Widespread retreat of coastal habitat is likely at warming levels above 1.5 °C

Several coastal ecosystems-most notably mangroves and tidal marshes-exhibit biogenic feedbacks that are facilitating adjustment to relative sea-level rise (RSLR), including the sequestration of carbon and the trapping of mineral sediment1. The stability of reef-top habitats under RSLR is similarly linked to reef-derived sediment accumulation and the vertical accretion of protective coral reefs2. The persistence of these ecosystems under high rates of RSLR is contested3. Here we show that the probability of vertical adjustment to RSLR inferred from palaeo-stratigraphic observations aligns with contemporary in situ survey measurements. A deficit between tidal marsh and mangrove adjustment and RSLR is likely at 4 mm yr-1 and highly likely at 7 mm yr-1 of RSLR. As rates of RSLR exceed 7 mm yr-1, the probability that reef islands destabilize through increased shoreline erosion and wave over-topping increases. Increased global warming from 1.5 °C to 2.0 °C would double the area of mapped tidal marsh exposed to 4 mm yr-1 of RSLR by between 2080 and 2100. With 3 °C of warming, nearly all the world's mangrove forests and coral reef islands and almost 40% of mapped tidal marshes are estimated to be exposed to RSLR of at least 7 mm yr-1. Meeting the Paris agreement targets would minimize disruption to coastal ecosystems.

Termite sensitivity to temperature affects global wood decay rates

Deadwood is a large global carbon store with its store size partially determined by biotic decay. Microbial wood decay rates are known to respond to changing temperature and precipitation. Termites are also important decomposers in the tropics but are less well studied. An understanding of their climate sensitivities is needed to estimate climate change effects on wood carbon pools. Using data from 133 sites spanning six continents, we found that termite wood discovery and consumption were highly sensitive to temperature (with decay increasing >6.8 times per 10°C increase in temperature)-even more so than microbes. Termite decay effects were greatest in tropical seasonal forests, tropical savannas, and subtropical deserts. With tropicalization (i.e., warming shifts to tropical climates), termite wood decay will likely increase as termites access more of Earth's surface.

Tallo: A global tree allometry and crown architecture database

Data capturing multiple axes of tree size and shape, such as a tree's stem diameter, height and crown size, underpin a wide range of ecological research-from developing and testing theory on forest structure and dynamics, to estimating forest carbon stocks and their uncertainties, and integrating remote sensing imagery into forest monitoring programmes. However, these data can be surprisingly hard to come by, particularly for certain regions of the world and for specific taxonomic groups, posing a real barrier to progress in these fields. To overcome this challenge, we developed the Tallo database, a collection of 498,838 georeferenced and taxonomically standardized records of individual trees for which stem diameter, height and/or crown radius have been measured. These data were collected at 61,856 globally distributed sites, spanning all major forested and non-forested biomes. The majority of trees in the database are identified to species (88%), and collectively Tallo includes data for 5163 species distributed across 1453 genera and 187 plant families. The database is publicly archived under a CC-BY 4.0 licence and can be access from: https://doi.org/10.5281/zenodo.6637599. To demonstrate its value, here we present three case studies that highlight how the Tallo database can be used to address a range of theoretical and applied questions in ecology-from testing the predictions of metabolic scaling theory, to exploring the limits of tree allometric plasticity along environmental gradients and modelling global variation in maximum attainable tree height. In doing so, we provide a key resource for field ecologists, remote sensing researchers and the modelling community working together to better understand the role that trees play in regulating the terrestrial carbon cycle.

Bridge to the future: Important lessons from 20 years of ecosystem observations made by the OzFlux network

In 2020, the Australian and New Zealand flux research and monitoring network, OzFlux, celebrated its 20th anniversary by reflecting on the lessons learned through two decades of ecosystem studies on global change biology. OzFlux is a network not only for ecosystem researchers, but also for those 'next users' of the knowledge, information and data that such networks provide. Here, we focus on eight lessons across topics of climate change and variability, disturbance and resilience, drought and heat stress and synergies with remote sensing and modelling. In distilling the key lessons learned, we also identify where further research is needed to fill knowledge gaps and improve the utility and relevance of the outputs from OzFlux. Extreme climate variability across Australia and New Zealand (droughts and flooding rains) provides a natural laboratory for a global understanding of ecosystems in this time of accelerating climate change. As evidence of worsening global fire risk emerges, the natural ability of these ecosystems to recover from disturbances, such as fire and cyclones, provides lessons on adaptation and resilience to disturbance. Drought and heatwaves are common occurrences across large parts of the region and can tip an ecosystem's carbon budget from a net CO2 sink to a net CO2 source. Despite such responses to stress, ecosystems at OzFlux sites show their resilience to climate variability by rapidly pivoting back to a strong carbon sink upon the return of favourable conditions. Located in under-represented areas, OzFlux data have the potential for reducing uncertainties in global remote sensing products, and these data provide several opportunities to develop new theories and improve our ecosystem models. The accumulated impacts of these lessons over the last 20 years highlights the value of long-term flux observations for natural and managed systems. A future vision for OzFlux includes ongoing and newly developed synergies with ecophysiologists, ecologists, geologists, remote sensors and modellers.

Nitrogen concentration and physical properties are key drivers of woody tissue respiration

BACKGROUND AND AIMS

Despite the critical role of woody tissues in determining net carbon exchange of terrestrial ecosystems, relatively little is known regarding the drivers of sapwood and bark respiration.

METHODS

Using one of the most comprehensive wood respiration datasets to date (82 species from Australian rainforest, savanna and temperate forest), we quantified relationships between tissue respiration rates (Rd) measured in vitro (i.e. 'respiration potential') and physical properties of bark and sapwood, and nitrogen concentration (Nmass) of leaves, sapwood and bark.

KEY RESULTS

Across all sites, tissue density and thickness explained similar, and in some cases more, variation in bark and sapwood Rd than did Nmass. Higher density bark and sapwood tissues had lower Rd for a given Nmass than lower density tissues. Rd-Nmass slopes were less steep in thicker compared with thinner-barked species and less steep in sapwood than in bark. Including the interactive effects of Nmass, density and thickness significantly increased the explanatory power for bark and sapwood respiration in branches. Among these models, Nmass contributed more to explanatory power in trunks than in branches, and in sapwood than in bark. Our findings were largely consistent across sites, which varied in their climate, soils and dominant vegetation type, suggesting generality in the observed trait relationships. Compared with a global compilation of leaf, stem and root data, Australian species showed generally lower Rd and Nmass, and less steep Rd-Nmass relationships.

