in different wetland areas, including lead, cadmium, mercury and other more harmless metals such as zinc and copper. We studied whether it is possible to trace the origin of lead poisoning in the marbled duck and the European copper duck, which also suffers from the hybridization of the introduced American copper duck with the European copper duck. It proved possible to trace the source of lead in the birds, with most of the lead coming from old shotgun pellets buried in the wetlands. Geese, among others, ingest large quantities of these hailstones, believing that they are small stones that help the birds to digest their food. To study the source of origin, we have been studying the presence of lead isotopes in different tissues of marbled and common scoters.

To read more about this project on heavy metals in ducks, see the following articles:

"Metal levels in the bones and livers of globally threatened marbled teal and white-headed duck from El Hondo, Spain External link."

Lead isotopes and lead shot ingestion in the globally threatened marbled teal (Marmaronetta angustirostris) and white-headed duck (Oxyura leucocephala). External link.

Sibling rivalry and intra-family conflict

Studying intra-family conflicts in animal organisms is actually a major field of research, as these conflicts are present in most animal groups. Perhaps the most visible conflict within a family is that between siblings, but there is also potential conflict between the sexes/parents and other conflicts within the family are at least potentially possible (see below). Sibling rivalry can take different forms in different animal groups. In birds, for example, it is common for the young in the nest to compete for the most advantageous place when the parents return to feed the young, but in birds of prey, for example, sibling rivalry can take other forms, such as more or less fierce fights that can sometimes end fatally (see below).

Another obvious conflict is that between parents and offspring. This may concern, for example, how much energy the parents want to invest in their litter or in a particular offspring in the sibling group. It is almost always the case that the offspring want more food (or more shelter, for a longer period of time) than the parents are interested in giving them. Therefore, it can sometimes be observed in nature how, for example, the parents of chicks more or less try to push them away when they have grown big enough, this is known as parent-offspring conflict.
Litter reduction in birds has been assumed to be a "tool" for the parents to make a trade-off in the reproductive effort. However, there are very few studies measuring the long-term effects of litter reduction.

At the University of Gävle, we are working on an interdisciplinary project between biology, mathematics and computer science, with the goal of creating a good model of intra-family conflict in birds of prey. In our group, Anders Johansson has experience in mathematical modeling of biological systems, Fredrik Åhman has done computer simulations of intra-family conflicts, and I myself have studied intra-family conflicts in a few different bird species. These studies include sibling competition in the black-and-white flycatcher (thesis), obligate siblicide in the tropical brown gannet, and facultative siblicide in colony-nesting cowbirds and raptors.

In the Doñana National Park in southern Spain, the influence of different ecological factors on siblicide behavior in birds of prey was studied. The study showed that siblicide behavior is most frequent in areas with higher local densities of brown thrashers, indicating that intra-family conflicts may be influenced by population density. In another study using videotaping of nests of brown cats, it was shown that sometimes conflict behavior was rewarding, while sometimes conflict behavior was most rewarding. Some behaviors within a family can be described as a game, where in this case the siblings are the different players.

We can now create ecological models of this system of intra-family conflicts in order to study which strategies may be most beneficial in the long run, i.e. those that constitute an evolutionarily stable strategy. It is important to study the ecological variables that may influence these strategies. An evolutionarily stable strategy (ESS) is robust to alternative strategies that occur with some randomness. Given that each strategy is a phenotype originating from the gene carried by the sibling, it would be theoretically possible to calculate the probability that an individual would choose a particular strategy when a certain condition arises. The condition can be conditioned by the environment or by the opponent in the litter/family in question. By describing a total fitness for all siblings, it is possible to determine the optimal parental investment strategy for each sibling. However, one may not be able to rely on such relatively primitive values as the degree of relatedness between siblings. Sibling relatedness can be more or less overlapping between different genotypes. There are therefore better methods that involve studying the reproductive success of a particular gene in different generations.

One such method is multiagent Q-learning. Reinforcement learning is a paradigm in computer science that involves programming different control systems (agents) so that they behave optimally, i.e. maximize a target function (such as profit or fitness). The method does not assume any a priori knowledge of how the environment and the agent interact, but the agent's policy (choice of action options) is refined based on the experience gained through "experimentation" in a simulated or real environment. A specific implementation of feedback learning is so-called Q-learning, which can also be used in game theoretic situations where multiple agents can interact and/or compete. In such a situation, under rather general conditions, learning will converge to an optimal equilibrium state where none of the agents' policies can be locally improved (Nash equilibrium). From an evolutionary perspective, such an equilibrium state should arise between the individuals of a typical family unit within a species where the objective function is given by the inclusive adaptation number. The point of using feedback learning in the simulations is that the method, which works well with different approximations of state and policy spaces, allows an efficient numerical determination together with specific and realistic models.