The tropical frontier in avian climate impact research


Corresponding author.

The impacts of climate change on tropical biodiversity are a subject of active debate. Global reviews show that climate change is having far-reaching effects on biodiversity (Sala et al. 2000, Walther et al. 2002, Root et al. 2003, Parmesan & Yohe 2003, Parmesan 2006, Rosenzweig et al. 2007, 2008, Miller-Rushing et al. 2010), but these studies tend to focus on temperate environments, with rare mention of changes in the tropics (Laurance et al. 2011, Wormworth & Sekercioglu 2011). Of the c. 30 000 studies reviewed for the IPCC 2007 report, fewer than 1% were from the tropics (Rosenzweig et al. 2008). The lack of research on climate impacts on tropical biodiversity, combined with the perception of a small absolute magnitude of projected temperature and rainfall changes (Sala et al. 2000, but see Stainforth et al. 2005, Chen et al. 2009), has helped fuel disagreement about the vulnerability of tropical species to ongoing and projected changes. Some studies argue that the effects of climate change will be small relative to the overwhelming impacts of habitat loss (Sala et al. 2000, Sodhi et al. 2004). In contrast, several modelling analyses predict that climate change will be an important extinction driver in the tropics (Williams et al. 2003, Thomas et al. 2004, Shoo et al. 2005, Colwell et al. 2008, Sekercioglu et al. 2008, Hole et al. 2009).

Tropical birds have received less study than temperate birds despite the fact that tropical latitudes harbour the vast majority of bird species (e.g. Sodhi et al. 2006). The lack of studies makes it difficult to measure and predict the impacts of climate change relative to other extinction drivers such as habitat loss, invasive species, disease and over-exploitation (Sodhi et al. 2011). We highlight examples of innovative studies on the effects of climate change on upland tropical birds and discuss further research avenues with a focus on methods that can provide useful results with minimal investment of time and money. In addition, we stress the need for increased climate monitoring, highlight the potential for literature-based traits analyses, and briefly discuss the conservation of upland tropical birds under climate change.

Rising temperatures from climate change have been shown to cause upslope range shifts in many studies of temperate animals (e.g. Tryjanowski et al. 2005) and plants (e.g. Lenoir et al. 2008) but fewer studies have documented altitudinal range shifts in the tropics (Pounds et al. 1999, 2005, Seimon et al. 2007, Raxworthy et al. 2008, Chen et al. 2009, 2011). For birds, Peh (2007) compared altitudinal ranges of generalist bird species (that are likely to be little affected by habitat loss) in Southeast Asian field guides from 1975 to 2000. Of 306 species studied, 84 shifted their upper range margin upslope with a stable lower margin, seven shifted their lower margin with a stable upper margin, and just three shifted both margins. The under-representation of tropical range shifts is due to low research effort in the tropics, a preponderance of short-term studies focused on presence–absence, and the difficulty of disentangling multiple drivers of range changes, such as habitat loss, invasive species and climate change (Brook et al. 2008).

Distributional shifts in the tropics demand attention because extinctions might be avoided if suitable refuges exist, species are able to disperse and species interactions are not seriously altered (Parmesan 2006). Mid-range emissions scenarios predict that, by 2100, large areas of the lowland tropics will either experience climates hotter than currently exist anywhere on Earth, or be over 1500 km from the equivalent of the current climate (New et al. 2009). In a process called lowland biotic attrition, lowland species that are found far from cool, upland refuges will be unable to shift and extinctions may result unless species can adapt (Colwell et al. 2008, Wright et al. 2009). Based on the mid-range A1B emissions scenario, 92% of protected areas are predicted to become climatically unsuitable by the year 2100 and in many tropical habitats with low topographic relief, vegetation shift will have to take place the fastest to keep up with warming (Loarie et al. 2009). Upland species that have narrow altitudinal ranges may suffer from range-shift gaps where they are unable to keep up with advancing climates up mountainsides (Colwell et al. 2008). In forested areas, birds may be less affected by range-shift gaps than some plants, insects, reptiles and amphibians that are poor dispersers or are strongly philopatric. However, habitat loss may substantially constrain distributional shifts that tropical animals will need to make under climate change (Forero-Medina et al. 2011). Mountaintop extinctions of high-elevation species may result when preferred climates shift off the tops of mountains (Williams et al. 2003) and low-elevation competitors expand their distributions upslope (Jankowski et al. 2010), in a process called the ‘escalator to extinction’ (Sekercioglu 2007). Lastly, tropical species may be particularly vulnerable to climate change because they experience minimal fluctuations in annual temperature and are already near their maximum thermal tolerance (Tewksbury et al. 2008).

