Each year the amount of carbon dioxide removed from the atmosphere by photosynthesis is nearly balanced by the amount of carbon dioxide returned to the atmosphere by respiration. Predicting just how close this balance is should be considered critical to understanding global carbon cycling and, by extension, climate change. However, this balance has proven difficult to predict, in large part due to the lack of a robust mechanistic model of plant respiration that is well parameterized under fluctuating environmental conditions. Individual environmental responses of respiration have been studied, and as a result it is well known that the biochemistry and physiology of respiration respond both directly and indirectly to temperature, mineral nutrition, water availability, O2, atmospheric trace gases and pollutants, and light. Of these, temperature has always been considered the most important environmental driver at the global scale and temperature is thought to control respiratory activity through time—from seconds to centuries. While incredibly important, the interactive effects among environmental variables (as they naturally fluctuate in ecological settings) on respiratory rates have not received adequate attention.
As an enzymatically driven process, respiration is expected to increase with temperature over ecologically meaningful ranges. When ambient air temperatures increase, so does molecular motion, membrane fluidity, stress on molecular bonds and the random chance of substrates colliding with the active sites of proteins. Although not a single enzyme reaction, there is a long tradition of modelling the temperature response of plant respiration using an Arrhenius type equation with a basal rate of respiration and an overall activation energy, resulting in roughly a doubling of the respiration rate with every 10 °C increase in temperature (i.e., Q10 ~2). While this approach simplifies carbon cycle models, it overlooks the well-known phenomenon of thermal acclimation—longer term adjustments in respiration rates at any given temperature that result in a more constant respiratory flux than would have been predicted from short-term temperature response curves (Larigauderie and Korner 1995). Respiratory acclimation has been observed both in the laboratory and in the field, but the degree of acclimation is variable and the lack of a complete mechanistic understanding continues to thwart efforts for models to account for it. Still, thermal acclimation likely results in a reduction in the long-term temperature sensitivity of respiration and in predicted rates of respiratory CO2 release in response to increasing ambient temperatures. As such, thermal acclimation of respiration is a critical missing piece of our current understanding of how plant carbon fluxes respond to their environment (Atkin et al. 2008, Smith and Dukes 2013).
Given the likely importance of respiratory thermal acclimation, it may be surprising to many that a recent meta-analysis turned up only 43 studies spanning a total of 103 species from alpine, Arctic and Antarctic, boreal, temperate and tropical climates, and including forbs, graminoids (sedges and grasses), and evergreen and deciduous shrubs, trees and lianas (Slot and Kitajima 2015). The lack of field studies under realistic warming scenarios is particularly noticeable. Furthermore, despite the well-known importance of boreal forests to the global carbon cycle (Goulden et al. 1998), only nine studies were included in the meta-analysis by Slot and Kitajima (2015) from this vast biome, and of these, only two Finnish studies of Scots pine (Pinus sylvestris) were done in the field with a realistic (2–3 °C) warming treatment (Wang et al. 1995, Zha et al. 2002). Two important studies from the University of Minnesota (Reich et al. 2016, Wei et al. 2016) add substantial new information regarding the likely respiratory responses of tree species and the temperate-boreal ecotone to predicted scenarios of future warming. The results from this work are encouraging for the modelling community, but surprising to the experimentalist.
Wei et al. (2016) utilize a unique experiment on the boreal-temperate forest ecotone in northern Minnesota with plots in two habitat types (open and understory) with two warming treatments (ambient and +3.4 °C) that were applied to both the soil and the air from 2009 to 2012 using a chamberless system. The goal of these researchers was to detect respiratory acclimation to relatively small changes in ambient temperature against a thermally fluctuating background while statistically controlling for both leaf development and local resource availability across seasons, years, locations and species. In addition, the authors collected simultaneous measurements of leaf traits (specific leaf area, leaf nitrogen and non-structural carbohydrates content) in order to probe for proxies that may be simpler to measure than respiration, and may provide insights regarding the mechanistic underpinnings of any observed responses, or result in alternative modelling approaches. The result is a huge data set that represents a very significant long-term effort from the research team.
