There is now clear evidence that the relatively modest climatic changes over the past century have already had significant impacts on the abundance, distribution, phenology and physiology of a wide range of species. Recent reviews have documented many instances of shifts in species distributions toward the poles or upward in elevation, and progressively earlier life cycle events such as flowering, reproduction and migration (Hughes 2000; McCarty 2001; Walther et al. 2002; Parmesan & Yohe 2003; Root et al. 2003). Australian examples are, however, notably lacking from these compilations. As climate changes in Australia are consistent with global trends, the lack of documented impacts is presumably not because Australian species have been unaffected, but rather because long-term datasets in which such trends could be detected are scarce (Westoby 1991). The following list of examples of species and ecosystems where there is evidence of recent, climate-associated change, is therefore modest compared to those compiled in North America and Europe and in many cases, the relative role of climate change versus other factors is poorly understood.
Australian vegetation has been profoundly altered during the last 200 years of European colonization. Century-scale changes in vegetation have been documented in various vegetation types in Australia, including open forests, grassy woodlands, tussock grasslands and rainforests, but few studies have involved repeated sampling at the same sites (reviewed by Lunt 2002). Although the potential role of recent climate and atmospheric change has been noted, most changes in composition and age structure have been attributed to grazing or to changes in fire regimes following European settlement (e.g. Lunt 1998; Lunt 2002).
A marked increase in woody biomass at the landscape scale has been reported for a wide variety of arid and semiarid environments, as well as tropical savannas and open woodlands (Archer et al. 1995 and references therein). This phenomenon, known as ‘vegetation thickening’, has been generally viewed as an example of vegetation recovery and succession following episodic disturbance events such as drought, fire and clearing (Gifford & Howden 2001). Although not the primary cause, atmospheric change may also be involved in thickening (Archer et al. 1995).
Vegetation thickening over the last 50–100 years has been recorded in Australia in a number of environments such as semiarid woodlands (Henry et al. 2002), and eucalypt savannas (Bowman et al. 2001). Where this has occurred in grazed rangelands it is known as the ‘woody weed’ problem. The potential role of CO2 fertilization in these changes has been investigated by Berry and Roderick (2002). They described landscapes in terms of the abundance of three different functional types of leaves and estimated the proportion of natural vegetation made up of each type using climatic and satellite data. Using continental scale maps of past and present vegetation, they estimated the change in proportion of the three leaf types that has occurred as a result of increased CO2, as opposed to land use change. They concluded that increasing CO2 would have exacerbated the woody weed problem.
Changes in rainfall patterns have also been implicated in some vegetation trends. For example, a historical survey of eucalypt savannas in Litchfield National Park, NT, where less than 1% of the study area has been modified by humans, showed that forest coverage has increased from 5% to nearly 10% over ∼50 years, whereas areas of grassland have decreased from approximately 7 to 2.5% (Bowman et al. 2001). Although the cessation of Aboriginal landscape burning may be the primary cause, the period of increased rainfall in the 1970s may also have been an influence. Above average rains in eastern Australia favour the increase of woody biomass in savannas. Climatic records over the past 125 years show that there was a trend of above-average rainfall in the second half of the study period (i.e. 1965/66–1994/95), although there was considerable variation in the total amount of rainfall for any wet season. It is possible that this wetter-than-average period has favoured the increase in woody biomass by increasing supplies of groundwater accessible to tree roots (Bowman et al. 2001). Similarly, Fensham et al. in press) documented a general increase in overstorey cover in central Queensland over the second half of the 20th century, during a period of higher than average rainfall. A contrasting trend in Tasmania toward an increased incidence of drought and alterations in seasonal rainfall patterns has been implicated in eucalypt dieback (Kirkpatrick et al. 2000). A positive relationship between the magnitude of drought and eucalypt dieback was also found in north Queensland savanna (Fensham & Holman 1999).