CONCLUSIONS

To the best of our knowledge, this is the first study to report control of respiration-nitrogen relationships by physical properties of tissues, and one of few to report respiration-nitrogen relationships in bark and sapwood. Together, our findings indicate a potential path towards improving current estimates of autotrophic respiration by integrating variation across distinct plant tissues.

Gross primary productivity and water use efficiency are increasing in a high rainfall tropical savanna

Despite their size and contribution to the global carbon cycle, we have limited understanding of tropical savannas and their current trajectory with climate change and anthropogenic pressures. Here we examined interannual variability and externally forced long-term changes in carbon and water exchange from a high rainfall savanna site in the seasonal tropics of north Australia. We used an 18-year flux data time series (2001-2019) to detect trends and drivers of fluxes of carbon and water. Significant positive trends in gross primary productivity (GPP, 15.4 g C m²  year^-2 ), ecosystem respiration (R_eco , 8.0 g C m²  year^-2 ), net ecosystem productivity (NEE, 7.4 g C m²  year^-2 ) and ecosystem water use efficiency (WUE, 0.0077 g C kg H₂ O^-1  year^-1 ) were computed. There was a weaker, non-significant trend in latent energy exchange (LE, 0.34 W m^-2  year^-1 ). Rainfall from a nearby site increased statistically over a 45-year period during the observation period. To examine the dominant drivers of changes in GPP and WUE, we used a random forest approach and a terrestrial biosphere model to conduct an attribution experiment. Radiant energy was the dominant driver of wet season fluxes, whereas soil water content dominated dry season fluxes. The model attribution suggested that [CO₂ ], precipitation and T_air accounting for 90% of the modelled trend in GPP and WUE. Positive trends in fluxes were largest in the dry season implying tree components were a larger contributor than the grassy understorey. Fluxes and environmental drivers were not significant during the wet season, the period when grasses are active. The site is potentially still recovering from a cyclone 45 years ago and regrowth from this event may also be contributing to the observed trends in sequestration, highlighting the need to understand fluxes and their drivers from sub-diurnal to decadal scales.

Influence of the 2015-2016 El Niño on the record-breaking mangrove dieback along northern Australia coast

This study investigates the underlying climate processes behind the largest recorded mangrove dieback event along the Gulf of Carpentaria coast in northern Australia in late 2015. Using satellite-derived fractional canopy cover (FCC), variation of the mangrove canopies during recent decades are studied, including a severe dieback during 2015–2016. The relationship between mangrove FCC and climate conditions is examined with a focus on the possible role of the 2015–2016 El Niño in altering favorable conditions sustaining the mangroves. The mangrove FCC is shown to be coherent with the low-frequency component of sea level height (SLH) variation related to the El Niño Southern Oscillation (ENSO) cycle in the equatorial Pacific. The SLH drop associated with the 2015–2016 El Niño is identified to be the crucial factor leading to the dieback event. A stronger SLH drop occurred during austral autumn and winter, when the SLH anomalies were about 12% stronger than the previous very strong El Niño events. The persistent SLH drop occurred in the dry season of the year when SLH was seasonally at its lowest, so potentially exposed the mangroves to unprecedented hostile conditions. The influence of other key climate factors is also discussed, and a multiple linear regression model is developed to understand the combined role of the important climate variables on the mangrove FCC variation.

Living on the edge: A continental-scale assessment of forest vulnerability to drought

Globally, forests are facing an increasing risk of mass tree mortality events associated with extreme droughts and higher temperatures. Hydraulic dysfunction is considered a key mechanism of drought-triggered dieback. By leveraging the climate breadth of the Australian landscape and a national network of research sites (Terrestrial Ecosystem Research Network), we conducted a continental-scale study of physiological and hydraulic traits of 33 native tree species from contrasting environments to disentangle the complexities of plant response to drought across communities. We found strong relationships between key plant hydraulic traits and site aridity. Leaf turgor loss point and xylem embolism resistance were correlated with minimum water potential experienced by each species. Across the data set, there was a strong coordination between hydraulic traits, including those linked to hydraulic safety, stomatal regulation and the cost of carbon investment into woody tissue. These results illustrate that aridity has acted as a strong selective pressure, shaping hydraulic traits of tree species across the Australian landscape. Hydraulic safety margins were constrained across sites, with species from wetter sites tending to have smaller safety margin compared with species at drier sites, suggesting trees are operating close to their hydraulic thresholds and forest biomes across the spectrum may be susceptible to shifts in climate that result in the intensification of drought.

Impact of an extreme monsoon on CO2 and CH4 fluxes from mangrove soils of the Ayeyarwady Delta, Myanmar

Mangrove ecosystems can be both significant sources and sinks of greenhouse gases (GHGs). Understanding variability in flux and the key factors controlling emissions in these ecosystems are therefore important in the context of accounting for GHG emissions. The current study is the first to quantify GHG emissions using static chamber measurements from soils in disused aquaculture ponds, planted mangroves, and mature mangroves from the Ayeyarwady Delta, Myanmar. Soil properties, biomass and estimated net primary productivity were also assessed. Field assessments were conducted at the same sites during the middle of the dry season in February and end of the wet season in October 2019. Rates of soil CO₂ efflux were among the highest yet recorded from mangrove ecosystems, with CO₂ efflux from the 8 year old site reaching 86.8 ± 17 Mg CO₂ ha^-1 yr^-1 during February, an average of 862% more than all other sites assessed during this period. In October, all sites had significant rates of soil CO₂ efflux, with rates ranging from 31.9 ± 4.4 Mg CO₂ ha^-1 yr^-1 in a disused pond to 118.9 ± 24.3 Mg CO₂ ha^-1 yr^-1 in the 8 year old site. High soil CO₂ efflux from the 8 year old site in February is most likely attributable to high rates of primary production and belowground carbon allocation. Elevated CO₂ efflux from all sites during October was likely associated with the extreme 2019 South Asian monsoon season which lowered soil pore salinity and deposited new alluvium, stimulating both autotrophic and heterotrophic activity. Methane efflux increased significantly (50-400%) during the wet season from all sites with mangrove cover, although was a small overall component of soil GHG effluxes during both measurement periods. Our results highlight the critical importance of assessing GHG flux in-situ in order to quantify variability in carbon dynamics over time.

