Tropical forests have a key role in the global carbon dynamics by accounting for one-third of the terrestrial gross primary production and one-half of the terrestrial stored carbon. Increasing atmospheric carbon dioxide concentration, rising temperatures, drought events, and deforestation are expected to affect ecosystem functioning through plant physiological responses and forest loss over the coming decades. How tropical forests will respond to increasing atmospheric carbon dioxide concentration and climate change are sources of uncertainty in predicting their future carbon stocks and net primary productivity. Among tropical forests, the ones under stressful conditions, such as the seasonally dry tropical forests that endure periodic droughts, can be vulnerable to these environmental changes because they are already at the edge of climate niches to sustain forest formations with high biomass. Brazil has been the largest source of carbon emissions from gross deforestation up to 2013: In 2013 alone, 192 000 hectares (1920 km²) of Caatinga forests (where the largest continuous extent of neotropical seasonally dry tropical forests is located) and 24 000 hectares (240 km²) of Atlantic forests were deforested. An aggravating factor is that only 6.2% of Brazilian seasonally dry tropical forests extent is protected. In this context, it is crucial to advance our understanding of the carbon sink of these forests.
Over recent decades, the terrestrial carbon sink has been increasing globally. This phenomenon is possibly explained by the increases in atmospheric carbon dioxide concentration (carbon dioxide fertilisation), which is expected to enhance plant growth. Carbon dioxide has a key role in photosynthesis and can potentially increase water use efficiency by reducing stomata conductance. In this context, the increases in atmospheric carbon dioxide concentration are thought to have enhanced photosynthesis more than rising temperatures have enhanced respiration. Hence, drought-related stress on plant growth would be understated by rising atmospheric carbon dioxide concentration. However, the mechanisms involved in the feedbacks among vegetation, atmospheric carbon dioxide, and climate are complex. For example, the increase in photosynthesis and water use efficiency led by rising atmospheric carbon dioxide concentration does not necessarily promote stand growth because of the effects of other co-occurring factors. The effects of increasing drought, rising temperatures, competition, and physiological acclimation to higher levels of carbon dioxide can constrain tree growth and also lead to tree mortality. Therefore, it is difficult to disentangle the effects of climate fluctuations and rising carbon dioxide on carbon dynamics because these factors covary and can interact over time.
While recent studies have shown a long-term decline in the Amazon rainforest carbon sink, mostly driven by climate-induced tree mortality, others have predicted that tropical rainforest carbon sink may be resilient to climate change in the next decades. However, it remains uncertain how Brazilian seasonal forests (which are already exposed to drought), such as deciduous forests and semideciduous forests, have responded to the increasing levels of atmospheric carbon dioxide and climate fluctuations over time.
In a paper published in the journal Science Advances on 18 December 2020, a team of scientists led by Vinícius Andrade Maia of the Departamento de Ciências Florestais at the Universidade Federal de Lavras, assess how these forests are behaving over time by using long-term seasonal forest census data from southeastern Brazil to investigate the long-term trends of carbon stocks, gains, losses, and net carbon sink.
Maia et al. draw their inferences from 95 census intervals nested within 32 sites, ranging between 1987 and 2020 (mean site total monitoring length, about 15 years). The spatial (number of sites and sampled area) and temporal (interval length and total monitoring time) sampling efforts varied among sites and forest types. The forest sites used are in advanced successional stages, free from fire, flood, landslides, and Human disturbances at least for decades before the first census of each site. The data encompass a wide environmental space and three forest types (deciduous, evergreen, and semideciduous forests), allowing Maia et al. to investigate whether forests under different climates have differed in their long-term trends. In addition to the rising atmospheric carbon dioxide concentration (parts per million) and carbon dioxide change (parts per million per year) over time, mean annual temperature and mean annual precipitation have shown an unstable temporal trend in their data. Therefore, Maia et al. used time (year) as a proxy of the effects of rising carbon dioxide, climate fluctuations, and other unmeasured confounding effects over time. Maia et al. did this because these factors can be codependent and correlated over time, being difficult to disentangle their individual effects. Maia et al. fitted statistical models to assess the general long-term trends of carbon dynamics, as well as the long-term trends by forest type, and to test whether climate mediates these long-term trends. More specifically, Maia et al. tested whether sites under different climate conditions have differed in their long-term trends. In this sense, Maia et al. expected the long-term trends of sites under drier and warmer conditions to differ from the long-term trends of sites under wetter and colder conditions.