Expansion of rainforest at the expense of eucalypt forest and grasslands in Queensland over the past two centuries has been well documented (Harrington & Sanderson 1994; Hopkins et al. 1996). In the Bunya Mountains in south-east Queensland, aerial photographs over a 40-year period (1951–1991) demonstrated that both eucalypt forest and rainforest had invaded and in some cases, completely engulfed, grasslands that occur in a range of landscapes (Fensham & Fairfax 1996). In North Queensland, comparisons of aerial photographs from the 1940s and the early 1990s indicated that rainforest had invaded large areas of wet sclerophyll eucalypt forest (Harrington & Sanderson 1994) with widespread thickening of the canopy across both vegetation types. Once again, cessation of active traditional Aboriginal land management has been suggested as the primary cause, but climatic changes cannot be discounted. Recent invasions of warm temperate rainforest species to higher elevations in northern NSW and expansion of Nothofagus into eucalypt woodland on plateaus in the Barrington Tops region have also been documented (Read & Hill 1985). At present it is unclear whether this migration is a response to recent warming or whether the vegetation is still responding to the major climate changes following the last glacial maximum.
Australia's alpine vegetation has been the subject of several long-term studies. Increases in cover of Sphagnum and decreases in shrub cover in sphagnum bogs in Kosciuszko National Park have occurred over a 32-year period (1959–1973 sites resampled in 1991, Clarke & Martin 1999). A similar trend in shrub senescence has been reported for shrublands in the Victorian Alps in plots fenced to exclude grazers since 1945 (Wahren et al. 1994). These trends have been interpreted mainly as slow recovery from grazing disturbance and postfire regeneration (Wahren et al. 1994). Other trends in the alpine zone, however, have been more confidently ascribed to recent warming trends. Encroachment by Eucalyptus pauciflora into subalpine grasslands near Mt Hotham, Victoria, has been documented by Wearne and Morgan (2001). All invading saplings were estimated to be less than or equal to 31 years old and the majority (54%) established between 1991 and 1995. Most sapling establishment (66%) occurred within 5 m of the forest–grass–land boundary and some of the recently established plants are now reproductively mature trees (1–8 m in height), suggesting an ecotonal change is underway.
Warming trends have also been implicated in agricultural changes. The average yield of wheat has increased by 0.5 tonnes per ha since 1952, representing a rise of ∼45% of the average annual yield (Nicholls 1997). Over the same period, the frequency of severe frosts decreased in eastern Australia (Stone et al. 1996) and median annual temperatures, averaged across Australia, increased by 0.58°C (1952–1992). Nicholls (1997) estimated that climate trends were responsible for 30–50% of the observed increase in wheat yields, with increases in the minimum temperature being the dominant influence, although other authors (Godden et al. 1998) have offered alternative explanations.
Some of the most compelling evidence that recent warming has already affected life cycles comes from long-term phenological monitoring of various species of plants, birds, insects and amphibians in the Northern Hemisphere (Hughes 2000; Parmesan & Yohe 2003; Root et al. 2003). Unfortunately, there are few comparable datasets in Australia, with the exception of several compilations of reproductive phenology of eucalypts in four Australian states (NSW, Queensland, Victoria and WA), collected by state forestry organizations for varying periods from 1925 to 1981 (Keatley et al. 2002). Records for four eucalypt species in Victoria, collected from 1940 to 1962, indicate that temperature accounted for greater variation in flowering commencement date than rainfall, but that over this period, no statistically significant trends in flowering date were evident (Keatley et al. 2002). One identified problem with the dataset, however, was that flowering observations took place only monthly, so it is possible that each observed flowering date may be up to 30 days later than the actual commencement date. Relationships between flowering date, temperature and rainfall for the dataset indicated that a warming of 1°C may result in earlier flowering for two of the four species, E. microcarpa and E. polyanthemos, by 41 and 43 days, respectively, and later flowering in E. leucoxylon and E. microcarpa (Keatley et al. 2002).
The landward transgression of mangroves into saltmarsh environments in the estuaries of Queensland, NSW, Victoria and SA over the past five decades is a widespread trend with saltmarsh losses ranging up to 80% (Saintilan & Williams 1999). This process is occurring in a range of geomorphic settings, and in some cases is reversing the trend of longer-term vegetation change. Although direct human disturbance is undoubtedly a factor in these trends (e.g. revegetation of areas cleared for agriculture, increases in nutrient levels and sedimentation), increases in rainfall and altered tidal regimes have also been implicated (Saintilan & Williams 1999). Along the eastern shore of the Gulf of St Vincent, SA, for example, mangroves extended at a rate of 17 m per year from 1949 to 1979 (Fotheringham 1994, cited in Saintilan & Williams 1999). One explanation for this trend is that increased average annual precipitation since 1945 in the area may have diluted salts within saltmarsh soils to the extent that mangrove colonization was enhanced (Saintilan & Williams 1999).