Approximately 10% of the world’s bird species are confined to small geographical and elevational ranges in tropical upland (≥ 500 m asl) habitats (C.H.S unpubl. data). Correlative distribution and abundance models suggest that many of these species are likely to be threatened by climate change (Jetz et al. 2007, Sekercioglu et al. 2008, Gasner et al. 2010, La Sorte & Jetz 2010a, Wormworth & Sekercioglu 2011), yet most are classified as least concern by the IUCN (Sekercioglu et al. 2008, BirdLife International 2009) because of uncertainties surrounding model predictions (Akçakaya et al. 2006). The causes of uncertainty in forecasts of climate change impacts on biodiversity are varied but, broadly speaking, uncertainty results from a lack of long-term empirical data on climate–biodiversity impacts combined with model-based uncertainty derived from biodiversity and climate modelling techniques, including a failure to incorporate biological processes (Araújo & Rahbek 2006, Heikkinen et al. 2007, Beaumont et al. 2008).

Monitoring climate change impacts

Climate change can cause compositional changes in tropical upland bird communities, but the shifting ecology of these novel communities remains to be investigated. Pounds et al. (1999) studied birds from 1979 to 1998 in a forested plot at Monteverde reserve, Costa Rica (1540 m). The authors documented the colonization of 15 low-elevation species (usually found below 1470 m) and showed that these avian community changes were correlated to decreased mist frequency from climate change. Furthermore, Pounds et al. (2005) observed that high-elevation species are declining (e.g. Resplendent Quetzal Pharomachrus mocinno) or moving upslope (e.g. Fiery-throated Hummingbird Panterpe insignis), probably in response to climate change and consequent changes in species interactions. This sort of documentation of bird community shifts from climate change is urgently needed from other tropical regions.

There are many ways to develop the pioneering work of Pounds et al. (1999, 2005). One efficient approach would be rapidly to survey bird communities along elevational gradients. In a recent project, J.B.C.H. (J.B.C.H, N.S.S, D.A.F. and B.W.B unpubl. data) recorded bird abundances with point counts and transect surveys on trails from the base to the summit of four mountains in Borneo. Abundances of 234 species were recorded at 275–4095 m in just 2 months. Abundance data are essential in climate impacts research for quantitative historical–current comparisons (Tingley & Beissinger 2009) and spatial modelling to predict potential changes in population size (Shoo et al. 2005). Most temperate studies that have been able to detect climate impacts on birds were long-term projects (reviewed in Crick 2004, Møller et al. 2010); thus, although most long-term projects are expensive and difficult to maintain, it will be important to repeat surveys at regular intervals, at least every 5 years (Magurran et al. 2010). If similar repeated, rapid surveys are done in different tropical regions, generalizations could perhaps be made as to which lowland species are likely to invade highland areas and which range-restricted highland endemic species are prone to decline. Such studies would need to control for the effects of habitat loss and land use.

