Scientific understanding of the biodiversity issues must be based on sufficient knowledge of the ecology and biology of the species and on the understanding of the dynamics of their populations. A modelling approach that is closely related to empirical studies is an invaluable tool to advance these aims. In this task, we develop a range of modelling approaches based on existing models, existing knowledge about the biology of forest-dwelling species, and on the results of Tasks 1 to 3 of this project. The four critical elements that have to be addressed by models are habitat loss, habitat fragmentation, change in habitat quality, and non-equilibrium dynamics.
The aims of this research are to develop the following modelling approaches
- Strategic models, which facilitate the conceptual and theoretical analysis of key questions
- Quantitative population dynamic models, which can be used to generate predictions for the management of forests and forest-dwelling organisms for conservation
- Landscape ecological models, which deal with structural properties of forest landscapes with minimal information on species.
The research group is described in the following table (PI=Principal Investigator, UH=University of Helsinki, FEI=Finnish Environment Institute)
Ilkka Hanski (UH) PI
post doc (UH)
Atte Moilanen (UH)
Paula Siitonen (FEI)
landscape ecological modelling
Scientific understanding of the biodiversity issues must be based on sufficient knowledge of the ecology and biology of the species and on the understanding of the dynamics of their populations. A modelling approach that is closely related to empirical studies is an invaluable tool to advance these aims. Therefore, the purpose of Task 4 is to develop a range of modelling approaches based on existing models, existing knowledge about the biology of forest-dwelling species, and on the results of Tasks 1 to 3 of this project. Our purpose is to develop both strategic models, which facilitate the conceptual and theoretical analysis of the key questions, and quantitative models, which can be used to generate predictions for the management of forests and forest-dwelling organisms in the context of forestry and conservation.
2 Existing modelling approaches
Complex changes in complex landscapes have complex biological consequences. The following four elements, which are all likely to be significant in the biology, conservation and management of boreal forest and forest-dwelling species, have been previously modeled separately.
Habitat loss is undoubtedly the most serious threat to the majority of threatened species (Temple 1986, Whitmore & Sayer 1992, Lawton & May 1994). At present, old-growth forests cover much less than 1% of the forested land in southern and central Finland, and around 15% in northern Finland. The species-area relationship may be used to predict, though very approximately, the fraction of species that will eventually be lost when a certain fraction of the area of the habitat has been lost (Whitmore & Sayer 1992, Simberloff 1992), though the species-area relationship does not suffice to predict when these extinctions will occur (May et al. 1994). Another shortcoming of the species-area relationship as a model of species loss is that it does not specify the actual mechanism leading to extinctions.
Habitat destruction usually involves two components, habitat loss and fragmentation of the remaining suitable habitat. 'Fragmentation effects' refer to the effects of habitat fragment size and isolation on the presence and density of the focal species (Andrén 1994). The biological consequences of habitat fragmentation have been addressed in metapopulation studies (Gilpin & Hanski 1991, Hanski & Gilpin 1997). Dynamic models have been developed to predict the consequences of particular changes in landscape structure (Akcakaya & Ferson 1992, Hanski 1994a, Lindenmayer & Possingham 1994). The incidence function approach, which has been developed in Helsinki (Hanski 1994a, 1994b, 1997), has been found to be a particularly effective modelling approach in the case of highly fragmented habitats (Hanski et al. 1996, Wahlberg et al. 1996, Moilanen et al. 1997). Other modelling studies have usually employed complex simulation models (Akcakaya & Ferson 1992, Lindenmayer & Possingham 1994), in which local dynamics and migration are modeled explicitly.
Change in habitat quality
The source-sink model (Pulliam 1988, 1996) has been developed to analyse the biology and dynamics of species living in heterogeneous landscapes with variation in habitat quality. A major message from these studies is that population density may be a very misleading indicator of habitat quality, because of the possibility that density is greatly affected by migration. Consequently, organisms may often occur in intrinsically unsuitable habitat, and in extreme cases even the majority of individuals within some larger region may occur in such a 'sink' habitat.
