Francis, C.M. & Piper, S.E. 1999. Demography of tropical forest birds. In: Adams, N.J. & Slotow, R.H. (eds) Proc. 22 Int. Ornithol. Congr., Durban: 295-296. Johannesburg: BirdLife South Africa.

Many tropical birds, particularly forest species, are threatened by loss of habitat due to clearing and development of tropical areas. To face these conservation challenges, and to develop appropriate management plans, it is essential to understand their demographic patterns. Such data are also necessary to understand the evolution of life histories in the large proportion of the world’s birds that live in the tropics. We convened this symposium with the intention of reviewing the current state of knowledge of the demography of tropical birds and to consider directions for further research in this field. As it turned out, the symposium focused on only a few components of demography, namely survival rates, due partly to some controversy surrounding estimates of survival rates for tropical birds, as well as the critical importance of these parameters for developing valid population models.

It is generally recognised that demographic parameters of tropical and south temperate forest passerines tend to differ from those of ecologically similar species in north temperate zones. Clutch sizes of tropical species are typically much smaller than those of similar-sized temperate species (Moreau, R.E. 1944. Ibis 86:286-347; Murray, B.G., Jr. 1985. Ornithological Monographs 36:505-519; Skutch, A.F. 1985. Ornithological Monographs 36: 575-594). It had also been widely assumed that survival rates of tropical birds were much higher than those of temperate species, but this assumption was recently challenged by Karr et al. (Karr, J.R., Nichols, J.D., Klimkiewicz, M.K. & Brawn, J.D. 1990. American Naturalist. 136:277-291). Because of the fundamental importance of survival rates for developing life history theory of tropical birds, the paper by Karr et al. (1990) prompted a number of new analyses and publications on tropical bird survival rates (Faaborg, J. & Arendt, W. 1995. Auk 112: 503-507; Johnston, J. P., Peach, W. J., Gregory, R. D. & White, S. A. 1997. American Naturalist 150: 771-789; Ricklefs, R. E. 1997. Ecological Monographs 67: 23-43). The papers by Brawn et al., and Francis et al. in this symposium both address this issue in more detail.

One conclusion from these studies is that survival rates of tropical birds are still not adequately understood, and further data and analyses are required. Most of the more intensive studies, based upon observations of colour-marked individuals, have yielded data on only certain ages or sexes in the population (e.g. adult males), while others, especially capture-recapture studies based upon mist-netting, have produced estimates for mixtures of birds differing in age, sex or breeding status. The paper by Piper and Oatley in this symposium discusses the fact that survival rates are likely to differ with breeding status (territorial adults vs. floaters), and demonstrates the consequences to demographic models of this heterogeneity, as well as the consequences of failure to model this heterogeneity for estimating survival. Some of these problems can be addressed through the use of appropriate statistical models, such as those that exclude ‘transients’ from the estimates, thus yielding a less biased estimate of survival for territorial residents. However, those models do not produce survival estimates for all segments of the population (e.g., young birds, floaters), and there is still need for further research to estimate the potential effects of heterogeneity in capture probabilities, net shyness, or emigration from the study areas on survival estimates. A challenge for the future is to develop field techniques that marry the benefits of intensive colour-ring and resighting studies with those of large-scale mist-netting studies to provide reliable parameter estimates for more than one species without incurring too heavy a burden of field work.

Another limitation of all of these studies is that the species for which survival rates have been estimated are certainly not a random sample, nor even a representative sample of the bird species in the community (for example, mist-netting studies are generally restricted to the more abundant understorey species that can be caught in mist nets). This greatly limits our ability to make statistical comparisons of survival rates among communities.

Despite these limitations, there does seem to be fairly good evidence that, on average, survival rates of birds in many areas of the tropics tend to be higher than those of ecologically similar species living in north temperate areas, although the differences may be less than had once been assumed. Also, there appears to be considerable variation among sites in the tropics, as well as among species within a site. This variation is much greater than variation in clutch size, indicating that demography of these birds cannot be explained as a simple trade-off between clutch size and survival rates (Brawn et al. this volume).

This highlights the need to expand demographic studies to estimate parameters in addition to clutch size and survival rates. Additional factors that are important for understanding the demography of tropical birds, and how they may differ from temperate species, include nesting success, numbers of nesting attempts per season, post-fledging survival, age at first breeding, and survival of pre-breeders. Unfortunately, many of these parameters are very difficult to estimate both for tropical and for temperate species.

There is also a need to go beyond estimating mean parameters across periods of several years. Most demographic parameters are likely to vary among years, especially in response to major stochastic effects such as those induced by the El Niño southern oscillation, monsoon rains, typhoons, or fires. The paper by Dettmann and Baker (Dettman, E.B. & Baker, G.B. 1998. Ostrich 69:23) in this symposium discusses the dramatic effects of a fire on the demography of a south temperate species. Measuring the effects of these types of perturbations requires large sample sizes and high resighting or recapture rates to estimate parameters with sufficient precision to measure annual variation. Demographic models must also be developed to allow for this stochastic variation, and not treat parameters such as clutch size or survival rates as fixed characters of a particular population.