The six predictions that are made by Wei et al. (2016) nicely demonstrate the complexity of the potential responses of respiration with shifts in temperature and allude to the strength of their experiment. Their first two predictions (H1 and H2 of their text) state that, since the underlying biochemistry has not changed, the respiratory temperature response, Q10, will remain constant, but that acclimation to higher growth temperatures will reduce the basal respiration rate at a constant reference temperature (RTref). Evidence exists both for and against these two hypotheses, but regardless they represent the key parameters that drive the overall respiratory response, and both are stated, testable hypotheses within this experiment. The third prediction (H3) addresses variation across space (open vs understory habitats), incorporating the light environment, and therefore indirect links to photosynthesis. The authors predict that trees from the open habitat will have both a higher Q10 (i.e., be more responsive to temperature) and demonstrate greater acclimation (lower RTref). To make an unstated link between H1 and H3, the authors suggest a change in the underlying respiratory biochemistry in the plants grown in open sites, presumably to support the increased rates of photosynthesis, substrate processing or perhaps photoprotection in these plants compared with their understory counterparts. The greater degree of thermal acclimation predicted for these plants would ultimately determine whether respiratory carbon losses exceed or fall below those of the control plants at any given temperature. After considering variation in space, the authors turn their attention to variation in time and here predict (H4) that mid-season measurements are likely to show the highest instantaneous temperature response (increased Q10) and greatest acclimation (reduced RTref) compared with either early or late season measurements. The justification is again linked to photosynthesis (and the implied impact of photosynthesis on respiratory physiology), as previous work has shown mid-season peaks in photosynthesis at these boreal forest sites (Reich et al. 2015). The authors further explore the importance of the duration of the study in their sixth prediction, stating that respiratory acclimation to temperature will strengthen through time, and, as a result, the response will be cumulative, with RTref decreasing with each successive year of the experiment. In addition, respiratory responses are predicted to be linked to leaf traits, such as decreased leaf N and non-structural carbohydrates, with climate warming (from their H5). Reductions in leaf nitrogen and non-structural carbohydrates presumably decrease respiratory enzyme concentrations, protein turnover rates and respiratory substrates, and therefore reduce in situ respiration at any temperature.
While the logic of each of these predictions on its own is simple to follow, the interactions between these variables are complex, and the overall responses of respiration and leaf traits across species, space, time and environmental conditions are difficult to predict. As such, the empirical approach used in this study seems particularly appropriate. Of course, the trade-off is the number of variables that need to be included in the statistical models used to analyse an experiment of this size, and therefore the large number of replicate measurements required to detect significance. An ANOVA of the main results across species shows 25 combinations of variables tested by >800 replicate measurements and concludes that time, space and species are all important interacting contributors to the observed respiratory variation—but surprisingly temperature treatment was not! Within each species, very little variation was observed in the Q10 and thus it was rare for any model variable to be significant. Respiration rates (RTref) were more likely to vary and show some effects of both space and time.
The results of this paper, and the expanded set of species from the experiment presented in Reich et al. (2016), suggest that models can continue to take a rather simplistic approach and use fixed Q10 values (although note there are well-known variations in Q10 at larger spatial scales (e.g., Tjoelker et al. 2001) that still need to be taken into account across plant functional types or perhaps climatic zones). These authors also suggest that acclimation is prevalent across tree species, and while species-specific differences may need to be considered, relatively simple functions can be used to implement these findings. The logic behind this suggestion stems from their finding that acclimation is both a time- and space-independent response in this boreal-temperate forest ecotone. Perhaps, the best news for those involved in efforts to model respiratory responses and their effect on climate change is the finding ‘that acclimation of leaf respiration to warming is a general response that is not strongly context- nor taxon-dependent for deciduous woody plants in seasonal temperate and boreal environments.’
