I was 11 years old when an event occurred to which I can trace my teenage wish to become a biologist. The date was August 18, 1964. I had the childhood luxury of spending all my summers in the countryside, where, following the example of some older boys, I had taken up collecting insects. On that day I caught a brown butterfly that I could not name with the help of my small handbook. Looking through a more advanced work which my parents had bought me in the autumn, I found that my mystery insect was a female of Hyponephele lycaon, a species that had not been recorded in Finland since 1936. My observation was communicated to Professor Esko Suomalainen, a geneticist well known for his pioneering work on parthenogenesis who had published a paper on the declining populations and eventual extinction of H. lycaon in Finland (Suomalainen 1958). My amazement was great when, a few days later, I found in the mail a copy of Suomalainen's paper, which I could not read as it was written in German, but which became an instant treasure to me as a small boy.
By a curious coincidence, 35 years later my own research had turned to the very questions explored by Suomalainen in his paper. Hyponephele lycaon was the first species of butterfly to become extinct in Finland in historic times, possibly because of a temporary cooling in the climate which may have been responsible for a decline in the extent of suitable habitat for this species. (The female that I had found in 1964 was probably a vagrant from outside Finland). Suomalainen was particularly interested in the possibility that the final demise of H. lycaon was due to harmful effects of inbreeding in the small dwindling populations in the 1930s. The coincidence is that, working on the second species of butterfly to go extinct from mainland Finland (Melitaea cinxia), we have now produced the first conclusive evidence of its kind that inbreeding indeed increases the risk of extinction of small natural populations (Saccheri et al. 1998, Nieminen et al. 2001).
I entered the University of Helsinki to study Zoology in 1972. As a first-year student, I initially felt destined to a career in a museum, studying entomology, but I soon realised that there was nothing to prevent me from becoming an ecologist instead of a systematist - and an ecologist I became. No doubt my mind was changed by the atmosphere of great excitement in ecology in the early 1970s, and though I was a student in Finland, outside the big centres of scientific discovery, current trends were eagerly assimilated. It was not just a single new direction that ecologists were exploring in those days; several new approaches were established and developed, from behavioural and evolutionary ecology to mathematical population and community ecology to ecosystem ecology. I was living in the midst of keen peers but had only foreign heroes; the transition from old field biology to modern ecology was rather abrupt in Finland.
I became associated with an ecological entomologist, Hannu Koskela, who was engaged in a study of the beetle community living in a habitat that only small boys and ecologists love to inspect, cattle droppings in pastures. A fascinating community with dozens of species coexisting in the same place and apparently using about the same resources! This was a challenge to community theory, and we joined forces to examine the niche relations in this assemblage of 'too many' beetles in a limited space (Hanski and Koskela 1977). I became excited by the opportunities for research with insect communities living in cattle droppings and in animal carcasses. My first piece of field work was based on the idea of experimentally transferring the entire community of carrion-feeding insects from one 'macrohabitat' (forest) to another habitat (field), and vice versa, to investigate the causes of obvious differences in the species composition in the respective communities. In another experiment I manipulated the environmental conditions for the carrion fly community by raising the ambient temperature with a heating cable buried in the soil below the experimental carcasses. These experiments were moderately successful, and they taught me that an ecologist with some imagination could do interesting things without expensive facilities. In 1976, the opportunity arose for me to see one of the big centres of scientific discovery, Oxford University, where I did my D.Phil., continuing with my studies on insect communities but also developing a lasting interest in spatial population structures and dynamics.
Insect communities and spatial population structures
In his Balzan article on biodiversity, Sir Robert May (1999) put the species richness of insects on Earth at 4 million species, of which less than a quarter have been scientifically described. Local species richness, which is what community ecologists are supposed to be able to deal with, is measured in thousands of species. For anyone fascinated by the sight of very many species in a very small space I can give no better advice than to go and look at the insect life in cattle droppings. At the same time, this community highlights the problems of unveiling the dynamics of ecological 'n-body problems', with thousands of direct and indirect interactions between pairs of species.