Information on reproductive rates is also needed to document changes in the breeding avifauna of a site and to allow quantification of reproductive fitness. Fundamental information can be efficiently collected with nest searching rapidly to improve our understanding of reproduction in upland tropical birds. For example, eight trained nest searchers located 700 nests in a Venezuelan upland tropical forest in a 4-month field season (T.E. Martin pers. comm.). Such large sample sizes allow monitoring of changes in reproductive output for many species that can be linked to changes in climate or, perhaps, competition. Video monitoring of nests can efficiently quantify baseline nest predation and brood parasitism and detect changes from invading nest predators and parasites over time, providing a clearer picture of any climate-driven change. As so few data are available, results from individual studies will be of great use but, again, efficacy will be markedly improved if studies are repeated over time (e.g. Martin 2007).

Intensive research methods such as mark-recapture studies are also needed in tropical uplands, but these are expensive, often logistically challenging and difficult to maintain, so studies should be carefully allocated to taxa and regions that are most likely to produce results that can be generalized. Long-term mark-recapture datasets are potentially important for understanding the effects of climate change on birds because they provide a statistically rigorous method for quantifying climate impacts on avian survival, enable us to measure breeding status and age distribution, allow population modelling, and enable robust inference on density and population trends (Grosbois et al. 2008). Mark-recapture studies have been undertaken on upland tropical birds (e.g. Parker et al. 2006) but long-term datasets are rare (e.g. Newmark 2006). Some of the difficulties of maintaining a long-term mark-recapture programme could be mitigated if programmes are linked to permanent research stations. As a starting point, we propose that mark-recapture programmes be established at least at one research station in each tropical region (Asian tropics, Afrotropics and Neotropics). Suitable locations for establishing these programmes include the Smithsonian’s Center for Tropical Forest Studies plots (, which are foci of long-term ecological research. Candidate sites at which baseline ecological research is already underway are La Planada, Colombia (1796–1891 m; Restrepo et al. 1999), and Doi Inthanon, Thailand (1660–1740 m; Khamyong et al. 2004). In Africa, where relevant studies on birds are fewest (Laurance et al. 2011), the Usambara Mountains, part of the Eastern Arc Mountains biodiversity hotspot, are an ideal candidate, with a long-term bird mark-recapture study that was established over two decades ago (Newmark 2006).

Wider monitoring programmes will also help to identify climate-induced shifts in avian distribution and abundance. These programmes draw on the large pool of skilled volunteer birdwatchers who can repeatedly and accurately collect occurrence data over large spatial and temporal scales. The North American Christmas bird count (La Sorte & Thompson 2007) and breeding bird atlas (Zuckerberg et al. 2009) and the British bird atlas (Thomas & Lennon 1999) have all been used to detect climate-related latitudinal shifts in bird distributions. Global monitoring schemes such as the Tropical Ecology Assessment and Monitoring Network (TEAM; and Global Observation Research Initiative in Alpine Environments (GLORIA; will also be important for comparing avian responses to climate change globally.

We know little about baseline competitive, parasitic and symbiotic interactions in tropical upland bird communities and virtually nothing about the changes to these dynamics caused by climate change. These can be profound: for example, due to upslope expansion, the cavity-nesting nest predator Keel-billed Toucan Ramphastos sulfuratus now nests alongside cavity-nesting Resplendent Quetzals at Monteverde, Costa Rica (Pounds et al. 1999), probably competing with them for cavities and preying on their eggs and young. Furthermore, the importance of abiotic (e.g. Ghalambor et al. 2006) and biotic (e.g. Price & Kirkpatrick 2009) factors in determining tropical range boundaries are still poorly understood. The only study that has tested the importance of biotic interactions in this context found that interspecific interactions are likely to be important for determining range boundaries in Monteverde (Jankowski et al. 2010). These authors also found that the mountaintop Slaty-backed Nightingale-thrush Catharus fuscater is tolerant of the middle elevation Black-headed Nightingale-thrush Catharus mexicanus, but not vice versa. This suggests that high-elevation species may be under asymmetric pressure from low-elevation species. This pattern seems to fit into taxon cycle theory, by which endemics are squeezed by generalists into higher elevations (Ricklefs & Bermingham 2002). Asymmetric competition from low-elevation generalists is likely to interact with other extinction pressures on high-elevation species under climate change. However, Jankowski et al. (2010) observed asymmetric competition in just one of two genera studied, and these results come from a single field site, so generalizations are difficult. While Jankowski et al. (2010) made progress on baseline interspecific interactions in upland tropical birds, avian interactions under climate change and their effects on tropical ecosystem function remain to be investigated (Mooney et al. 2009).