Land transformation in tropical savannas preferentially decomposes newly added biomass, whether C3 or C4 derived

As tropical savannas are undergoing rapid conversion to other land uses, native C₃ -C₄ vegetation mixtures are often transformed to C₃ - or C₄ -dominant systems, resulting in poorly understood changes to the soil carbon (C) cycle. Conventional models of the soil C cycle are based on assumptions that more labile components of the heterogenous soil organic C (SOC) pool decompose at faster rates. Meanwhile, previous work has suggested that the C₄ -derived component of SOC is more labile than C₃ -derived SOC. Here we report on long-term (18 months) soil incubations from native and transformed tropical savannas of northern Australia. We test the hypothesis that, regardless of the type of land conversion, the C₄ component of SOC will be preferentially decomposed. We measured changes in the SOC and pyrogenic carbon (PyC) pools, as well as the carbon isotope composition of SOC, PyC and respired CO₂ , from 63 soil cores collected intact from different land use change scenarios. Our results show that land use change had no consistent effect on the size of the SOC pool, but strong effects on SOC decomposition rates, with slower decomposition rates at C₄ -invaded sites. While we confirm that native savanna soils preferentially decomposed C₄ -derived SOC, we also show that transformed savanna soils preferentially decomposed the newly added pool of labile SOC, regardless of whether it was C₄ -derived (grass) or C₃ -derived (forestry) biomass. Furthermore, we provide evidence that in these fire-prone landscapes, the nature of the PyC pool can shed light on past vegetation composition: while the PyC pool in C₄ -dominant sites was mainly derived from C₃ biomass, PyC in C3-dominant sites and native savannas was mainly derived from C₄ biomass. We develop a framework to systematically assess the effects of recent land use change vs. prior vegetation composition.

Net landscape carbon balance of a tropical savanna: Relative importance of fire and aquatic export in offsetting terrestrial production

The magnitude of the terrestrial carbon (C) sink may be overestimated globally due to the difficulty of accounting for all C losses across heterogeneous landscapes. More complete assessments of net landscape C balances (NLCB) are needed that integrate both emissions by fire and transfer to aquatic systems, two key loss pathways of terrestrial C. These pathways can be particularly significant in the wet-dry tropics, where fire plays a fundamental part in ecosystems and where intense rainfall and seasonal flooding can result in considerable aquatic C export (ΣF_aq ). Here, we determined the NLCB of a lowland catchment (~140 km² ) in tropical Australia over 2 years by evaluating net terrestrial productivity (NEP), fire-related C emissions and ΣF_aq (comprising both downstream transport and gaseous evasion) for the two main landscape components, that is, savanna woodland and seasonal wetlands. We found that the catchment was a large C sink (NLCB 334 Mg C km^-2  year^-1 ), and that savanna and wetland areas contributed 84% and 16% to this sink, respectively. Annually, fire emissions (-56 Mg C km^-2  year^-1 ) and ΣF_aq (-28 Mg C km^-2  year^-1 ) reduced NEP by 13% and 7%, respectively. Savanna burning shifted the catchment to a net C source for several months during the dry season, while ΣF_aq significantly offset NEP during the wet season, with a disproportionate contribution by single major monsoonal events-up to 39% of annual ΣF_aq was exported in one event. We hypothesize that wetter and hotter conditions in the wet-dry tropics in the future will increase ΣF_aq and fire emissions, potentially further reducing the current C sink in the region. More long-term studies are needed to upscale this first NLCB estimate to less productive, yet hydrologically dynamic regions of the wet-dry tropics where our result indicating a significant C sink may not hold.

Effect of elevated magnesium sulfate on two riparian tree species potentially impacted by mine site contamination

Globally, mining activities have been responsible for the contamination of soils, surface water and groundwater. Following mine closure, a key issue is the management of leachate from waste rock accumulated during the lifetime of the mine. At Ranger Uranium Mine in northern Australia, magnesium sulfate (MgSO₄) leaching from waste rock has been identified as a potentially significant surface and groundwater contaminant which may have adverse affects on catchment biota. The primary objective of this study was to determine the effect of elevated levels of MgSO₄ on two riparian trees; Melaleuca viridiflora and Alphitonia excelsa. We found that tolerance to MgSO₄ was species-specific. M. viridiflora was tolerant to high concentrations of MgSO₄ (15,300 mg l^-1), with foliar concentrations of ions suggesting plants regulate uptake. In contrast, A. excelsa was sensitive to elevated concentrations of MgSO₄ (960 mg l^-1), exhibiting reduced plant vigour and growth. This information improves our understanding of the toxicity of MgSO₄ as a mine contaminant and highlights the need for rehabililitation planning to mitigate impacts on some tree species of this region.

Effect of land-use and land-cover change on mangrove blue carbon: A systematic review

Mangroves shift from carbon sinks to sources when affected by anthropogenic land-use and land-cover change (LULCC). Yet, the magnitude and temporal scale of these impacts are largely unknown. We undertook a systematic review to examine the influence of LULCC on mangrove carbon stocks and soil greenhouse gas (GHG) effluxes. A search of 478 data points from the peer-reviewed literature revealed a substantial reduction of biomass (82% ± 35%) and soil (54% ± 13%) carbon stocks due to LULCC. The relative loss depended on LULCC type, time since LULCC and geographical and climatic conditions of sites. We also observed that the loss of soil carbon stocks was linked to the decreased soil carbon content and increased soil bulk density over the first 100 cm depth. We found no significant effect of LULCC on soil GHG effluxes. Regeneration efforts (i.e. restoration, rehabilitation and afforestation) led to biomass recovery after ~40 years. However, we found no clear patterns of mangrove soil carbon stock re-establishment following biomass recovery. Our findings suggest that regeneration may help restore carbon stocks back to pre-disturbed levels over decadal to century time scales only, with a faster rate for biomass recovery than for soil carbon stocks. Therefore, improved mangrove ecosystem management by preventing further LULCC and promoting rehabilitation is fundamental for effective climate change mitigation policy.

Australian vegetated coastal ecosystems as global hotspots for climate change mitigation

Policies aiming to preserve vegetated coastal ecosystems (VCE; tidal marshes, mangroves and seagrasses) to mitigate greenhouse gas emissions require national assessments of blue carbon resources. Here, we present organic carbon (C) storage in VCE across Australian climate regions and estimate potential annual CO₂ emission benefits of VCE conservation and restoration. Australia contributes 5–11% of the C stored in VCE globally (70–185 Tg C in aboveground biomass, and 1,055–1,540 Tg C in the upper 1 m of soils). Potential CO₂ emissions from current VCE losses are estimated at 2.1–3.1 Tg CO₂-e yr^-1, increasing annual CO₂ emissions from land use change in Australia by 12–21%. This assessment, the most comprehensive for any nation to-date, demonstrates the potential of conservation and restoration of VCE to underpin national policy development for reducing greenhouse gas emissions.