Both mean annual temperature and mean annual precipitation fluctuated over time. Mean annual temperature showed a positive trend (0.04°C by year), while mean annual precipitation showed a negative trend (−10.2 mm by year). Moreover, carbon dioxide change increased almost linearly with time. In general, the carbon stocks increased over time until 2013 (roughly 0.67% by year) and then started to decline. During most of the time, the net carbon sink was above zero (positive balance between carbon gains and losses), with a slight negative trend; however, in 2013, the net carbon sink became negative (carbon losses exceeded carbon gains), which explains the carbon stock decline after 2013. In general, the net carbon sink decreased by 0.13 tonnes of carbon per hectare per year. Carbon gains decreased (about 2.6% by year) and carbon losses increased (about 3.4% by year) almost linearly over time.
The long-term trends by forest type revealed that the carbon stocks of the semideciduous forests increased over time, while the carbon stocks of the deciduous and evergreen forests showed a stable trend, whereby the deciduous forests showed a slight (but nonsignificant) decrease. All forest types showed a decline in their net carbon sinks over time, with a stronger decline in deciduous forests. Carbon gains decreased and carbon losses increased in all forest types, but these trends were more pronounced in the deciduous forests than in other forest types.
In the final model, including climate, soil, and time, the forests under different climate conditions differed in their long-term trends. Carbon stocks increased over time, except for the driest and warmest sites, in a way that the positive temporal trend of carbon stocks became weaker with decreasing mean annual precipitation and increasing mean annual temperature. Net carbon sink decreased over time, in a way that its negative temporal trend became weaker as mean annual precipitation increases and mean annual temperature decreases. Carbon gains decreased with time, whereby its negative trend became weaker with decreasing mean annual temperature and increasing mean annual precipitation, and became positive under wet conditions. At the same time, carbon losses increased over the years; the temporal trend of carbon losses became weaker with increasing mean annual precipitation and mean annual temperature.
The effects of climate have changed over time. Carbon stocks increased with mean annual precipitation and decreased with mean annual temperature; these effects became stronger from past to present. The effects of mean annual precipitation and mean annual temperature on the net carbon sink were near zero in the past; however, from past to present, the effect of mean annual precipitation became positive while the effect of mean annual temperature became negative. Meanwhile, the positive effect of mean annual precipitation and the negative effect of mean annual temperature on carbon gains increased over time. The positive effect of mean annual precipitation and the negative effect of mean annual temperature on carbon losses became weaker from past to present. The only significant effect from soil variables was found for soil organic matter, which had negative effects on carbon gains and carbon losses. Site area, which was used as a proxy of edge effects, has not displayed significant effects on the carbon dynamics variables.
The net carbon sink in southeastern Brazil’s seasonal forests is declining over time by decreasing carbon gains and increasing carbon losses. The carbon sink became a carbon source in 2013, which explains the decline in carbon stocks after this year. Among the forests under different climates, the driest and warmest sites are experiencing the most severe decrease in carbon gains, the most severe increase in carbon losses, and, therefore, the most severe decline in the net carbon sink. The severe decrease in the carbon sink of the driest and warmest forests suggests that these forests may have reached a climatic stress threshold. The carbon stocks of the driest and warmest sites remain stable with a slight (but nonsignificant) decrease; however, if their net carbon sink remains in a negative balance, then their carbon stocks will also decline in the near future, as observed in the general trend after 2013.
Under wet conditions, the carbon gains increased over time, consistent with the hypothesised pantropical increase in tree growth caused by carbon dioxide fertilisation. However, the long-term decrease in carbon gains experienced by the sites under intermediate climate and by the driest and warmest sites is inconsistent with carbon dioxide fertilisation effects and with findings in the Amazon forests. Recent evidence suggests that atmospheric carbon dioxide increases are not necessarily translated into larger amounts of sequestered carbon by old-growth forest trees. A large portion of this carbon dioxide surplus can be emitted back into the atmosphere by processes such as soil respiration. Alternative hypotheses might also explain this effect. High levels of carbon dioxide can increase tree-to-tree competition by enhancing the growth of some species or individuals, which, at the stand level, can decrease carbon gains by constraining the growth of the suppressed trees. In addition, trees growing under increasing carbon dioxide can acclimate to high carbon dioxide availability, by reducing their photosynthetic capacity below the expected for a given carbon dioxide level. The potential increases in liana density caused by increasing carbon dioxide and decreasing mean annual precipitation. can also decrease carbon gains by stimulating tree-liana competition for light, water, and nutrients. Meanwhile, the potential direct effects of climate fluctuations, such as decreasing mean annual precipitation and increasing mean annual temperature, on carbon gains may have suppressed the effects of carbon dioxide fertilisation on photosynthesis and water use efficiency. Because water is an important resource for photosynthesis and high temperatures enhance respiration, drought and rising temperatures can cause physiological stress (e.g. carbon starvation and hydraulic failure) and decrease tree growth.
