S41. Summary: Avian dispersal in fragmented habitats

Desrochers et al. present a review of the many techniques that have been used to study dispersal patterns, particularly in relation to landscape structure (paper presented by S. Hannon). These methods range from pattern studies of patch occupancy and mark-recapture methods to direct observations of movements. Desrochers and colleagues introduce a few novel and promising experimental approaches 'forcing' individuals to seek their way through a particular set of landscape elements. One method is translocation of individuals, another – without the necessity of capturing birds – is to lure them to a playback source. The paper by Foppen & Graveland, on the other hand, focuses on one technique in particular: the estimation of dispersal flow rates by means of mark/recapture (or resighting) efforts. Their method combines recently developed models for mark/recapture analysis (based on SURGE) with the older Jackson method, a simple technique which allows one to infer the magnitude of dispersal flow between landscape cells from the increase in local recovery rate as cells are combined in larger units.

The three remaining papers focus on the causes and consequences of dispersal as well as the possible impacts of fragmentation, all of them based on mark-recapture studies, often with detailed knowledge on individual life histories. One of the most often cited fitness consequences of low dispersal is inbreeding depression. Daniels and Walters (paper presented by J. Walters) review the evidence and conclude that although inbreeding depression effects appear to be almost ubiquitous among those bird species that have been studied in this respect, the evidence for inbreeding avoidance through dispersal is more equivocal. The intuitively appealing assumption that the likelihood of mating with a relative decreases with dispersal distance does not appear to hold in some species, at least not within the common dispersal range. Nevertheless there is increasing evidence that dispersal behaviour can be affected by the spatial distribution of close kin. Fitzpatrick et al. present results from their long-term study on Florida Scrub-Jays and present a convincing case that survival costs of dispersal are associated with the cost of establishing a new territory. Their mark-recapture data in different landscapes show that in a highly fragmented landscape jays travel longer distances, rather than shorter, which results in a similar 'social' dispersal distance i.e. in number of territories. This does not mean that fragmentation is not important, since patterns of occupation show a clear distance limit to the colonization of isolated patches.

Finally, the paper by Lens et al. touches on the evolutionary aspects of habitat fragmentation and dispersal by comparing effects of fragmentation on forest birds (and a few other organisms) in two different parts of the world: a traditionally fragmented landscape in Belgium and recently fragmented wooded hilltops in Kenya. Kenya bird populations show increased levels of fluctuating asymmetry, skewed sex ratios and low interpopulation dispersal rates. In the Belgian study populations fragmentation is associated with lower population densities in many species, but otherwise population parameters are relatively unaffected. Interpatch dispersal rates are fairly high even though dispersers may suffer heavier losses compared to nonfragmented sites.

In the general discussion at the end of the symposium a number of additional issues were raised that were not discussed in depth in any of the papers but are highly relevant to the study of dispersal in relation to fragmentation. A. van Noordwijk raised the issue of considering fitness consequences in the appropriate time-scale, since extinction/colonization events may select for dispersal even if short-term selection favours philopatry. Some studies (including the Florida Scrub Jay) suggest the existence of different dispersal strategies, but the degree to which dispersal behaviour may have a heritable component is unknown for most birds (Fitzpatrick, Grubb). It is noteworthy that at least two talks at the conference reported on highly heritable variation in potentially related traits, viz. partial migration (P. Berthold, plenary lecture) and exploratory behaviour associated with consequences for social dominance (P. Drent and C. Marchetti, symposium 16).

J.N. Smith and others questioned whether estimates of dispersal/philopatry are biased by landscape pattern. This is an important problem but perhaps not applicable to the studies presented here, since they are based on constant mistnetting or observer effort (Matthysen). The problem of including failed dispersers in data analysis and interpretation was mentioned (D. Heg) and acknowledged. This is not so much a problem in estimating dispersal flows, where only successful dispersal matters, but it does matter for the study of fitness consequences of dispersal decisions. Radiotracking, with its limits in applicability, can offer a solution, and for this problem, in particular, experiments manipulating dispersal decisions of known individuals may be necessary (Matthysen).

All five papers present different approaches and emphasise different aspects of the study of dispersal and habitat fragmentation, and thereby provide a representative – though far from complete – view of the current state of knowledge in this field. All of them have a strong empirical orientation, which to some degree reflects a conscious bias by the convenors at the planning stage of this symposium. However, we hope that it also reflects a growing awareness of the importance of basic data to match the fast-growing body of theoretical and conceptual work. There is an urgent need for a deeper understanding of what we like to call 'the natural history of dispersal' which includes dispersal patterns and their variation in time and space, the underlying behavioural mechanisms and the ecological consequences.

The need for advanced empirical research calls for development or application of new methods, and fortunately there seems to be no shortage of those in dispersal research, including molecular fingerprinting and sexing techniques, demographic modelling, analysis of fluctuating asymmetry, and experimental methods to study individual movements at small spatial scales. Whether the last technique can be 'scaled up' and extrapolated to proper dispersal behaviour remains to be investigated further.

As a general conclusion, there is no doubt now that habitat fragmentation influences population viability in general and dispersal patterns in particular. However, much remains to be done to find out how, why and to what extent these effects operate. Some important points for future research were already mentioned in the general discussion. A particularly challenging question is whether habitat fragmentation alters the cost/benefit balance of particular dispersal decisions. For instance, a more patchily distributed habitat may alter the distribution of close kin, the options for territory establishment, or the opportunities for mate and site selection. Explicit consideration of the social and genetic structure of the population will be an obvious requirement.