One note of interest is the form of the respiratory model used. Several recent studies have examined this topic in detail, and one such study (Heskel et al. 2016) concludes that the exponential models, like the Q10 formulation used here, fall short, overestimating respiration rates at higher temperatures, while potentially underestimating net carbon gain. The recommendation of the cross-biome study of Heskel et al. (2016) is to use a simple polynomial model due to the goodness of fit and ease of implementation. Wei et al. (2016) test four different models and ultimately conclude that they all make similar predictions across the range of the experimental treatments, and therefore they choose to stick with the traditional exponential Q10 model. Moving forward, it may be important for the community to coalesce around a single common formulation to facilitate cross-site and cross-model comparisons.
As convenient as the overall results are for large-scale modelling efforts, and perhaps even future field campaigns, they are ecologically surprising. For example, trees grown in open sites were found to have higher photosynthetic rates and higher growth rates, yet acclimation resulted in no observed changes in respiration or respiratory substrates. This implies that there is a decoupling of respiration from both the demand for respiratory products and the supply of respiratory substrates. The lack of accumulation of non-structural carbohydrates despite the increased photosynthetic rates and reduced respiration rates is particularly challenging to interpret. In addition, leaf traits did not change with temperature. Overall, these results then suggest that the effect of warming is perhaps not realized at the leaf level, but instead at the whole-plant level, perhaps through changes in cell division and expansion, resulting in more leaves with identical respiratory rates, carbohydrate and nitrogen concentrations. Thus, increased leaf-level efficiency as benefits go up and costs go down (increased leaf-level photosynthesis relative to respiration) under warmer conditions may result in greater canopy-level carbon gain that is either offset by non-leaf respiratory losses (e.g., stem, branch, twig and root respiration, carbon exudates and volatile carbon compounds), or results in compounded increases in growth (Figure 1). Clearly, linked leaf, plant and canopy-level carbon flux measurements would be an ideal way to pursue the implications of these changes.
Another ecologically surprising result from this study is that the temperature effects appear to accumulate with time over the course of the experiment. In two of the three study species (Acer rubrum and Betula papyrifera), acclimation resulted in continued decreases of the rate of respiration in the elevated vs control treatments in each of the 4 years. In fact, in Year 1, respiration at the growth temperature in red maple was still significantly higher in the elevated compared with the ambient treatment, but by Year 5 it was significantly lower. A similar response was seen in birch but not in aspen. This perhaps again suggests an increase in carbon-use efficiency and/or compound growth benefit from higher photosynthetic rates and resource investment in a larger canopy.
The implications of this work are numerous and important. If respiratory temperature responses (Q10) can be considered constant, and respiratory acclimation is complete (such that respiration rates at any given growth temperature acclimate within a few days and absolute rates of carbon exchange remain constant), our lack of a mechanistic model is, on one hand, less troubling, and large-scale respiratory fluxes can continue to be estimated with simple empirical formulations. On the other hand, the implications for the mechanisms underlying this acclimation are puzzling.
How does overall growth and metabolic activity increase in response to warming while respiration decreases? If the proximate cause is seen as increased molecular motion and stress on molecular bonds, how is this overcome without changing either the underlying biochemistry or consumption and accumulation of substrates, particularly in light of probable increased demand for respiratory products? Perhaps, the answer is in scaling leaf-level results to ecosystem fluxes or, in the even shorter term, dynamic changes that result in rapid increases and decreases in respiratory activity to support a burst of metabolic activity associated with cell expansion and division that then creates the principal investment for a compound growth increase. While shifts in respiratory rates at the leaf level may seem small, the aggregate change—from single leaves to communities, ecosystems and the globe—represents an enormous amount of respiratory activity and carbon release. To this end, Wei et al. (2016) have produced a large and valuable data set for three species grown in the field under naturally complex and fluctuating conditions. Their findings suggest that our simple models may be working just fine. They also suggest we still do not know why, and have plenty of work to do.
We thank Danielle Way for her editorial advice and assistance.
Conflict of interest