The driest and warmest sites are experiencing the most severe increase in carbon losses over time. This effect is possibly explained by the combined effects of increasing carbon dioxide, drought, and higher temperatures on tree mortality. Drought and high temperatures can directly drive tree mortality through physiological stress, which can be potentialized by the effects of carbon dioxide on tree mortality. Increasing carbon dioxide can accelerate the speed at which trees reach large heights, which would increase the rate at which they are exposed to dry upper canopy, lightning, and windthrow and to the physiological aspects associated with larger sizes. In addition, increases in liana density provoked by rising carbon dioxide and drought can also enhance tree mortality by increasing tree-liana competition for light and water. The decrease in carbon gains and the increase in carbon losses of the driest and warmest sites over time, which have a distinct flora, naturally associated with dry and warm conditions, suggest that these forests may have reached a stress threshold due to the effects of increasing drought, temperature, and carbon dioxide.
Carbon stocks and carbon gains decreased with mean annual temperature and increased with mean annual precipitation over space, as expected by theory. These effects are evidence of the physiological responses to the harsh conditions imposed by high temperatures and drought, which are related to physiological stress. However, the wettest and coldest sites have both greater carbon gains and greater carbon losses, consistent with the high-gain high-loss dynamic pattern. Therefore, the spatial effects of mean annual temperature and mean annual precipitation on the net carbon sink were near zero most of time and became negative for mean annual temperature and positive for mean annual precipitation from past to present. These shifts in the spatial effects of climate occurred because the driest and warmest sites experienced the most severe increase in their carbon losses and the most severe decrease in their carbon gains over time. Thus, the net carbon sink of the driest and warmest sites became smaller than the net carbon sink of the wettest and coldest sites. Although mean annual temperature was more important than mean annual precipitation to differentiate the forest types, mean annual precipitation was more important than mean annual temperature to predict the net carbon sink, carbon gains, and carbon losses. Soil variables were important to differentiate forest types; however, only soil organic matter displayed significant effects on carbon dynamics. Forests on soils with lower levels of soil organic matter tend to have higher carbon gains and higher carbon losses. Soil organic matter is related to soil quality and higher productivity; thus, these effects are not conclusive and should be further investigated. In addition, site area, which was a proxy of edge effects, has not displayed significant effects in the carbon dynamics. Therefore, the results suggest that climate is the most important predictor of the spatial patterns of carbon dynamics in Maia et al.'s study region.
Maia et al. recognise the limitations of their data, such as space-for-time and unbalanced spatial-temporal sampling efforts. However, the negative trend and the negative balance of the carbon sink were a clear pattern in our data and results. In general, these forests are shifting from carbon sinks to carbon sources. Currently, the forests under intermediate climate conditions and the forests under the driest and warmest conditions are already carbon sources, probably because they may have reached a stress threshold. Meanwhile, the carbon sink of the wettest and coldest forests is continually declining. The driest and warmest forests naturally have lower carbon stocks, which will decline in the near future if their net carbon sink remains in a negative balance. These long-term trends in carbon dynamics are likely to be influenced by climate fluctuations and rising carbon dioxide. However, because these factors are correlated and can interact over time, their mechanistic individual effects and the effects of other unmeasured drivers remain uncertain and should be further investigated. Atmospheric carbon dioxide concentration, temperature, and drought events are expected to continue increasing in upcoming decades, implying that the ecosystem functioning of southeastern Brazil tropical seasonal forests may be under threat.
Political actions to mitigate greenhouse gas emissions, together with conservation policies to protect these ecosystems, are needed. Maia et al. also argue that the driest and warmest sites (deciduous forests, seasonally dry tropical forests) should be further included in conservation policies and that revegetation strategies in agricultural areas can be useful to offset the decline in the carbon sink and stocks. Beyond the political implications, our findings are also useful to improve the predictions of future global carbon sink and to bring knowledge to the carbon cycle of tropical forests.
Follow Sciency Thoughts on Facebook.
Follow Sciency Thoughts on Twitter.