Some broad issues may not be hopelessly intractable, however. The question regarding the coexistence of large numbers of similar species in the same community is one such general issue. I spent weeks and weeks pondering this question in the late 1970s, without realising that valuable advice might have been obtained locally. Charles Elton, one of the greatest names in the history of ecology, had retired in Oxford in 1967, following a bitter and unsuccessful fight to keep his Bureau of Animal Population alive after his retirement. In the late 1970s, Elton used to come to his office once a week, which gave me the chance to talk to him on several occasions. Regrettably, I had not read his presidential address to the British Ecological Society, published in 1949. In this paper, Elton was decades ahead of his time when he complained about ecological theories being based upon notions of mean density. In his words, ecologists "must learn to take account of the fact that populations are split up into groups or centres of action..." In 1981, Atkinson and Shorrocks (1981) and I (Hanski 1981), working independently on fruit flies and dung beetles, respectively, showed with simple models how right Elton was. The idea is quite straightforward. If species have spatially aggregated distributions, most individuals occur in high-density patches, Elton's centres of action, which situation necessarily increases the level of competition experienced by an average individual in the population. Secondly, if different species are aggregated at least to some extent independently of each other, spatial aggregation will intensify intraspecific competition more than interspecific competition, and hence spatial aggregation will facilitate coexistence. The point is that, as Elton had observed, spatial aggregation is ubiquitous in real populations, and different species practically never show completely correlated spatial distributions. Here, therefore, is a mechanism that very generally makes it easier for many species to coexist. In my contributions, I combined theory and simple models with observational and experimental field work, which produced conclusive results (Hanski 1990). It gave me much satisfaction to demonstrate for myself and to others how the aggregation mechanism worked both in theory and in real communities.
The line of research described above clarified the population dynamic consequences of spatial aggregation, but it left largely open the question about the causes of the observed population aggregation, in other words the dynamics of spatial aggregation itself. In the case of insect populations, this is a hard problem, because population aggregation is here largely due to interaction between complex structure of the habitat and complex behaviour of individuals. Much later, in the 1990s, ecologists focusing on more tractable systems, plant communities inhabiting relatively uniform habitats, have made great progress in modelling both the consequences as well as the population dynamic causes of spatial population structure (Bolker and Pacala 1997, Law and Dieckmann 2000). One of the challenges that remains is to incorporate in the models heterogeneous habitat structure.
What I have written above about spatial population structures and their population dynamic consequences applies to small-scale spatial structure, observed at the scale of single local populations. In 1959, Professor L.R. Taylor working at the Rothamsted Experimental Station initiated an ambitious long-term project to investigate simultaneous changes in time and space of population size in more than 800 species of moths and aphids. Over one hundred sampling stations were established across the UK, which produced, in the due course of time, an unparalleled picture of large-scale spatial dynamics in insect populations (the 'Rothamsted survey' is still going on, and the concept has been imported to several other European countries). Working together with Ian Woiwod and Joe Perry from Rothamsted, I was able to clarify several key questions about population dynamics using the results of the Rothamsted survey. These data conclusively demonstrate regulation even of the populations of insects that appear to oscillate wildly (Woiwod and Hanski 1992), they show how changes in population size are spatially autocorrelated at a large scale, most likely because of similarly correlated environmental conditions (Hanski and Woiwod 1994a), and new insight was gained into the structure of stochasticity in insect population dynamics (Hanski and Woiwod 1994b). Another unique population data set with which I have had a chance to work with describes the multiannual population oscillations of small rodents in Fennoscandia, which exhibit one of the best-described cases of complex dynamics in animal populations (Hanski et al. 1993, Hanski and Henttonen 2001). In the 1980s, there was a period when observational studies were not highly valued in population and community ecology, the focus having shifted to experiments. Today, luckily, a more balanced view prevails, and we realise that there is no substitute for long-term spatially extended data series in ecology, which both tell us what we need to explain and which provide the test-bench against which to assess our theories.
