Spatial Ecology workshop on
Extinction
 

Programme (4 - 8 November 2000)

 Saturday, 4 November
 18.00  Bus leaves from the department in Helsinki
 20.00  Estimated arrival time at Tvärminne Zoological Station
 20.30  Light evening meal
 
 Sunday, 5 November
9.00 - 10.00 Breakfast
10.00 - 10.10 ILKKA HANSKI: Opening words
10.10 - 11.00 CHRIS COSNER: Factors influencing local extinctions in habitat fragments: some hints from simple spatial models |abstract|
11.00 - 11.30 Coffee break
11.30 - 12.15 KENT HOLSINGER: Demography and extinction in plant populations |abstract|
12.15 - 13.00 BERNT-ERIK SAETHER: Predicting time to extinction in passerine bird populations |abstract|
13.00 - 14.00 Lunch
14.00 - 14.30 JOHAN KOTZE: Range size-body size relationships and the status of carabid beetles in Belgium, Denmark and the Netherlands |abstract|
14.30 - 15.15 GRANT HAMILTON: Population system persistence in a catastrophic system: the wild rabbit in arid and semi-arid Australia |abstract|
15.15 - 15.45 Coffee break
15.45 - 16.30 ANTHONY IVES: Extreme cycles in the dynamics of midges in lake Myvatn: the possibility of resource-consumer cycles |abstract|
16.30 - 17.15 JÖRGEN RIPA: On stochastic population models and the route to extinction |abstract|
18.00 Workshop dinner
 
Monday, 6 November
8.00 - 9.00 Breakfast
9.00 - 10.00 MICHAEL WHITLOCK: Genetic load and extinction in small and subdivided populations |abstract|
10.00 - 10.15 Coffee break
10.15 - 11.00 LAURENCE LOEWE: How many beneficial mutations are needed to stop Mullers Ratchet in MTDNA? |abstract|
11.00 - 12.00 STEINAR ENGEN: Migration and spatio-temporal variation in population dynamics in a heterogeneous environment |abstract|
12.00 - 13.00 Lunch
13.00 - 14.00 JORDI BASCOMPTE: Environmental stochasticity and metapopulation extinction |abstract|
14.00 - 14.30 Coffee break
14.30 - 15.30 OTSO OVASKAINEN: Stochastic versus deterministic metapopulation models |abstract|
15.30 - 16.15 WOLFGANG WEISSER: Extinctions in aphid metapopulations |abstract|
17.00 - 18.00 Dinner
18.00 - Sauna
 
Tuesday, 7 November
8.00 - 9.00 Breakfast
9.00 - 10.00 ILKKA HANSKI: Time delays in metapopulation dynamics and the extinction debt in communities |abstract|
10.00 - 10.30 Coffee break
10.30 - 11.15 BOB O'HARA: Uncertainty, stochasticity and more stochasticity in estimating the extinction risk of metapopulations |abstract|
11.15 - 12.00 MIKKO HEINO: Extinctions in temporally correlated environments |abstract|
12.00 - 13.00 Lunch
13.00 - 13.30 MAR CABEZA: Design of reserve networks and the persistence of biodiversity |abstract|
13.30 - 14.00 CHRISTOPH HAAG: Extinction in a Daphnia magna metapopulation |abstract|
14.00 - 14.30 Coffee break
14.30 - 15.15 MIKAEL FORTELIUS: Ecological anatomy of an extinction event 10 million years ago |abstract|
15.15 - 16.00 JUKKA JERNVALL: Contribution of inferred abundance to evolutionary trends in the dental capability of western Eurasian ungulates |abstract|
17.00 - 18.00 Dinner
18.00 - Optional round table discussion
 
Wednesday, 8 November
8.00 - 9.00 Breakfast
9.00 Bus leaves from Tvärminne Zoological Station
11.00 Approximate arrival time in Helsinki

TOP

Abstracts

 

Jordi Bascompte
Title: Environmental stochasticity and metapopulation extinction.
Abstract: In this talk I will present a general model of a metapopulation subjected to human induced mortality in stochastic environments. The model is particularly relevant to species with weak density dependence, spatial structure, high dispersal, and subjected to highly variable biotic conditions. I will extend well known theory on geometric mean fitness to obtain analytical expressions for the minimum number of local populations required for metapopulation persistence. These expressions provide a clear way to relate the magnitude of environmental fluctuations, the minimum number of local populations, and the maximum levels of human induced mortality for metapopulation persistence. Different scenarios can be interpreted under this theoretical background, ranging from conservation biology, where the challenge is the preservation of the metapopulation, to biological control, where the challenge is the eradication of a pest.

