Periodic Reporting for period 4 - REALM (Re-inventing Ecosystem And Land-surface Models)
Período documentado: 2023-04-01 hasta 2024-09-30
Terrestrial ecosystems respond to changes in climate and the atmospheric environment, which they in turn help to regulate. But the lack of established theory for many key processes has hindered development of reliable models of land ecosystem processes. REALM set out to improve this situation by developing new theory and using the fast-growing body of relevant observations at scales from leaf-level trait and photosynthesis measurements to global satellite data sets.
REALM has developed a universal theory and a parameter-sparse model for gross primary production, GPP (total ecosystem photosynthesis) that outperforms previous models despite its far greater simplicity and transparency. The model reproduces both diurnal and seasonal cycles and temporal trends in GPP at flux-measurement sites using the same parameter values for all biomes. Further new theory explains the "leaf economics spectrum" and successfully predicts how leaf economics vary along environmental gradients. Seasonal patterns of green vegetation cover, as observed by satellites, can also be predicted by a universal optimality principle; as can geographic patterns in the maximum allocation of biomass to leaves.
This new body of theory has improved understanding of the interaction of carbon (C) and nitrogen (N) cycles. Many models assume that N supply (from soil and N deposition) to plants determines leaf-level N and thence, photosynthetic capacity. But experiments show that N addition mainly affects biomass allocation, with a positive effect of leaf N but little or none on photosynthesis. This insight has led to (ongoing) research on the effects of phosphorus (P) addition. A comprehensive theory of C-N-P cycle interactions is within sight.
This revised understanding of plant-nutrient interactions has practical consequences. For example, the decline in leaf N over recent decades does not (as was once claimed) signify increasing N limitation of primary production. Instead, it follows from optimal resource use by plants, responding to warming and rising atmospheric CO2.
Further insights have been gained by coupling the new GPP model to a minimal representation of how heterotrophic respiration (primarily CO2 release from soil due to decomposition of dead organic matter) varies with the soil environment. The observed growth in the amplitude of seasonal variations in high-latitude atmospheric CO2, and the seasonal geographic pattern of net ecosystem exchange of CO2 between the atmosphere and land, are predicted accurately.
Three reasons have underpinned the success of REALM in developing improved ecosystem model components for ESMs.
1) A first-principles approach based on eco-evolutionary optimality hypotheses has generated clear and testable governing equations for the most important plant and ecosystem processes.
2) Huge increases in the amount and variety of relevant observations have allowed these equations to be tested robustly. This approach is more effective than "end-of-pipe" benchmarking of multi-component models, which rarely (if ever) allows problems with specific process representations to be diagnosed and resolved.
3) The "legacy" models of plant and ecosystem processes are essentially physical, with the role of biology limited to providing (over-)generalized parameter values for "plant functional types" (PFTs). The REALM approach puts biology at the centre by harnessing the power of natural selection, so that key adaptive processes need no longer be treated as add-ons that make models even more complex. Instead, they are integral to the way that REALM models represent plant and ecosystem function.
REALM has developed a universal theory and a parameter-sparse model for gross primary production, GPP (total ecosystem photosynthesis) that outperforms previous models despite its far greater simplicity and transparency. The model reproduces both diurnal and seasonal cycles and temporal trends in GPP at flux-measurement sites using the same parameter values for all biomes. Further new theory explains the "leaf economics spectrum" and successfully predicts how leaf economics vary along environmental gradients. Seasonal patterns of green vegetation cover, as observed by satellites, can also be predicted by a universal optimality principle; as can geographic patterns in the maximum allocation of biomass to leaves.
This new body of theory has improved understanding of the interaction of carbon (C) and nitrogen (N) cycles. Many models assume that N supply (from soil and N deposition) to plants determines leaf-level N and thence, photosynthetic capacity. But experiments show that N addition mainly affects biomass allocation, with a positive effect of leaf N but little or none on photosynthesis. This insight has led to (ongoing) research on the effects of phosphorus (P) addition. A comprehensive theory of C-N-P cycle interactions is within sight.
This revised understanding of plant-nutrient interactions has practical consequences. For example, the decline in leaf N over recent decades does not (as was once claimed) signify increasing N limitation of primary production. Instead, it follows from optimal resource use by plants, responding to warming and rising atmospheric CO2.
Further insights have been gained by coupling the new GPP model to a minimal representation of how heterotrophic respiration (primarily CO2 release from soil due to decomposition of dead organic matter) varies with the soil environment. The observed growth in the amplitude of seasonal variations in high-latitude atmospheric CO2, and the seasonal geographic pattern of net ecosystem exchange of CO2 between the atmosphere and land, are predicted accurately.
Three reasons have underpinned the success of REALM in developing improved ecosystem model components for ESMs.
1) A first-principles approach based on eco-evolutionary optimality hypotheses has generated clear and testable governing equations for the most important plant and ecosystem processes.