The metapopulation perspective
One of the truly novel ideas that was introduced in ecology in the late 1960s is the dynamic theory of island biogeography (MacArthur and Wilson 1963, 1967). Originally developed to explain the dynamics and the pattern of occurrence of species on islands with dissimilar areas and distances from the mainland, the 'island theory' soon captivated the minds of ecologists at large and was widely adopted by conservation biologists. In a nutshell, the theory explained the distribution of species on islands in terms of area-dependent extinction and isolation-dependent colonization; the number of species to be found in any particular island, or in a nature reserve, would represent, according to this theory, the outcome of the opposing forces of extinction and colonization.
In the late 1980s and early 1990s the island theory began to give way to a new perspective, and a new set of theories, based on the concept of a metapopulation. The metapopulation concept had originally been developed by Richard Levins (1969) at the time when MacArthur and Wilson were working on their island theory, in the late 1960s, but curiously the idea remained dormant for some 15 years. It is also one of the mysteries in the history of population ecology why MacArthur and Levins, who knew each other well, did not produce straightaway a more general theory that would have encompassed their respective models as special cases. The essential difference between the two conceptual frameworks is that the island theory includes a mainland, a permanent source of colonists, whereas the Levins model is concerned with the dynamics of species in networks of habitat fragments without a mainland, and hence a species may become permanently extinct in the Levins model. The island model was specifically focused on the effects of island area and isolation on extinction and colonization, whilst the original Levins model assumes an infinite number of identical habitat fragments. My own contribution in the 1990s was to produce a synthesis of the two models, allowing for spatial variation in habitat fragment areas and connectivities in finite patch networks without a mainland (Hanski 1998, 1999). We now have a relatively well-developed theory (Ovaskainen and Hanski 2001) and models that can be parameterized with empirical data (Moilanen 1999) and applied to real metapopulations to generate quantitative predictions about the distribution of species in highly fragmented landscapes (Wahlberg et al. 1996, Moilanen et al. 1998). We can answer questions about the capacity of fragmented landscapes to support viable metapopulations of particular species (Hanski and Ovaskainen 2000), and though the models do not yet incorporate all relevant processes, such as regional stochasticity, they will soon do so. My hope is that the kind of 'spatial realism' that the current ecological models of metapopulation dynamics incorporate would next be transported to genetic and evolutionary models.
The major shortcoming of the metapopulation theory that I have helped construct is that it applies primarily to 'highly fragmented' landscapes, by which I mean landscapes in which the habitat of interest occurs as a network of discrete fragments. It remains a big challenge to develop comparable theory for landscapes with a less well-defined structure. On the other hand, another line of theoretical research on spatial dynamics assumes no landscape structure at all (Hanski 1998), these theories being concerned with the conditions under which population dynamic processes alone may generate and maintain spatial variation in population density. One of the challenges for research in this area of ecology is to produce a more unified theory that would incorporate in one framework the different approaches that we now have to spatial population dynamics.
The shrinking world
Several factors have contributed to the shift from the island theory to metapopulation theories (Hanski and Simberloff 1997), of which one no doubt is the ongoing worldwide transformation of natural landscapes. With real landscapes losing 'mainlands', large continuous areas of natural habitat, the metaphor of islands located outside a large mainland is becoming a poor metaphor. While we humans with a few other species experience globalisation, for most species the Earth is shrinking and breaking up into small, disconnected pieces. It is inevitable that increasing numbers of species will become extinct, not just hundreds of species but thousands, tens of thousands and hundreds of thousands of species. Perhaps a few million species will become extinct in the coming decades and centuries, in the order of half of the current stock of species.
During my first year at Oxford, I had the honour of having Charles Elton, already an old man, attending my seminar. I talked about the community of dung beetles and about the problem of coexistence of great numbers of species in spite of seemingly limited differences in their ecologies. Elton asked me which species I considered as most important in the community. I could not name such species, an answer that appeared to please him. Nor should we attempt to name, today, the species that are important enough to be allowed to survive. This question is not about science, of course, but there is no reason why scientists should not express their anxiety about short-term human greediness permanently eradicating much of the natural beauty in the world as well as parts of the living machinery tested by billion-year-old evolution. As we should know, our own existence is ultimately dependent on this machinery.