TOP

Mar Cabeza
Title: Design of reserve networks and the persistence of biodiversity.
Abstract: Site-selection algorithms are used in reserve design to identify networks of sites which maximize diversity given some constraints (e.g. cost, total area, etc.). These algorithms are mostly based on a snapshot of species occurrence and they ignore the question of how well species persist in the selected sites. I will present a theoretical approach that considers spatiotemporal dynamics in order to assess effectiveness of site-selection algorithms in terms of species persistence. Once a set of sites is selected by a common algorithm, we apply a metapopulation model to simulate species occurrence through time. We look at the species extinctions at the selected sites in two situations: first, when all sites are included in the simulation, and second, when the non-selected-sites are excluded from the simulation (i.e. assumed to be destroyed if they do not become a reserve). The distinction is done to emphasize the importance of the non-selected sites for the spatial dynamics of the system. The results show that species persistence may be strongly dependent on patches not included in the selection. In summary, our results support the call for the integration of spatial population modelling in reserve network design.

TOP

Chris Cosner
Title: Factors influencing local extinctions in habitat fragments: some hints from simple spatial models.
Abstract: The dynamics of a population inhabiting a small habitat patch can be affected in various ways by factors related to patch size and geometry, the permeability of the patch boundary, the nature of the landscape between patches, and the dispersal characteristics of the population. In the case of a single species or a small number of interacting species in a single patch it is often possible to gain insight into the role of spatial effects via diffusion models. This talk will describe some of the unexpected predictions of reaction-diffusion models in the context of persistence or extinction of populations in a single patch. For example, changing conditions on the patch boundary in a way that would benefit a population if it were alone can sometimes harm the population if it competes with another species. The change in boundary conditions can sometimes cause a reversal of competitive dominance which could potentially lead to a local extinction. There are a number of other such effects, some of them surprising, which emerge from the way that population dynamics and dispersal characteristics interact with patch size or boundary conditions. In many cases geometric aspects of the models can be accounted for via the principal eigenvalues of differential operators which describe the diffusion or other forms of dispersal of the population.

TOP

Steinar Engen
Title: Migration and spatio-temporal variation in population dynamics in a heterogeneous environment.
Abstract: A continuous stochastic population model in space and time is presented. The model includes migration, spatially correlated environmental noise and accounts for spatial heterogeneity in habitat quality using a general spatial autocorrelation in the local carrying capacities. The model is analyzed by Fourier methods based on linearizations which are realistic when the population fluctuations are not too large. An expression for the spatial autocorrelation for the population density is derived together with some relatively simple relations for the scaling of these autocorrelations under the assumption that the migration rate is constant. For strong local density regulation, the spatial density pattern is approximately the same as the pattern of spatial heterogeneity in habitat quality. For weak density regulation the scaling increases with migration rate and distance and is approximately the same as the scaling of the environmental noise if the migration rate and distance are large. A required distinction between the spatial autocorrelation function and the function describing the spatial variation in the synchrony of population fluctuations is made. The effects of migration distance, migration rate, heterogeneity in carrying capacities, strength of local density regulation as well as the magnitude and spatial autocorrelation in the environmental noise are discussed. Numerical examples illustrating the general findings are presented. If the carrying capacities are lognormally distributed in space, and individuals migrate in subsequent small steps, the model generates exactly a lognormal spatial distribution of abundances. For this model it is possible to derive the joint spatial and temporal autocorrelations, which in turn may be used to find approximations for the time to local extinction or quasi-extinction in areas of different sizes.