Other topical research directions

An additional, little-explored research strategy would be to combine elevational range data with species trait information from the literature to evaluate whether traits can predict colonization success of low-elevation species or extirpation vulnerability in highland residents. Results from this kind of analysis could help direct monitoring to species that may be most threatened by climate change or most likely to become ‘problem’ species. Previous work suggests that range size, specialization, mobility and local abundance are related to resistance to extinction (Kattan 1992, Sekercioglu 2011), and elevational range, dispersal ability, reproductive output, migratory behaviour and climatic niche breadth are likely to influence a species’ ability to respond to climate change (Isaac et al. 2009, Laurance et al. 2011). Species traits analyses could be readily implemented with existing data and would yield results that could be compared across tropical regions.

Accurately determining the relationship between key climate variables and species abundance will also depend on substantially increasing the collection of site-specific, long-term climate data. In tropical uplands, interpolated spatial climate layers are often impacted by poor spatial and temporal coverage of weather stations (Raxworthy et al. 2008), and steep topography where climates change rapidly over small horizontal distances. Automated portable weather stations that are established and carefully maintained at long-term study sites will improve the precision and accuracy of present-day climate data and provide scope for downscaling future climate projections to ecologically relevant spatial scales (≤ 5 km). Furthermore, improved weather station coverage will strengthen biodiversity–climate impact studies that rely on correlative approaches such as range-shift analyses, species distribution modelling and mark-recapture-derived survival analyses. In addition, spatial models that incorporate fine-scale climate data from portable weather stations can delineate key cool refuges and prioritize protection and reforestation in light of future range shifts (Shoo et al. 2011).

Conservation planning

Greater information from the tropics could be used to inform conservation status evaluations and active adaptive management programmes. Although uncertainties surrounding models of climate–biodiversity impacts have so far precluded most conservation status assessments from including climate change (Akçakaya et al. 2006), combining advanced modelling techniques with new empirical data should dramatically improve the precision of predictions, and eventually allow conservation status evaluations in light of climate change. For example, coupled population and distribution models (Brook et al. 2009, Fordham et al. in press) and mechanistic process-based models (La Sorte & Jetz 2010b) show promise for substantially reducing uncertainty, but neither approach has been applied to tropical birds.

New data should be integrated rapidly into active adaptive management plans to increase our chances of mitigating extinctions and test management hypotheses (Wilhere 2002). For example, results could be used to design species-specific conservation programmes for critically threatened species or ‘hotspot’ habitats. Species traits analyses and removal experiments can be used to identify potential problem colonists and cautiously make predictions for other regions. Once altitudinal movements from climate change are better understood, models can be used to identify potential refuges (usually nearby higher elevation sites) and management action can be adjusted accordingly (Shoo et al. 2011). At a broader scale, systematic reserve planning can be used to combine new empirical data with spatial models (Hole et al. 2009) to design optimally connected networks of protected areas that maintain suitable climate space and encourage dispersal. Overall, management under climate change will have to be dynamic and adaptive, with ever-changing strategies and biodiversity goals, as novel communities emerge and species are lost (Manning et al. 2009).


M. Breed, W. Hochachka and two anonymous reviewers provided helpful comments on the manuscript. J.B.C.H. is funded by the Loke Wan Tho Memorial Foundation and an EIPR scholarship at the University of Adelaide.