Policies aiming to preserve vegetated coastal ecosystems (VCE) to mitigate greenhouse gas emissions require national assessments of blue carbon resources. Here the authors assessed organic carbon storage in VCE across Australian and the potential annual CO₂ emission benefits of VCE conservation and find that Australia contributes substantially the carbon stored in VCE globally.

Estimating the full greenhouse gas emissions offset potential and profile between rehabilitating and established mangroves

Mangrove forests are extremely productive, with rates of growth rivaling some terrestrial tropical rainforests. However, our understanding of the full suite of processes underpinning carbon exchange with the atmosphere and near shore-waters, the allocation of carbon in mangroves, and fluxes of non-CO₂ greenhouse gases (GHGs) are limited to a handful of studies. This constrains the scientific basis from which to advocate for greater support for and investment in mangrove restoration and conservation. Improving understanding is urgently needed given the on-going landuse pressures mangrove forests face, particularly throughout much of Southeast Asia. The current study reduces uncertainties by providing a holistic synthesis of the net potential GHG mitigation benefits resulting from rehabilitating mangroves and established forests. Rehabilitating sites from two contrasting locations representative of high (Tiwoho) and low (Tanakeke) productivity systems on the island of Sulawesi (Indonesia) were used as case studies to compare against established mangroves. A carbon budget, allocation and pathways model was developed to account for inputs (carbon sequestration) and outputs (GHG emissions of CO₂, N₂O and CH₄) to estimate Net Ecosystem Production (NEP) and Net Ecosystem Carbon Balance (NECB). Our results indicate that while Tiwoho's rehabilitating sites and established mangroves represent a significant carbon sink (-10.6 ± 0.9 Mg CO₂e ha^-1 y^-1 and 16.1 Mg CO₂e ha^-1 y^-1 respectively), the low productivity of Tanakeke has resulted in minimal reductions to date (0.7 ± 0.3 Mg CO₂e ha^-1 y^-1). Including NEP from mangrove-allied primary producer communities (e.g. benthic algae) and the portion of dissolved inorganic carbon exported from mangroves (EX_DIC) that remains within the water column may drive overall removals considerably upwards in established forests to -37.2 Mg CO₂e ha^-1 y^-1. These values are higher than terrestrial forests and strengthen the evidence base needed to underpin the use of forest carbon financing mechanisms for mangrove restoration.

Hydroperiod, soil moisture and bioturbation are critical drivers of greenhouse gas fluxes and vary as a function of landuse change in mangroves of Sulawesi, Indonesia

The loss and degradation of mangroves can result in potentially significant sources of atmospheric greenhouse gas (GHG) emissions. For mangrove rehabilitation carbon projects, quantifying GHG emissions as forests regenerate is a key accounting requirement. The current study is one of the first attempts to systematically quantify emissions of carbon dioxide (CO₂), nitrous oxide (N₂O) and methane (CH₄) from: 1) aquaculture ponds, 2) rehabilitating mangroves, and 3) intact mangrove sites and frame GHG flux within the context of landuse change. In-situ static chamber measurements were made at three contrasting locations in Sulawesi, Indonesia. The influence of key biophysical variables known to affect GHG flux was also assessed. Peak GHG flux was observed at rehabilitating (32.8 ± 2.1 Mg CO₂e ha^-1 y^-1) and intact, mature reference sites (43.8 ± 4.5 Mg CO₂e ha^-1 y^-1) and a dry, exposed disused aquaculture pond (30.6 ± 1.9 Mg CO₂e ha^-1 y^-1). Emissions were negligible at low productivity rehabilitating sites with high hydroperiod (mean 1.0 ± 0.1 Mg CO₂e ha^-1 y^-1) and an impounded, operational aquaculture pond (1.1 ± 0.2 Mg CO₂e ha^-1 y^-1). Heterogeneity in biophysical conditions and geomorphic position exerted a strong influence on GHG flux, with the longer hydroperiod and higher soil moisture content of seaward fringing mangroves correlated with decreased fluxes. A greater abundance of Mud lobster mounds and root structures in landward mangroves correlated to higher flux. When viewed across a landuse change continuum, our results suggest that the initial conversion of mangroves to aquaculture ponds releases extremely high rates of GHGs. Furthermore, the re-institution of hydrological regimes in dry, disused aquaculture ponds to facilitate tidal flushing is instrumental in rapidly mediating GHG flux, leading to a significant reduction in baseline emissions. This is an important consideration for forest carbon project proponents seeking to maximise creditable GHG emissions reductions and removals.

Community structure dynamics and carbon stock change of rehabilitated mangrove forests in Sulawesi, Indonesia

To date, discourse associated with the potential application of "blue carbon" within real-world carbon markets has focused on blue carbon as a mitigation strategy in the context of avoided deforestation (e.g., REDD+). Here, we report structural dynamics and carbon storage gains from mangrove sites that have undergone rehabilitation to ascertain whether reforestation can complement conservation activities and warrant project investment. Replicated sites at two locations with contrasting geomorphic conditions were selected, Tiwoho and Tanakeke on the island of Sulawesi, Indonesia. These locations are representative of high (Tiwoho, deep muds and silty substrates) and low (Tanakeke, shallow, coralline sands) productivity mangrove ecosystems. They share a similar management history of clearing and conversion for aquaculture before restorative activities were undertaken using the practice of Ecological Mangrove Rehabilitation (EMR). Species diversity and mean biomass carbon storage gains after 10 yr of regrowth from the high productivity sites of Tiwoho (49.2 ± 9.1 Mg C·ha^-1 ·yr^-1 ) are already almost of one-third of mean biomass stocks exhibited by mature forests (167.8 ± 30.3 Mg C·ha^-1 ·yr^-1 ). Tiwoho's EMR sites, on average, will have offset all biomass C that was initially lost through conversion within the next 11 yr, a finding in marked contrast to the minimal carbon gains observed on the low productivity, low diversity, coral atoll EMR sites of Tanakeke (1.1 ± 0.4 Mg C·ha^-1 ·yr^-1 ). These findings highlight the importance of geomorphic and biophysical site selection if the primary purpose of EMR is intended to maximize carbon sequestration gains.