In some areas of the NT, dramatic expansion of some tidal creek systems has occurred since the 1940s. In the Lower Mary River system, two creeks have extended more than 4 km inland, invading freshwater wetlands (Woodroffe & Mulrennan 1993; Bayliss et al. 1997; Mulrennan & Woodroffe 1998). Rates of extension of saltwater ecosystems inland in excess of 0.5 km year−1 have been measured (Knighton et al. 1992). The tidal networks are still expanding and only in their lower reaches does the maximum extension appear to have been reached. The smallest tributaries are being eliminated as mangroves (Avicennia sp.) spread along creek boundaries and trap large quantities of fine sediment. The saltwater intrusion has had dramatic effects on the vegetation of formerly freshwater wetlands with more than 17 000 ha adversely affected and a further 35–40% of the plains immediately threatened (Mulrennan & Woodroffe 1998). These changes most likely have multiple causes, but both sea level rise and increases in rainfall may have contributed (Woodroffe & Mulrennan 1993; Bayliss et al. 1997).
One of the few long-term datasets for Australian fauna is that for the sleepy lizard, Tiliqua rugosa, collected in mid-north SA by Bull and colleagues since 1983. Over the study period 1983–1997, the last months of winter (July and August) have become warmer and drier, and the spring months when lizard activity is highest (September and October) have become wetter (Bull & Burzacott 2002). These climatic shifts have been associated with changes in the timing of lizard pairing behaviour. Specifically, pairing tended to start earlier following the warmer, drier winters of the later years of the study and persisted for longer. As part of the same study, distributions of two reptile ticks for which the sleepy lizard is a host have also been monitored. The abrupt parapatric boundary between the two species has shifted 1–2 km over the same period, with the more mesic-adapted Aponomma hydrosauri advancing into the distribution of the more arid-adapted Amblyomma limbatum. The density of ticks on lizards in regions flanking the boundary zone has also increased for Ap. hydrosauri and decreased for Am. limbatum (Bull & Burzacott 2001).
In the alpine zone, there is evidence of shifts in vertebrate ranges to higher elevations over the 30-year period to 1999. Wildlife Atlas records indicate a higher maximum altitudinal distribution for all three macropod species and for four species of feral mammals (Green & Pickering 2002). Other evidence for increasing activity by feral mammals at higher altitudes supports this trend. In the 1970s, Snowy Plains (1370 m a.s.l.) was regarded as climatically marginal for rabbits, yet during the summer of 1998–1999 the National Parks and Wildlife Service was forced to institute a rabbit control program at Perisher Valley (Green & Pickering 2002). During the period 1980–1988, Green (1988) conducted regular small-mammal trapping near the treeline on the South Ramshead with access up a spur from Dead Horse Gap (1580 m a.s.l.) without once recording evidence of horses above the gap. This route is now regularly used by horses to gain access to the alpine zone (K. Green, pers. obs.).
A trend toward earlier arrival of migratory bird species in the alpine zone in the 1980s and/or 1990s, compared with the 1970s, has also been documented (Green & Pickering 2002). For the 11 bird species for which there are sufficient data, the earliest record was in the 1990s for five species and in the 1980s for four. The particular foraging techniques and biology of the individual bird species were associated with the trends in their arrival. The species recorded as arriving earlier include three species of honeyeaters that depend on the flowering of shrubs. The Australian kestrel Falco cenchroides is largely dependent on snow-free ground for foraging. The ground-feeding flame robin, Petroica phoenicea, and Richard's pipit, Anthus novaeseelandiae, arrive early in spring and feed on insects immobilized on snow; the earlier presence of these insects is associated with sufficient warmth at their point of origin for metamorphosis and flight. Olive whistlers, Pachycephala olivacea, and striated pardalotes, Pardalotus striatus, glean active insects off shrubs and trees and movements of fan-tailed cuckoos, Cuculus flabelliformis, are attuned to the breeding timetable of their hosts. The two species that appear not to arrive earlier despite changes in snow cover over the three decades are the grey fantail, Rhipidura fuliginosa, which catches insects in flight, and the silvereye, Zosterops lateralis, which is involved in long migratory flights, the timing of which may be independent of local events (Green & Pickering 2002).