TOP

Mikael Fortelius
Title: Ecological anatomy of an extinction event 10 million years ago.
Abstract: The so-called "Vallesian Crisis" seen in the mammal faunas of Western Europe at about 10 Ma is usually described as a local extinction event resulting in a significant (perhaps 50%) reduction in species richness. Critical analysis of the data suggest a dramatic decrease in sampling intensity at the same time, however, leaving the degree (indeed, existence) of reduction in species richness open to question. Despite this, and despite the lack of independent evidence for any major change in the environment at the time, the extinction event itself is not in doubt. It can be shown that the ecological composition of the mammal community changed markedly at that time, with disproportionate loss of the smallest large mammal species of omnivores and carnivores, and general replacement of native lineages by immigrants from Eastern Europe and Asia, resulting in considerable reduction in endemism. The Vallesian crisis also marks the transition in Western Europe from an ungulate fauna dominated by rhinoceroses and suids to one dominated by horses and, especially, ruminants. Thus, a major extinction event did clearly take place, but the evidence from community structure is considerably more robust and easier to interpret than are the uncertain species richness counts. This may be generally true of extinction events in the fossil record.

TOP

Christoph Haag
Title: Extinction in a Daphnia magna metapopulation.
Abstract: Daphnia magna occurs in thousands of rockpools in the archipelago of Southern Finland. The pools show a metapopulation dynamic with frequent extinction and recolonisation, the overall yearly turnover rate being approximately 20%. However, newly founded population have a considerably higher extinction risk (40%) than already established pools (15%). We have investigated two factors, which might explain this pattern. First, we found evidence for strong inbreeding following colonisation events. In an experiment strong inbreeding depression in the presence of competition between inbred and outbred individuals led to the fixation or near fixation of the outbred individuals. In the absence of competition inbreeding depression was less strong but of a total of 99 populations seven inbred and no outbred populations went extinct during a period of three months. Second, less suitable pools might be colonised from time to time for a short period but more permanent populations might fail to establish in such pools. Indeed we found that the establishment of populations in empty pools fails in most cases (20% success). If however limestone is added to empty pools, increasing the Ca++-content and buffering the pH, the rate of establishment increases to 80%. Our data suggest that both, inbreeding depression and water chemistry influence the extinction of local Daphnia magna populations. In our future work we plan to follow these two factors over time.

TOP

Grant Hamilton
Title: Population system persistence in a catastrophic system: the wild rabbit in arid and semi-arid Australia.
Abstract: Local populations of wild rabbit in arid and semi-arid zones of Australia undergo frequent and widespread local extinctions due to highly irregular rainfall, but rabbit population systems persist despite this. In this system, instability results from environmental catastrophes that cause the distribution of resources per individual to fluctuate widely. A consequence of these fluctuations is that all suitable habitat patches are not occupied at any one time. In a system underpinned by recurrent extinctions on this scale, system persistence is driven by the potential for recolonisation. Previous studies have shown that two distinct, adjoining rabbit population systems occur in southern central Queensland, a panmictic western system and a structured eastern system. The high degree of genetic structuring evident in the eastern system cannot be explained by traditional models of isolation by distance or geographical barriers. Given that the habitat matrix between local populations is able to support small, transient populations, an extension of traditional modelling approaches is needed to explain underlying processes. The region was considered to be saturated with resource patches of varying quality, and connectivity among local populations was modelled. Model output was validated by regression against independent genetic population structure data, and explained over 80% of the variability within the system.

TOP

Ilkka Hanski
Title: Time delays in metapopulation dynamics and the extinction debt in communities
Abstract: Metapopulation size responds with a delay to perturbations, such as a change in the structure of a fragmented landscape. In the community context, extinction debt refers to the number of species whose new equilibrium following environmental change is global extinction, but which haven't yet had time to go extinct. We have worked on deterministic spatially realistic metapopulation models, which assume a finite number of patches and spatial variation in patch areas and connectivities. The models assume some relationship between local extinction rate and patch area, and between colonization rate and connectivity. In a spatially realistic version of the Levins model, the time delay is a product of four factors: the strength of the perturbation, the ratio between the metapopulation capacity of the landscape and the threshold value set by the properties of the species, the characteristic population turnover rate, and an additional factor determined by the structure of the landscape. Time delay is especially long when the perturbation is large, when the species is close to the threshold for persistence, when the characteristic turnover rate is slow, and when there are only a few dynamically important patches in the network. In communities, the extinction debt is expected to be especially great if many species are close to their extinction threshold following habitat loss. A corollary is that landscapes that have recently experienced substantial habitat loss and fragmentation are expected to show a transient 'overabundance' of rare species.
Papers:
Hanski & Ovaskainen, 2000, The metapopulation capacity of a fragmented landscape. Nature 404:755-758.
Ovaskainen & Hanski, Spatially structured metapopulation models: metapopulation capacity and threshold conditions for persistence. Manuscript.
Ovaskainen & Hanski, Time delay in metapopulation response to perturbation. Manuscript.
Hanski & Ovaskainen, Extinction debt at extinction threshold. Manuscript.