Seasonal, interannual and decadal drivers of tree and grass productivity in an Australian tropical savanna

Tree-grass savannas are a widespread biome and are highly valued for their ecosystem services. There is a need to understand the long-term dynamics and meteorological drivers of both tree and grass productivity separately in order to successfully manage savannas in the future. This study investigated the interannual variability (IAV) of tree and grass gross primary productivity (GPP) by combining a long-term (15 year) eddy covariance flux record and model estimates of tree and grass GPP inferred from satellite remote sensing. On a seasonal basis, the primary drivers of tree and grass GPP were solar radiation in the wet season and soil moisture in the dry season. On an interannual basis, soil water availability had a positive effect on tree GPP and a negative effect on grass GPP. No linear trend in the tree-grass GPP ratio was observed over the 15-year study period. However, the tree-grass GPP ratio was correlated with the modes of climate variability, namely the Southern Oscillation Index. This study has provided insight into the long-term contributions of trees and grasses to savanna productivity, along with their respective meteorological determinants of IAV.

Exotic grass invasion alters microsite conditions limiting woody recruitment potential in an Australian savanna

Andropogon gayanus Kunth. is a large African tussock grass invading Australia’s tropical savannas. Invasion results in more intense fires which increases the mortality rate of adult woody plants. Invasion may also affect community structure by altering the recruitment potential of woody plants. We investigated the effects of A. gayanus invasion on ground-level microclimate, and the carbon assimilation potential and recruitment potential of two Eucalyptus species. We compared microclimatic variables from the early wet-season and into the mid-dry season to coincide with the period of growth of A. gayanus. We assessed Eucalyptus recruitment by monitoring seedling establishment, growth and survival of experimentally sown seed, and estimating seedling density resulting from natural recruitment. A. gayanus invasion was associated with increased grass canopy height, biomass and cover. Following invasion, the understorey microclimate had significantly reduced levels of photon flux density, increased air temperatures and vapour pressure deficit. The conditions were less favourable for woody seedling with aboveground biomass of seedlings reduced by 26% in invaded plots. We estimated that invasion reduced daily carbon assimilation of woody seedlings by ~30% and reduced survivorship of Eucalyptus seedlings. Therefore, A. gayanus invasion reduces recruitment potential, contributing to the transformation of savanna to a grassland ecosystem.

Invasive Andropogon gayanus (Gamba grass) alters litter decomposition and nitrogen fluxes in an Australian tropical savanna

The African grass Andropogon gayanus Kunth. is invading Australian savannas, altering their ecological and biogeochemical function. To assess impacts on nitrogen (N) cycling, we quantified litter decomposition and N dynamics of grass litter in native grass and A. gayanus invaded savanna using destructive in situ grass litter harvests and litterbag incubations (soil surface and aerial position). Only 30% of the A. gayanus in situ litter decomposed, compared to 61% of the native grass litter, due to the former being largely comprised of highly resistant A. gayanus stem. In contrast to the stem, A. gayanus leaf decomposition was approximately 3- and 2-times higher than the dominant native grass, Alloteropsis semilata at the surface and aerial position, respectively. Lower initial lignin concentrations, and higher consumption by termites, accounted for the greater surface decomposition rate of A. gayanus. N flux estimates suggest the N release of A. gayanus litter is insufficient to compensate for increased N uptake and N loss via fire in invaded plots. Annually burnt invaded savanna may lose up to 8.2% of the upper soil N pool over a decade. Without additional inputs via biological N fixation, A. gayanus invasion is likely to diminish the N capital of Australia’s frequently burnt savannas.

Vulnerability of native savanna trees and exotic Khaya senegalensis to seasonal drought

Seasonally dry ecosystems present a challenge to plants to maintain water relations. While native vegetation in seasonally dry ecosystems have evolved specific adaptations to the long dry season, there are risks to introduced exotic species. African mahogany, Khaya senegalensis Desr. (A. Juss.), is an exotic plantation species that has been introduced widely in Asia and northern Australia, but it is unknown if it has the physiological or phenotypic plasticity to cope with the strongly seasonal patterns of water availability in the tropical savanna climate of northern Australia. We investigated the gas exchange and water relations traits and adjustments to seasonal drought in K. senegalensis and native eucalypts (Eucalyptus tetrodonta F. Muell. and Corymbia latifolia F. Muell.) in a savanna ecosystem in northern Australia. The native eucalypts did not exhibit any signs of drought stress after 3 months of no rainfall and probably had access to deeper soil moisture late into the dry season. Leaf water potential, stomatal conductance, transpiration and photosynthesis all remained high in the dry season but osmotic adjustment was not observed. Overstorey leaf area index (LAI) was 0.6 in the native eucalypt savanna and did not change between wet and dry seasons. In contrast, the K. senegalensis plantation in the wet season was characterized by a high water potential, high stomatal conductance and transpiration and a high LAI of 2.4. In the dry season, K. senegalensis experienced mild drought stress with a predawn water potential -0.6 MPa. Overstorey LAI was halved, and stomatal conductance and transpiration drastically reduced, while minimum leaf water potentials did not change (-2 MPa) and no osmotic adjustment occurred. Khaya senegalensis exhibited an isohydric behaviour and also had a lower hydraulic vulnerability to cavitation in leaves, with a P50 of -2.3 MPa. The native eucalypts had twice the maximum leaf hydraulic conductance but a much higher P50 of -1.5 MPa. Khaya senegalensis has evolved in a wet-dry tropical climate in West Africa (600-800 mm) and appears to be well suited to the seasonal savanna climate of northern Australia. The species exhibited a large phenotypic plasticity through leaf area adjustments and conservative isohydric behaviour in the 6 months dry season while operating well above its critical hydraulic threshold.

Climate change and long-term fire management impacts on Australian savannas

Tropical savannas cover a large proportion of the Earth's land surface and many people are dependent on the ecosystem services that savannas supply. Their sustainable management is crucial. Owing to the complexity of savanna vegetation dynamics, climate change and land use impacts on savannas are highly uncertain. We used a dynamic vegetation model, the adaptive dynamic global vegetation model (aDGVM), to project how climate change and fire management might influence future vegetation in northern Australian savannas. Under future climate conditions, vegetation can store more carbon than under ambient conditions. Changes in rainfall seasonality influence future carbon storage but do not turn vegetation into a carbon source, suggesting that CO₂ fertilization is the main driver of vegetation change. The application of prescribed fires with varying return intervals and burning season influences vegetation and fire impacts. Carbon sequestration is maximized with early dry season fires and long fire return intervals, while grass productivity is maximized with late dry season fires and intermediate fire return intervals. The study has implications for management policy across Australian savannas because it identifies how fire management strategies may influence grazing yield, carbon sequestration and greenhouse gas emissions. This knowledge is crucial to maintaining important ecosystem services of Australian savannas.