Baxter et al. (2001) have documented recent observations of the black-necked stork, Ephippiorhynchus asiaticus, in north-east South Australia, well to the south of its customary range, and suggested that monsoon-flooding in Queensland and far northern SA during in the summer–autumn of 2001 may be an explanation. Recent observations of the magpie goose, Anseranas semipalmata, a vagrant well outside its usual range, have also been noted (Baxter et al. 2001). Although annual floods down the major rivers in the Channel Country in the summers of 1999–2001 were much larger than average annual flows and were linked to a La Niña phase, they were not as great as the largest floods experienced in previous La Niña periods such as 1974–1976 and 1989–1991. No records of magpie goose or black-necked stork were reported from north-eastern SA during these previous La Niña events. Baxter et al. (2001) conclude that the intrusion of these species into the area is highly unusual and may not have occurred in the recent past.
The range of Pteropus poliocephalus, the grey-headed flying fox, has contracted south from its northern boundary by approximately 750 km since the 1930s (Tidemann 1999). Pteropus poliocephalus was a common breeding species in the Rockhampton area in 1930 but by the early 1960s the northern extent of its breeding range had contracted to Maryborough. Although satellite camps still extend to around Maryborough, most camps in this area are now dominated by P. alecto, the black flying fox, a more tropical species. Along with the southward contraction in range by P. poliocephalus, P. alecto has apparently extended its range south by a similar distance, with its southern limit progressively extending from Maryborough, to the NSW–Queensland border and then to Bowraville, NSW; its breeding range now extends to Maclean, NSW. Increases in rainfall and temperature in eastern Australia over the period have possibly favoured P. alecto, which cannot tolerate frosts (Tidemann 1999). The range shifts of both species have occurred such that the area of sympatry between them has remained much the same.
There have been some major changes in seabird breeding distribution since the late 19th century in the transition zone between tropical and temperate seabird species in the region between the Houtman Abrolhos and the Naturaliste and Leeuwin Capes, off the coast of WA (Dunlop 2001). At least eight species have formed new breeding locations well to the south of their historical range and/or have seen marked population increases at their more southerly colonies. Such changes have occurred in the wedge-tailed shearwater (Puffinus pacificus), bridled tern (Sterna anaethetus), roseate tern (Sterna dougalli), crested tern (Sterna bergii) and brown noddy (Anous stolidus). Some of the shifts began as early as the 1920s (bridled tern), others in the 1950s and 60s (roseate tern, red-tailed tropicbird), whereas others did not begin until the last decade of the 20th century (brown noddy, sooty tern). The rate of establishment and/or growth of new colonies seems to have accelerated since the early 1980s. Long-term monitoring studies at three wedge-tailed shearwater colonies suggest that their fisheries decline during El Niño conditions. Brown noddies and sooty terns also fare poorly on the Abrolhos during periods when the Leeuwin Current is weak. During the 1996–1997 El Niño event there was almost complete breeding failure in hundreds of thousands of brown and lesser noddies, sooty terns and roseate terns on Pelsaert Island due to an ENSO-triggered collapse in the pelagic seabird fisheries. Although the records are patchy, there is also some evidence that the prospecting behaviour by the birds that precedes the colonization of new, more southerly breeding sites, tends to occur in association with the major El Niño events. The El Niño phase of the Southern Oscillation increased in frequency with every couple of decades of the 1900s such that what was once an unusual climatic pattern has become more common in the last two decades of the century, with major events in 1982–1983, 1987–1988, 1991–1994 and 1996–1997. Because the behaviour of the Leeuwin Current is strongly influenced by ENSO, it is likely that this is the ultimate cause of the shifts in the seabird fisheries and changing population dynamics (Dunlop 2001).