Depressing as it is, there is little hope that the present trajectories of global change will take a sudden turn. In the case of habitat loss and fragmentation, the task of population ecologists is to gain better understanding of the biological consequences - and to make sure that this understanding is conveyed as accurately as possible to society at large. Apart from that, one can only hope that, at some point in the foreseeable future, this society will be ready for action to slow down and ultimately to stop and reverse the current trends of environmental deterioration.
Thoughts on population ecology
I have suggested that the current popularity of metapopulation theories is partly due to the obvious need to better understand the population consequences of habitat loss and fragmentation. It is also important that the current models have been successful in giving guidance to empiricists in the planning of their research. In a similar manner, the island theory of MacArthur and Wilson became so popular so quickly in its time largely because it gave a simple explanation to a universal pattern in the distribution of species, the species-area relationship, and made empirical research on this pattern so much more interesting. In both cases, a key component is the mapping of the population dynamic processes of extinction and colonization onto the structure of the landscape, described in terms of the areas and connectivities of islands, reserves or any other fragments of habitat. That such area and isolation effects occur is almost a truism, given basic understanding of population dynamics and dispersal, but the challenge is to produce a theory that will explain these effects in quantitative terms.
Much of theoretical ecology suffers from a lack of meaningful bridges to empirically-based research. Combining theory and research on real populations is difficult in ecology, because ecological systems are characterized both by historical contingency (populations and communities are perturbed by processes operating at many scales) and great dynamic complexity (many species with innumerable direct and indirect interactions). Frustrated by the failures to develop a predictive theory at the scale of individual communities, some ecologists have advocated an approach dubbed macroecology. Macroecology is focused on large-scale statistical patterns in the distribution and abundance of species and in measures of community structure, such as the distribution of body sizes. Unfortunately, documenting these patterns at large spatial scales may not increase our understanding of the relevant ecological processes. In this respect, I consider that the area and isolation effects on extinction and colonization, which are the cornerstones of much of the current metapopulation theory, represent 'macroecology' at a proper scale, not too small to require the hugely difficult reductionist analysis of local populations and communities, but small enough to retain a handle on the relevant processes.
One approach to strengthening the dialogue between theory and empirical research in population ecology is via large-scale projects that employ a particular population, metapopulation or a community as an 'ecological research facility' to address broad issues. In such a context, one may accumulate a sufficient body of information and knowledge to deal with the problems of historical contingency and complex interactions, and one is in general in a good position to promote a balanced advancement of theory and empirical research. In my research group, we have focused on one such ecological research facility for the past 10 years, a large metapopulation of the Glanville fritillary butterfly (Melitaea cinxia) in the Åland Islands in the northern Baltic (Hanski 1999, chapters 11 and 12). The backbone of this research is a data base on thousands of habitat fragments suitable for the species, supplemented with a range of remote-sensed information to describe the environment. We monitor all local populations, which number several hundreds in each year, and we have gradually accumulated knowledge of more than 1000 population turnover events, extinction and colonization. It is clear that developing predictive models is greatly facilitated by such a mass of knowledge and information, and it is simply very helpful for the more narrowly-defined research projects of students and researchers to have that knowledge available as a context into which to place their own results. Many other ecological research projects are even bigger than ours, but it is probably fair to say that the great bulk of population ecological research continues to be conducted without the benefit of knowing the ecological context. Much potentially useful information will be lost because of fragmentation of our pooled research effort. The community of ecologists must learn to define its research priorities and to find the means of implementing the research at a scale that is required to advance our science (Lawton 2000).
Atkinson, W.D. and Shorrocks, B. 1981. Competition on a divided and ephemeral resource: a simulation model. Journal of Animal Ecology 50, 461-471.