Fire in Australian savannas: from leaf to landscape

Savanna ecosystems comprise 22% of the global terrestrial surface and 25% of Australia (almost 1.9 million km2) and provide significant ecosystem services through carbon and water cycles and the maintenance of biodiversity. The current structure, composition and distribution of Australian savannas have coevolved with fire, yet remain driven by the dynamic constraints of their bioclimatic niche. Fire in Australian savannas influences both the biophysical and biogeochemical processes at multiple scales from leaf to landscape. Here, we present the latest emission estimates from Australian savanna biomass burning and their contribution to global greenhouse gas budgets. We then review our understanding of the impacts of fire on ecosystem function and local surface water and heat balances, which in turn influence regional climate. We show how savanna fires are coupled to the global climate through the carbon cycle and fire regimes. We present new research that climate change is likely to alter the structure and function of savannas through shifts in moisture availability and increases in atmospheric carbon dioxide, in turn altering fire regimes with further feedbacks to climate. We explore opportunities to reduce net greenhouse gas emissions from savanna ecosystems through changes in savanna fire management.

Resource-use efficiency explains grassy weed invasion in a low-resource savanna in north Australia

Comparative studies of plant resource use and ecophysiological traits of invasive and native resident plant species can elucidate mechanisms of invasion success and ecosystem impacts. In the seasonal tropics of north Australia, the alien C4 perennial grass Andropogon gayanus (gamba grass) has transformed diverse, mixed tree-grass savanna ecosystems into dense monocultures. To better understand the mechanisms of invasion, we compared resource acquisition and usage efficiency using leaf-scale ecophysiological and stand-scale growth traits of A. gayanus with a co-habiting native C4 perennial grass Alloteropsis semialata. Under wet season conditions, A. gayanus had higher rates of stomatal conductance, assimilation, and water use, plus a longer daily assimilation period than the native species A. semialata. Growing season length was also ~2 months longer for the invader. Wet season measures of leaf scale water use efficiency (WUE) and light use efficiency (LUE) did not differ between the two species, although photosynthetic nitrogen use efficiency (PNUE) was significantly higher in A. gayanus. By May (dry season) the drought avoiding native species A. semialata had senesced. In contrast, rates of A. gayanus gas exchange was maintained into the dry season, albeit at lower rates that the wet season, but at higher WUE and PNUE, evidence of significant physiological plasticity. High PNUE and leaf (15)N isotope values suggested that A. gayanus was also capable of preferential uptake of soil ammonium, with utilization occurring into the dry season. High PNUE and fire tolerance in an N-limited and highly flammable ecosystem confers a significant competitive advantage over native grass species and a broader niche width. As a result A. gayanus is rapidly spreading across north Australia with significant consequences for biodiversity and carbon and retention.

Natural abundance (δ¹⁵N) indicates shifts in nitrogen relations of woody taxa along a savanna-woodland continental rainfall gradient

Water and nitrogen (N) interact to influence soil N cycling and plant N acquisition. We studied indices of soil N availability and acquisition by woody plant taxa with distinct nutritional specialisations along a north Australian rainfall gradient from monsoonal savanna (1,600-1,300 mm annual rainfall) to semi-arid woodland (600-250 mm). Aridity resulted in increased 'openness' of N cycling, indicated by increasing δ(15)N(soil) and nitrate:ammonium ratios, as plant communities transitioned from N to water limitation. In this context, we tested the hypothesis that δ(15)N(root) xylem sap provides a more direct measure of plant N acquisition than δ(15)N(foliage). We found highly variable offsets between δ(15)N(foliage) and δ(15)N(root) xylem sap, both between taxa at a single site (1.3-3.4 ‰) and within taxa across sites (0.8-3.4 ‰). As a result, δ(15)N(foliage) overlapped between N-fixing Acacia and non-fixing Eucalyptus/Corymbia and could not be used to reliably identify biological N fixation (BNF). However, Acacia δ(15)N(root) xylem sap indicated a decline in BNF with aridity corroborated by absence of root nodules and increasing xylem sap nitrate concentrations and consistent with shifting resource limitation. Acacia dominance at arid sites may be attributed to flexibility in N acquisition rather than BNF capacity. δ(15)N(root) xylem sap showed no evidence of shifting N acquisition in non-mycorrhizal Hakea/Grevillea and indicated only minor shifts in Eucalyptus/Corymbia consistent with enrichment of δ(15)N(soil) and/or decreasing mycorrhizal colonisation with aridity. We propose that δ(15)N(root) xylem sap is a more direct indicator of N source than δ(15)N(foliage), with calibration required before it could be applied to quantify BNF.

Savanna vegetation-fire-climate relationships differ among continents

Ecologists have long sought to understand the factors controlling the structure of savanna vegetation. Using data from 2154 sites in savannas across Africa, Australia, and South America, we found that increasing moisture availability drives increases in fire and tree basal area, whereas fire reduces tree basal area. However, among continents, the magnitude of these effects varied substantially, so that a single model cannot adequately represent savanna woody biomass across these regions. Historical and environmental differences drive the regional variation in the functional relationships between woody vegetation, fire, and climate. These same differences will determine the regional responses of vegetation to future climates, with implications for global carbon stocks.

Termite mound emissions of CH4 and CO2 are primarily determined by seasonal changes in termite biomass and behaviour

Termites are a highly uncertain component in the global source budgets of CH(4) and CO(2). Large seasonal variations in termite mound fluxes of CH(4) and CO(2) have been reported in tropical savannas but the reason for this is largely unknown. This paper investigated the processes that govern these seasonal variations in CH(4) and CO(2) fluxes from the mounds of Microcerotermes nervosus Hill (Termitidae), a common termite species in Australian tropical savannas. Fluxes of CH(4) and CO(2) of termite mounds were 3.5-fold greater in the wet season as compared to the dry season and were a direct function of termite biomass. Termite biomass in mound samples was tenfold greater in the wet season compared to the dry season. When expressed per unit termite biomass, termite fluxes were only 1.2 (CH(4)) and 1.4 (CO(2))-fold greater in the wet season as compared to the dry season and could not explain the large seasonal variations in mound fluxes of CH(4) and CO(2). Seasonal variation in both gas diffusivity through mound walls and CH(4) oxidation by mound material was negligible. These results highlight for the first time that seasonal termite population dynamics are the main driver for the observed seasonal differences in mound fluxes of CH(4) and CO(2). These findings highlight the need to combine measurements of gas fluxes from termite mounds with detailed studies of termite population dynamics to reduce the uncertainty in quantifying seasonal variations in termite mound fluxes of CH(4) and CO(2).