TOP

Mikko Heino
Title: Extinctions in temporally correlated environments
Abstract: Most ecological models use white noise as the null model for environmental stochasticity, despite the increasing consensus on that environmental variations tend to have positive temporal correlations ("red"). However, including temporal correlations to stochastic population models is not entirely trivial. Positively autocorrelated red environmental noise is characterized by a strong dependence of expected sample variance on sample length. This dependence has to be taken into account when assessing extinction risk under red and white uncorrelated environmental noise. To facilitate a comparison between red and white noise, their expected variances can be scaled to be equal, but only at a chosen time scale. Simple one-dimensional population dynamics models show that the different but equally reasonable choices of the time scale yield qualitatively different results on the dependence of extinction risk on the colour of environmental noise: extinction risk might increase as well as decrease when the temporal correlation of noise increases.

TOP

Kent Holsinger
Title: Demography and extinction in plant populations.
Abstract: Populations subject to deterministic declines are certain to go extinct within a relatively small period of time, and the time to extinction is relatively insensitive to the current size of the population. Even populations that grow in size, on average, will persist for a relatively short time if the variance in annual growth rate is more than about twice the mean. As a result, isolated populations are not likely to persist indefinitely unless they are actively managed. A demographic approach to management of rare plant populations suggests that conservation managers focus their efforts on those life-history stages between which transitions most limit population growth. Demographic analyses of long-lived plants typically use matrix models to assess the sensitivity of mean population growth rates to changes in different life-history stages. Because the variance in growth rates cannot be ignored and because rare, catastrophic events can have a large impact on persistence probabilities, however, analyses of demographic models are better regarded as giving conservation managers tools to guide their intuition than as providing them with quantitative estimates of persistence probability.

TOP

Anthony Ives
Title: Extreme cycles in the dynamics of midges in Lake Myvatn: the possibility of resource-consumer cycles.
Abstract: Midges, primarily the species Tanytarsus gracilentus, in Lake Myvatn, Iceland, show quasi-cycles of four orders of magnitude with a period of roughly 5-7 years. Time-series analyses suggest that these cycles are generated by consumer-resource interactions. T. gracilentus is a benthic filter feeder that reaches sufficient densities to eliminate its resource base. Analyses incorporating both population density and wing length (as a surrogate for resource availability) show the pattern expected of consumer-resource cycles; fluctuations in density lag behind fluctuations in wing length. The analyses also demonstrate the importance of accounting for observation (measurement) error to describe midge dynamics accurately. Midges provide the primary food for several breeding bird populations, and understanding the processes causing midge outbreaks and crashes will help conservation management of these species.

TOP

Jukka Jernvall
Title: Contribution of inferred abundance to evolutionary trends in the dental capability of western Eurasian ungulates.
Abstract: The advance of databases on large-scale species-locality-occurrences (SPLOC:s) offer a means to link ecological-level processes with patterns observed in palaeontological data. In recent communities, abundant species may contribute disproportionately to the community organization and biomass. While abundance is difficult to estimate from compilations of locality data directly, abundance can be approximated indirectly using the frequency of occurrence of taxa. Within a region and time-unit, fossil species occurring at many localities may be regarded as "common", and distinguished from "rare" species with few locality occurrences. We used SPLOC:s to test whether trends in the evolution of dental capability (hypsodonty times the number of cutting edges) differ between common and rare large plant-eating mammals over the 20-million year interval preceding the Quaternary Ice Age. Initially, rare taxa show higher dental capability than common taxa but the increase in dental capability is more pronounced in the common taxa. The trends for hypsodonty alone and for different aspects of lophedness are all shown stronger by the common species than by the rare ones. The common taxa contribute only about 10% or less to the total species list, a proportion observable in recent communities as well. Therefore, to the extent that species-locality-occurrences reflect abundance, the stronger trends seen among common taxa may be more representative of the community level changes over the Neogene than the overall pattern. Conversely, the use of simple taxon presence lists may underestimate the
ecological changes inferred from palaeontological data.

