To respond to millennial-scale climatic change in the past, species shifted their ranges. Landscape, regional, and global models need to incorporate realistic dispersal and migration functions to capture this process. These models are limited, however, by lack of data-based functions of individual dispersal and establishment. Moreover, the extent and connectivity of habitat is being changed by humans. Species face a double-bind: smaller, less connected habitats and a changing climate. Fragmentation will result in the extinction of species because reduced interactions change the long-term dynamics, and this effect will be exacerbated by climatic change. The interaction also creates a double-whammy for researchers: it is difficult to differentiate the effects of climatic change from those of fragmentation due to disequilibrium (e.g., an 'extinction debt'). Understanding the effects of fragmentation on dispersal is critical to assessing impacts of global change.

Successful modeling of dispersal and migration should identify how landscape structures will interact with climatic change and ecological processes to determine future distributions. Current DGVMs compare ubiquitous dispersal versus no dispersal, and current regional and landscape dispersal models have functions based on assumed seed shadows. Data-based functions are needed improve landscape scale models and be scaled to DGVMs. The sensitivity of the migration process to establishment processes also needs to be determined.

Objectives

To compile data sets on dispersal and recruitment patterns from which dispersal probability distributions, or kernels, can be derived. Kernels will describe dispersal per se; process-based establishment functions need to be included to obtain full spatial models of the recruitment process.

To use these data in model comparisons on different landscapes in order to examine how much landscape fragmentation can impede species migration. The degree of impediment can be used to create dispersal functional groups and identify dispersal-limited or recruitment-limited taxa or groups.

To examine scaling properties for the dispersal functions on differing landscapes and to derive rules for use in DGVMs or regional scale vegetation models.

To tie into empirical work on dispersal and on Quaternary migrations of vegetation.

Implementation

To address objective (1), a core network of empiricists and modellers will work on synthesizing dispersal data sets. Data sets with consistent format will be analyzed to create dispersal kernels. Needs for further data for missing types of dispersal will be identified. Approaches to generate data sets, such as mechanistic aerodynamic models, or animal behavior observations will also be investigated. Particular attention will be devoted to seeking funds and an agency for support of the data base.

The modelling tasks involved in objectives (2) to (4) will be carried out by another partly overlapping network. The approach will involve the use of common or comparable protocols for creating heterogeneous landscapes (random, hierarchical, or fractal patterns) on which dispersal is simulated with shared initial kernels. Modeling could then be extended to real landscapes with contrasting patterns. Further comparisons will assess the sensitivity to dispersal of a set of process- and rule-based landscape models of vegetation dynamics. Finally, another set of comparisons should address scaling procedures, by analyzing the effects in theoretical models of different renormalization rules used to aggregate cells. The results of this theoretical approach will be used to 1) describe specific spatial processes of interaction between landscape pattern and dispersal kernels; 2) create dispersal functional groups and identify dispersal-limited or recruitment-limited taxa or groups; 3) contribute dispersal modules to the generic landscape models developed in Task 2.2.4, and 4) develop dispersal rules for DGVMs. Current analyses indicate that paleoecological data will provide minimum estimates of potential migration rates. These data, when combined with local climate/environmental constraints, represent a valuable test of regional and landscape models.

Throughout the development of the Task, contacts among researchers will be maintained through 1) a small listserver to be rapidly created for forum discussion and data exchange; 2) a web page, including an early warning bibliographic system on work in press. Exchanges with pest dispersal modellers will be encouraged throughout.

Milestones

October 1996 Global Changeand Plant Migration workshop, Bateman's Bay

26-28 October 1997 (initial activity workshop), to refine the goals in concert with others working on landscape processes

December 1997 Iowa City working group meeting, to run trials of simulations on different "neutral" landscapes, to discuss various approaches to creating kernels from simulated data and to initiate the discussion on scaling

12-14 March 1998 Barcelona GCTE-LUCC science conference: mini-workshop to expand research and communication network, discuss common research directions, plant-animal interactions, landscape patterns, scaling, scenarios and validation

August 1999 (IALE conference, Snowmass, USA, 29/7 - 3/8): workshop to synthesize results on model comparisons

2000: Activity workshop: present full models of migration with both climate change and landscape heterogeneity