Stable isotopes reveal the contribution of corticular photosynthesis to growth in branches of Eucalyptus miniata

The deciduous bark habit is widespread in the woody plant genus Eucalyptus. Species with deciduous bark seasonally shed a layer of dead bark, thereby maintaining smooth-bark surfaces on branches and stems as they age and increase in diameter. This has a significant cost in terms of fire protection, because smooth-barked species have thinner bark than rough-barked species that accumulate successive layers of dead bark. Eucalypts are closely associated with fire, suggesting that the smooth-bark habit must also provide a significant benefit. We suggest that this benefit is corticular photosynthesis. To test this, we quantified the contribution of corticular photosynthesis to wood production in smooth-barked branches of Eucalyptus miniata growing in tropical savanna in northern Australia. We covered branch sections with aluminum foil for 4 years to block corticular photosynthesis and then compared the oxygen and carbon stable isotope composition of foil-covered and uncovered branch sections. We developed theory to calculate the proportion of wood constructed from corticular photosynthate and the mean proportional refixation rate during corticular photosynthesis from the observed isotopic differences. Coverage with aluminum foil for 4 years increased wood δ(13)C by 0.5‰ (P = 0.002, n = 6) and wood δ(18)O by 0.5‰ (P = 0.02, n = 6). Based on these data, we estimated that 11% ± 3% of wood in the uncovered branch sections was constructed from corticular photosynthate, with a mean δ(13)C of -34.8‰, and that the mean proportional refixation rate during corticular photosynthesis was 0.71 ± 0.15. This demonstrates that corticular photosynthesis makes a significant contribution to the carbon economy of smooth-barked eucalypts.

Invasive Andropogon gayanus (gamba grass) is an ecosystem transformer of nitrogen relations in Australian savanna

Invasion by the African grass Andropogon gayanus is drastically altering the understory structure of oligotrophic savannas in tropical Australia. We compared nitrogen (N) relations and phenology of A. gayanus and native grasses to examine the impact of invasion on N cycling and to determine possible reasons for invasiveness of A. gayanus. Andropogon gayanus produced up to 10 and four times more shoot phytomass and root biomass, with up to seven and 2.5 times greater shoot and root N pools than native grass understory. These pronounced differences in phytomass and N pools between A. gayanus and native grasses were associated with an altered N cycle. Most growth occurs in the wet season when, compared with native grasses, dominance of A. gayanus was associated with significantly lower total soil N pools, lower nitrification rates, up to three times lower soil nitrate availability, and up to three times higher soil ammonium availability. Uptake kinetics for different N sources were studied with excised roots of three grass species ex situ. Excised roots of A. gayanus had an over six times higher-uptake rate of ammonium than roots of native grasses, while native grass Eriachne triseta had a three times higher uptake rate of nitrate than A. gayanus. We hypothesize that A. gayanus stimulates ammonification but inhibits nitrification, as was shown to occur in its native range in Africa, and that this modification of the soil N cycle is linked to the species' preference for ammonium as an N source. This mechanism could result in altered soil N relations and could enhance the competitive superiority and persistence of A. gayanus in Australian savannas.

A canopy-scale test of the optimal water-use hypothesis

Common empirical models of stomatal conductivity often incorporate a sensitivity of stomata to the rate of leaf photosynthesis. Such a sensitivity has been predicted on theoretical terms by Cowan and Farquhar, who postulated that stomata should adjust dynamically to maximize photosynthesis for a given water loss. In this study, we implemented the Cowan and Farquhar hypothesis of optimal stomatal conductivity into a canopy gas exchange model, and predicted the diurnal and daily variability of transpiration for a savanna site in the wet-dry tropics of northern Australia. The predicted transpiration dynamics were then compared with observations at the site using the eddy covariance technique. The observations were also used to evaluate two alternative approaches: constant conductivity and a tuned empirical model. The model based on the optimal water-use hypothesis performed better than the one based on constant stomatal conductivity, and at least as well as the tuned empirical model. This suggests that the optimal water-use hypothesis is useful for modelling canopy gas exchange, and that it can reduce the need for model parameterization.

A test of the optimality approach to modelling canopy properties and CO2 uptake by natural vegetation

Photosynthesis provides plants with their main building material, carbohydrates, and with the energy necessary to thrive and prosper in their environment. We expect, therefore, that natural vegetation would evolve optimally to maximize its net carbon profit (NCP), the difference between carbon acquired by photosynthesis and carbon spent on maintenance of the organs involved in its uptake. We modelled N(CP) for an optimal vegetation for a site in the wet-dry tropics of north Australia based on this hypothesis and on an ecophysiological gas exchange and photosynthesis model, and compared the modelled CO2 fluxes and canopy properties with observations from the site. The comparison gives insights into theoretical and real controls on gas exchange and canopy structure, and supports the optimality approach for the modelling of gas exchange of natural vegetation. The main advantage of the optimality approach we adopt is that no assumptions about the particular vegetation of a site are required, making it a very powerful tool for predicting vegetation response to long-term climate or land use change.

Stem and leaf gas exchange and their responses to fire in a north Australian tropical savanna

We measured stem CO2 efflux and leaf gas exchange in a tropical savanna ecosystem in northern Australia, and assessed the impact of fire on these processes. Gas exchange of mature leaves that flushed after a fire showed only slight differences from that of mature leaves on unburned trees. Expanding leaves typically showed net losses of CO2 to the atmosphere in both burned and unburned trees, even under saturating irradiance. Fire caused stem CO2 efflux to decline in overstory trees, when measured 8 weeks post-fire. This decline was thought to have resulted from reduced availability of C substrate for respiration, due to reduced canopy photosynthesis caused by leaf scorching, and to priority allocation of fixed C towards reconstruction of a new canopy. At the ecosystem scale, we estimated the annual above-ground woody-tissue CO2 efflux to be 275 g C m(-2) ground area year(-1) in a non-fire year, or approximately 13% of the annual gross primary production. We contrasted the canopy physiology of two co-dominant overstory tree species, one of which has a smooth bark on its branches capable of photosynthetic re-fixation (Eucalyptus miniata), and the other of which has a thick, rough bark incapable of re-fixation (Eucalyptus tetrodonta). Eucalyptus miniata supported a larger branch sapwood cross-sectional area in the crown per unit subtending leaf area, and had higher leaf stomatal conductance and photosynthesis than E. tetrodonta. Re-fixation by photosynthetic bark reduces the C cost of delivering water to evaporative sites in leaves, because it reduces the net C cost of constructing and maintaining sapwood. We suggest that re-fixation allowed leaves of E. miniata to photosynthesize at higher rates than those of E. tetrodonta, while the two invested similar amounts of C in the maintenance of branch sapwood.