Since 1980, the Australasian gannet (Morus serrator) population has increased threefold in Australian waters, from 6600 breeding pairs to approximately 20 000 pairs in 1999–2000, a rate of 6% per year (Bunce et al. 2002). In colonies where nesting space is not limiting, the breeding population has expanded at rates as high as 24% per year. Bunce et al. (2002) suggest that the population increase may be associated with the increased ENSO activity over this time because increased upwellings of nutrient-rich cold subantarctic waters during ENSO events are positively correlated with increases in several commercially important fish stocks. A gradual long-term warming trend in Bass Strait and waters off south-eastern Australia may also have positively affected the distribution and local availability of pilchards and other prey species, as has been shown in other parts of the world (Bunce et al. 2002). Although these correlations are suggestive of a climate influence on gannet populations, it is possible that changes in the fishing practices of several major commercial fisheries in south-eastern Australia, resulting in an increase in discarded bycatch, may also be important (Bunce et al. 2002).
Since the late 1970s there has been a global increase in the number and scale of coral-bleaching events and the extent, timing and severity of many such events have been correlated with warmer than normal seawater temperatures (Jones et al. 1997; Lough 2000). In 1998, tropical sea surface temperatures were the highest on record, topping off a 50-year trend for some tropical oceans (Reaser et al. 2000). In the same year, coral reefs around the world suffered the most extensive and severe bleaching on record. The mortalities that followed these events were higher than any in the previous 3000 years (Aronson et al. 2002). The geographic extent, increasing frequency, and regional severity of mass bleaching events are an apparent result of a steadily rising baseline of marine temperatures, combined with regionally specific El Niño and La Niña events (Hoegh-Guldberg 1999; Lough 2000).
One of the best records of recurrent bleaching events comes from the inshore fringing reefs of Magnetic Island on the Great Barrier Reef, where bleaching has been observed in the summers of 1979–1980, 1981–1982, 1986–1987, 1991–1992 and 1993–1994 (Jones et al. 1997). Average daily seawater temperatures exceeded 31°C for 14 days and 31.5°C for 2 days during the bleaching event of 1991–1992 and exceeded 31°C for 10 days and 31.5°C for 2 days during the 1993–1994 event.
A severe and widespread bleaching on the Great Barrier Reef occurred from February to April, 1998 with inshore reefs being the worst affected (Berkelmans & Oliver 1999). Although Australian reefs were less affected than many elsewhere (3% loss compared with 46% in the Indian Ocean; Wilkinson 2000) damage was nonetheless severe at many sites. At the worst affected sites, most staghorn and other fast-growing corals were killed, whereas the very old corals had high rates of survival. The level of thermal stress at the majority of bleaching sites was unmatched in the period 1903–1999 (Lough 2000). Reefs elsewhere around Australia were similarly affected. Lowered seawater salinity as a result of river flooding between Ayr and Cooktown early in 1998 probably exacerbated the effects of warming on the inshore reefs (Berkelmans & Oliver 1999). At Scott Reef off the north-west coast, most corals died to a depth of 30 m and have since recovered only slightly (Sweatman et al. 2002).
Extensive coral bleaching re-occurred in the summer of 2001–2002 and the Great Barrier Reef was again subject to a complex mosaic of relatively hotter and cooler areas. Bleaching was more extensive than in 1998 and the inshore reefs were again the most affected. In the cooler areas no damage was found but significant coral mortality was seen in the hottest patches (T. J. Done, unpubl. obs., Sweatman et al. 2002). Surprisingly, the ubiquitous hard corals previously thought to be the most sensitive (e.g. family Pocilloporidae) survived relatively well whereas others (Acroporidae and Faviidae) suffered significant injury and mortality (T. J. Done, cited in Sweatman et al. 2002)
Evidence of warming oceans also comes from examination of annual variation in the density of calcium carbonate (CaCO3) skeletons in some massive coral species such as Porites (Lough & Barnes 1997; Lough 2000). Some of these corals live several hundred years and can be used retrospectively to monitor coral growth in an analogous way to the study of tree-rings. For each 1°C of temperature increase, calcification increases by ∼0.3 g cm−2 year−1 and linear extension of the coral increases by ∼3 mm year−1. Average calcification rates of Porites measured in more than 200 small coral colonies from 29 reefs along the Great Barrier Reef show decreases from north to south as the average annual seawater temperature decreases. When examined over 50-year periods, a more than two-centuries long record shows constant rates of calcification until the most recent period, when calcification significantly increases by ∼4%, matching the observed rise in seawater temperatures (Lough 2000).