A particularly critical element in the modelling of the biological consequences of habitat destruction is the time delay that is likely to occur in the response of populations, making it questionable to use modelling approaches based on the assumption of equilibrium (Hanski et al. 1996). Delayed extinctions due to past habitat destruction have been aptly dubbed as the 'debt of extinctions' (Tilman et al. 1994), due to the temporary survival of species that are, for practical purposes, 'living dead' (Hanski 1997). A key question in the case of managed boreal forests in Finland, which have experienced dramatic changes at the landscape level for the past few hundred years, is which fraction of the surviving species are actually such living dead.
At the level of individual populations that have become completely isolated in the course of habitat fragmentation, we may use standard models of population extinction (Foley 1997 and references therein) to predict the delayed extinctions.
3 Research tasks
At present, no sensible modelling approach exists that would incorporate all the four components described above. The only approach that could claim to include everything is numerical simulation of the dynamics of species in realistic landscapes, characterized by some GIS-based landscape description. But the obvious and great problem with this approach is the great difficulty of justifying the numerous model assumptions, estimating the many model parameters, and testing the model. In this situation, we plan to use the following modelling strategy, which should allow Task 4 to achieve as much as possible:
A. Use the more specific modelling approaches for specific purposes
It seems useful to develop all the four components described above to answer specific questions and to provide partial answers to more comprehensive questions. For instance, the incidence function approach (Hanski 1994a, 1994b) can be applied in a rather straightforward manner to model the dynamics of species living in discrete microhabitats, such as fungi, mosses and insects associated with decaying tree trunks (Section 4.1). The dynamics of species living in networks of old-growth fragments surrounded by an inhospitable matrix (Sections 4.2 and 4.3) can also be modeled with the incidence function approach.
We are ready to apply the existing modelling approaches, especially the incidence function model, to existing and new data. Therefore, some modelling results will be available very soon, and the dialogue between modelling and empirical studies in this project will be established from the very beginning.
B. Develop a more comprehensive simulation approach but keep it as simple as possible
It seems inevitable that a modelling framework based on brute force numerical simulation is needed, in spite of its problems, to model the dynamics of species in landscapes with varying degree of habitat loss and fragmentation. Such models could also be used to make predictions about non-equilibrium dynamics. The big challenge here is to keep the model so simple that it can be sufficiently justified, parameterized, and tested. We expect that this modelling framework is functional in 2-3 years, though it will be further developed during the duration of this project.
C. Develop new strategic modelling approaches
The scientifically most challenging and ambitious task is to develop modelling approaches that lie between the two extremes described above: approaches that would include two or more of the components described above but without entirely resorting to brute force numerical simulation. What this approach will consist of is presently not known. This part of the project aims at new scientific discoveries and insights.
D. Landscape ecological models
Paula Siitonen (unpubl.) has been developing a tool for landscape ecological planning of forest regions which is largely based on the physical properties of fragmented landscapes, especially the sizes and spatial locations of different types of habitat fragments, but also including such key microhabitat elements as the amount of decaying wood in the forest. She will continue to develop this approach in this project as a part of her Ph.D. thesis. Her study includes modelling of fragmentation history and species dispersal (see Task 2). Her approach is complementary to the approach described above, which is based on modelling of the dynamics of focal forest-dwelling species. Below is a short description of the landscape ecological modelling approach.
The purpose is to develop a simple method for landscape ecological planning. The method is a GIS-based tool, which uses forestry inventory data, satellite images and field inventories. The method can be used to select the stepping stones, movement corridors and core areas for the landscape ecological forest plan. Alternative landscape ecological plans can be compared by this method. For the selection of stepping stones, we need to find out which habitat patches best complement an existing old-growth forest network and increase spatial and temporal connectivity of fragmented habitats. Optimality in patch selection means finding a minimum set of sites that includes maximum efficiency of representation in terms of the number or area of sites. The forest patches that increase spatial connectivity are important because they facilitate the dispersal and movement of species in fragmented landscapes.
The method can be integrated into the GIS-based forestry planning system of the Finnish Forest and Park Service or other forest planning systems. The approach here is linked to the practical tool for the assessment of the forest biodiversity which Paula Siitonen has developed in 1993-1996 for the FPS and the Forestry Development Centre Tapio. This tool has been used in years 1995-1996 as a part of the landscape ecological planning in ca 35 planning units (10,000-60,000 hectares in sizes) of state-owned and private forests.