Viewpoint: Assessing the carbon sequestration potential of mesic savannas in the Northern Territory, Australia: approaches, uncertainties and potential impacts of fire

Tropical savannas cover a quarter of the Australian landmass and the biome represents a significant potential carbon sink. However, these savannas are subject to frequent and extensive fire. Fire regimes are likely to affect the productivity and carbon sequestration potential of savannas, through effects on both biomass and carbon emissions. The carbon sequestration potential has been estimated for some savanna sites by quantifying carbon storage in biomass and soil pools, and the fluxes to these pools. Using different techniques, previous work in these savannas has indicated that net ecosystem productivity [NEP, net primary productivity (NPP) less heterotrophic respiration] was about -3 t C ha^-1 y^-1 (i.e. a carbon sink). However, the impacts of fire were not accounted for in these calculations. Estimates of NEP have been combined with remotely-sensed estimates of area burnt and associated emissions for an extensive area of mesic savanna in Arnhem Land, NT, Australia. Combining NEP estimates with precise fire data provides an estimate of net biome productivity (NBP), a production index that includes carbon loss through disturbance (fire), and is thus a more realistic indicator of sequestration rate from this biome. This preliminary analysis suggests that NBP is approximately -1 t C ha^-1 y^-1 (i.e. a carbon sink). A reduction in the annual area burnt is likely to increase the sink size. Uncertainties surrounding these estimates of NBP and the implications of these uncertainties for land management in these extensive landscapes are discussed.

Carbon balance of a tropical savanna of northern Australia

Through estimations of above- and below-ground standing biomass, annual biomass increment, fine root production and turnover, litterfall, canopy respiration and total soil CO(2) efflux, a carbon balance on seasonal and yearly time-scales is developed for a Eucalypt open-forest savanna in northern Australia. This carbon balance is compared to estimates of carbon fluxes derived from eddy covariance measurements conducted at the same site. The total carbon (C) stock of the savanna was 204+/-53 ton C ha(-1), with approximately 84% below-ground and 16% above-ground. Soil organic carbon content (0-1 m) was 151+/-33 ton C ha(-1), accounting for about 74% of the total carbon content in the ecosystem. Vegetation biomass was 53+/-20 ton C ha(-1), 39% of which was found in the root component and 61% in above-ground components (trees, shrubs, grasses). Annual gross primary production was 20.8 ton C ha(-1), of which 27% occurred in above-ground components and 73% below-ground components. Net primary production was 11 ton C ha(-1) year(-1), of which 8.0 ton C ha(-1) (73%) was contributed by below-ground net primary production and 3.0 ton C ha(-1) (27%) by above-ground net primary production. Annual soil carbon efflux was 14.3 ton C ha(-1) year(-1). Approximately three-quarters of the carbon flux (above-ground, below-ground and total ecosystem) occur during the 5-6 months of the wet season. This savanna site is a carbon sink during the wet season, but becomes a weak source during the dry season. Annual net ecosystem production was 3.8 ton C ha(-1) year(-1).

Daily and seasonal patterns of carbon and water fluxes above a north Australian savanna

Daily and seasonal fluxes of carbon dioxide and water vapor above a north Australian savanna were recorded over a complete dry season-wet season annual cycle using the eddy covariance technique. Wet season rates of photosynthesis and transpiration were larger than those measured in the dry season and were dominated by the presence of the grassy understory. As the dry season progressed and the grass understory died, ecosystem rates of assimilation and water vapor flux declined substantially. By the end of the dry season, canopy assimilation and evapotranspiration rates were 20-25% of wet season values. Assimilation was light saturated in the dry season but not in the wet season. Stomatal control of transpiration increased between the wet and dry season. This was revealed by the decline in the slope of E with increasing leaf-to-air vapor pressure difference (D) between wet and dry seasons, and also by the significant decrease in the ratio of boundary to canopy conductance observed between the wet and dry seasons. A simple pan-tropical modeling of leaf area index or wet season canopy CO2 flux was undertaken. It was shown that with readily available data for foliar N content and the ratio of rainfall to potential evaporation, leaf index and wet season canopy CO2 flux can be successfully estimated for a number of tropical ecosystems, including north Australian savannas.

Monsoonal influences on evapotranspiration of savanna vegetation of northern Australia

Data from savannas of northern Australia are presented for net radiation, latent and sensible heat, ecosystem surface conductance (G _s) and stand water use for sites covering a latitudinal range of 5° or 700 km. Measurements were made at three locations of increasing distance from the northern coastline and represent high- (1,750 mm), medium- (890 mm) and low- (520 mm) rainfall sites. This rainfall gradient arises from the weakened monsoonal influence with distance inland. Data were coupled to seasonal estimates of leaf area index (LAI) for the tree and understorey strata. All parameters were measured at the seasonal extremes of late wet and dry seasons. During the wet season, daily rates of evapotranspiration were 3.1-3.6 mm day^-1 and were similar for all sites along the rainfall gradient and did not reflect site differences in annual rainfall. During the dry season, site differences were very apparent with evapotranspiration 2-18 times lower than wet season rates, the seasonal differences increasing with distance from coast and reduced annual rainfall. Due to low overstorey LAI, more than 80% of water vapour flux was attributed to the understorey. Seasonal differences in evapotranspiration were mostly due to reductions in understorey leaf area during the dry season. Water use of individual trees did not differ between the wet and dry seasons at any of the sites and stand water use was a simple function of tree density. G _s declined markedly during the dry season at all sites, and we conclude that the savanna water (and carbon) balance is largely determined by G _s and its response to atmospheric and soil water content and by seasonal adjustments to canopy leaf area.