TOP

Johan Kotze
Title: Range size-body size relationships and the status of carabid beetles in Belgium, Denmark and the Netherlands.
Abstract: The relationship between body size and range size is characteristically positive, i.e. larger species have wider ranges than smaller species. Range size and body size are also related to probability of extinction. For example, species of small geographic range size and large body size have been claimed to exhibit high probabilities of extinction. We tested these predictions using carabid beetles (Carabidae, Coleoptera) from three northern European countries; Belgium, Denmark and the Netherlands. There was no strong relationship between beetle range size and body size. However, carabid range, speciality, habitat preference and body size strongly influenced carabid beetle status (i.e. increasing, decreasing or stagnant) in these countries. Geographically restricted, stenotopic and small beetles are decreasing while geographically widespread, eurytopic larger beetles are generally increasing. Beetles in various dry habitats, i.e. dry grasslands and dry open habitats are decreasing while beetles in coastal and bog habitat are increasing in these countries. These results are discussed in relation to recent habitat alteration in Belgium, Denmark and the Netherlands.

TOP

Laurence Loewe
Title: How many beneficial mutations are needed to stop Mullers Ratchet in mtDNA?
Abstract: Recent observations of mitochondrial DNA in pedigrees revealed surprisingly high mutation rates. To reconcile these levels with significantly lower expectations based on phylogenetic analyses, hotspots and long-term removal of mutations by weak selection have been invoked. The latter, however, would imply an increased genetic load. Together with recent estimates of slightly deleterious mutation rates in the human genome this may seem to endanger long-term survival. My simulations of such populations have shown that biologically meaningful sets of parameters exist which do lead to extinction, if only deleterious mutations, selection and drift are considered. However, this may be avoided by advantageous mutations. These may compensate directly for specific deleterious mutations or may lead to new capabilities improving fitness. Furthermore, compensatory adaptation in quantitative traits may prolong time to extinction or even prevent deterioration of absolute fitness at all. Different mutational effects and other parameters are explored by stochastic computer simulations in order to find the critical frequency of advantageous mutations necessary to stop Mullers Ratchet under these circumstances.

TOP

Bob O'Hara
Title: Uncertainty, stochasticity and more stochasticity in estimating the extinction risk of metapopulations.
Abstract:Predicting extinctions is an uncertain business. Population dynamics are intrinsically stochastic. The environment is also stochastic, which adds to the stochasticity of the population. These elements of randomness arise from the stochastic nature of the world we observe. But we should also consider uncertainties in the models we build, in the parameters of the models, and in the models themselves. I have taken patch occupancy data from Finnish butterfly metapopulations, and used Bayesian methods to fit a stochastic metapopulation model (the Incidence Function model) to the data. The parameter estimates are expressed as probability distributions, and these are used to make predictions about the time to extinction of the metapopulations. The predictions give us a common currency to compare the different forms of variability. For the data, predictions are made for the model fitted both with and without environmental stochasticity, and for each of these, with both point estimates of the parameters obtained from the analysis (implying an assumption of complete knowledge about the parameters), and also from the full posterior distribution, which allows us to include our uncertainty in the parameter estimates in the predictions. The effects of these different forms of variability on the predictions will be discussed.

TOP

Otso Ovaskainen
Title: Stochastic versus deterministic metapopulation models
Abstract:Although real biological systems are of stochastic nature, deterministic models may in many cases give reasonable approximations for the observed phenomena. However, this is not always the case, and deterministic models should thus be applied with caution. I have studied spatially realistic metapopulation models (meaning that I take explicitly into account the areas and locations of the habitat patches, but ignore any further details) both through the deterministic and the stochastic approaches. The main shortcoming of the stochastic model is that the size of the problem increases as 2 to the power of n, where n is the number of habitat patches. Consequently, the stochastic model is not tractable (for anything but simulation) for large n. On the other hand, the standard deterministic description of a metapopulation lacks some of the most important features, such as the assessment of time to metapopulation extinction, and correlations in patch-occupancy probabilities. I demonstrate how the deterministic, spatially realistic Levins model may be applied to give information about the quasi-stationary distribution of the corresponding stochastic model. I also demonstrate how the correlation structure induced by the stochastic model may be incorporated into the deterministic model. As an application, I study the effect of spatially correlated extinction risk.