Bolker, B. and Pacala, S. W. 1997. Using moment equations to understand stochastically driven spatial pattern formation in ecological systems. Theoretical Population Biology 52, 179-197.
Elton, C. 1949. Population interspersion: an essay on animal community patterns. Journal of Animal Ecology 37, 1-23.
Hanski, I. 1981. Coexistence of competitors in patchy environments with and without predation. Oikos 37, 306-312.
Hanski, I. 1990. Dung and carrion insects. In: B. Shorrocks and I. Swingland, eds. Living in a patchy environment, pp. 127-145. Oxford University Press, Oxford.
Hanski, I. 1998. Metapopulation dynamics. Nature 396, 41-49.
Hanski, I. 1999. Metapopulation Ecology. Oxford University Press, Oxford.
Hanski, I. and Henttonen, H. 2001. Population cycles of small rodents in Fennoscandia. In: Berryman, A.A., ed. Population cycles. Chicago University Press, in press.
Hanski, I. and Koskela, H. 1977. Niche relations amongst dung-inhabiting beetles. Oecologia 28, 203-231.
Hanski, I. and Ovaskainen, O. 2000. The metapopulation capacity of a fragmented landscape. Nature 404, 755-758.
Hanski, I. and Simberloff, D. 1997. The metapopulation approach, its history, conceptual domain, and application to conservation. In: Hanski, I. and Gilpin, M.E., eds. Metapopulation biology, pp. 5-26. Academy Press, San Diego.
Hanski, I., Turchin, P. Korpimäki, E. and Henttonen, H. 1993. Population oscillations of boreal rodents: regulation by mustelid predators leads to chaos. Nature 364, 232-235.
Hanski, I. and Woiwod, I.P. 1994a. Spatial synchrony in the dynamics of moth and aphid populations. Journal of Animal Ecology 62, 656-668.
Hanski, I. and Woiwod, I.P. 1994b. Mean-related stochasticity and population variability. Oikos 67, 29-39.
Law, R. and Dieckmann, U. 2000. A dynamical system for neighborhoods in plant communities. Ecology 81, 2137-2148.
Lawton, J.H. 2000. Community ecology in a changing world. Ecology Institute, Oldendorf/Luhe.
Levins, R. 1969. Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America 15, 237-240.
MacArthur, R.H. and Wilson, E.O. 1963. An equilibrium theory of insular zoogeography. Evolution 17, 373-387.
MacArthur, R.H. and Wilson, E.O. 1967. The theory of island biogeography. Princeton University Press, Princeton.
May, R. 1999. A panoramic view. Premi Balzan 1998. Milano.
Moilanen, A. 1999. Patch occupancy models of metapopulation dynamics: efficient parameter estimation with implicit statistical inference. Ecology 80, 1031-1043.
Moilanen, A., Smith, A.T. and Hanski, I. 1998. Long-term dynamics in a metapopulation of the American pika. American Naturalist 152, 530-542.
Nieminen, M., Singer, M.C., Fortelius, W., Schöps, K. and Hanski, I. 2001. Experimental confirmation that inbreeding depression increases extinction risk in butterfly populations. American Naturalist, in press.
Ovaskainen, O. and I. Hanski. 2001. Spatially structured metapopulation models: metapopulation capacity and threshold condition for persistence. Manuscript.
Saccheri, I., Kuussaari, M., Kankare, M., Vikman, P., Fortelius, W. and Hanski, I. 1998. Inbreeding and extinction in a butterfly metapopulation. Nature 392, 491-494.
Suomalainen, E. 1958. Über das Vorkommen und spätere Verschwinden von Epinephele Lycaon Rott. (Lep., Satyridae) in Finnland. Annales Entomologici Fennici 24, 168-181.
Wahlberg, N., Moilanen, A. and Hanski, I. 1996. Predicting the occurrence of endangered species in fragmented landscapes. Science 273, 1536-1538.
Woiwod, I.P. and Hanski, I. 1992. Patterns of density dependence in moths and aphids. Journal of Animal Ecology 61, 619-629.