TOP

Jörgen Ripa
Title: On stochastic population models and the route to extinction.
Abstract: This talk has two topics. First, I will discuss how populations actually do go extinct, and what is predicted by population models. The focus question is whether populations go extinct from low densities, as we might think, or in catastrophic crashes from high densities, due to resource depletion. Models predict both, depending on the maximum and density dependence of the population growth rate and details in the population growth process. This analysis is performed by elaborate stochastic population dynamic modeling. Second, I will relate these results to what can be predicted from simple linear modeling and discuss the usefulness and possible limitations of linear models in theoretical population ecology.

TOP

Bernt-Erik Saether
Title: Predicting time to extinction in passerine bird populations
Abstract: Predicting time to extinction is of central importance in population viability analysis. Recently concerns have been raised about the practical applicability of such methods because of large uncertainties in model predictions. Such uncertainties arise from stochastic effects on the population dynamics as well as it may difficult to obtain accurate estimates of essential model parameters Here we introduce the term Population Prediction Interval (PPI) to quantitatively account for such uncertainties. A PPI is a stochastic interval that includes the unknown variable to be predicted with probability (1- p), where p is the cover probability. A prediction interval may be considered similar to a confidence interval in statistics, except that we draw inference about a stochastic quantity rather than a parameter. We illustrate how different parameters (e.g. strength of density regulation, specific growth rate, demographic and environmental stochasticity) as well as the uncertainty in them affect the prediction of time to extinction in small passerine populations.

TOP

Wolfgang Weisser
Title: Extinctions in aphid metapopulations.
Abstract: Meta-populations of many aphid species are characterised by frequent extinction and colonisation events. Extinctions may occur at several spatial scales, on the level of individual plant ramets (aphid 'colonies'), on the level of the plant (genet), on the level of isolated sites with a number of genets, and also at higher spatial levels. While the rates of extinctions are higher than in many other described metapopulations, the causes of population extinctions are often difficult to establish. Using the aphid species Metopeurum fuscoviride, Macrosiphoniella tanacetaria, and Acyrthosiphon pisum as model organisms, the causes of extinctions are investigated, on the level of individual aphid colonies. Experimental evidence is presented to show that natural enemies can play an important role in aphid extinctions. In contrast to the results from these experimental manipulations, however, non-manipulative field studies often fail to find clear evidence for an involvement of natural enemies. Thus, the quantitative importance of natural enemies, and in fact of other extinction mechanisms, mostly remains unclear in population studies of aphid-predator systems. This raises the general question how population studies with the aim to determine the importance of extinction factors should be performed. For insect predator-prey systems, is seems that a number of general methodological problems currently hinder quantitative studies of extinction causes.

TOP

Michael Whitlock
Title: Genetic load and extinction in small and subdivided populations.
Abstract: Beyond the identification of taxa and evolutionary units, the main contribution of genetics to conservation biology has been the recognition that the accumulation of deleterious alleles due to genetic drift in small populations threatens the continued persistence of endangered species. Here I extend previous results on the fixation of deleterious alleles to include the effects of compensatory and beneficial mutations, sexual selection, and population structure. Beneficial and compensatory mutations can, in populations larger than a certain effective population size, maintain fitness at an equilibrium which should allow the continued persistence of populations. Based on good guesses about certain unknown genetic parameters, this critical effective size may be relatively small, in the hundreds or thousands. Furthermore, the effects of new mutations are likely to also include effects on mating success (as demonstrated by some new data from Drosophila), and I sow that this effect can substantially reduce the load due to new mutations. In contrast, the subdivision of a species into geographically separated populations can result in a higher rate of fixing deleterious alleles and therefore extinction.

TOP

Back to Extinction Workshop page

Back to Spatial Ecology home page