2) Huge increases in the amount and variety of relevant observations have allowed these equations to be tested robustly. This approach is more effective than "end-of-pipe" benchmarking of multi-component models, which rarely (if ever) allows problems with specific process representations to be diagnosed and resolved.
3) The "legacy" models of plant and ecosystem processes are essentially physical, with the role of biology limited to providing (over-)generalized parameter values for "plant functional types" (PFTs). The REALM approach puts biology at the centre by harnessing the power of natural selection, so that key adaptive processes need no longer be treated as add-ons that make models even more complex. Instead, they are integral to the way that REALM models represent plant and ecosystem function.
The following is a summary of key REALM findings and advances beyond the prior state-of-the-art:
• A universal theory for GPP, independent of PFTs, allows diurnal, seasonal and multiannual GPP variations to be well represented using far fewer parameters than current models.
• Plants and ecosystems adapted to more arid climates use water more conservatively – but can continue to function at lower soil moisture levels.
• Forests adapted to warm climates produce biomass more efficiently. This is consistent with optimality theory, but opposite to assumptions in current models.
• The allocation of carbon to leaves can be predicted from modelled GPP and rainfall, following a simple optimality principle.
• Leaves are displayed close to the time when they are most productive. This assumption allows seasonal foliage dynamics to be predicted accurately.
• The plasticity of leaf traits has been quantified based on extensive re-sampling of species along environmental gradients. Physiological traits show much greater plasticity than morphological traits.
• A unifying theory predicts the leaf economics spectrum, and how leaf mass-per-area (LMA) varies across environments. This theory also correctly predicts the divergent latitudinal trends of LMA in evergreen and deciduous plants.
• The main effect of N addition to ecosystems is not to increase leaf-level photosynthetic capacity, but to shift carbon allocation from below to above ground.
• The observed recent decline in leaf N is a predictable consequence of rising CO2 and warming. It does not imply increasing N limitation of photosynthesis.
• "Greening" trends have predominated over most of the land surface during recent decades. They are a predictable response to rising CO2 and climate change.
• Biomass production by crops follows the same principles as wild plants. A global model for wheat yields based on this insight outperforms existing models and correctly predicts yield responses to climate variation.
Dissemination of findings has included journal publications and a consistent presence at international meetings, most importantly the annual General Assembly of the European Geosciences Union. We have reached out to the land-surface modelling community in various other ways, including a published article showing that GPP and plant respiration as formulated by REALM can beneficially be incorporated in the Noah-MP model framework.
Two projects have built on the success of REALM. Schmidt Sciences now funds the multinational project Land Ecosystem Models based On New Theory, obseRvations and ExperimeEnts (LEMONTREE: 2021–2026: Colin Prentice is science strategy leader). LEMONTREE involves working with weather and climate modelling groups to trial the implementation of REALM-derived model components in their land-surface frameworks. The European Horizon programme funds the project Improved CarbOn cycle represeNtation through multi-sCale models and Earth observation for Terrestrial ecOsystems (CONCERTO: 2025–2029). One goal is the implementation of REALM process representations into several global land-surface models. Wildfire risk modelling within the Leverhulme Centre for Wildfires, Environment and Society (Colin Prentice is Director) is also using REALM model components.
We envisage that research conducted within REALM and its successor projects will transform the practice of global vegetation and land-surface modelling, including ecohydrological and agricultural aspects. Thereby, it will establish the foundations for a more robust understanding of the role of terrestrial ecosystems in Earth System dynamics and climate change.
• A universal theory for GPP, independent of PFTs, allows diurnal, seasonal and multiannual GPP variations to be well represented using far fewer parameters than current models.
• Plants and ecosystems adapted to more arid climates use water more conservatively – but can continue to function at lower soil moisture levels.
• Forests adapted to warm climates produce biomass more efficiently. This is consistent with optimality theory, but opposite to assumptions in current models.
• The allocation of carbon to leaves can be predicted from modelled GPP and rainfall, following a simple optimality principle.
• Leaves are displayed close to the time when they are most productive. This assumption allows seasonal foliage dynamics to be predicted accurately.
• The plasticity of leaf traits has been quantified based on extensive re-sampling of species along environmental gradients. Physiological traits show much greater plasticity than morphological traits.
• A unifying theory predicts the leaf economics spectrum, and how leaf mass-per-area (LMA) varies across environments. This theory also correctly predicts the divergent latitudinal trends of LMA in evergreen and deciduous plants.
• The main effect of N addition to ecosystems is not to increase leaf-level photosynthetic capacity, but to shift carbon allocation from below to above ground.
• The observed recent decline in leaf N is a predictable consequence of rising CO2 and warming. It does not imply increasing N limitation of photosynthesis.
• "Greening" trends have predominated over most of the land surface during recent decades. They are a predictable response to rising CO2 and climate change.
• Biomass production by crops follows the same principles as wild plants. A global model for wheat yields based on this insight outperforms existing models and correctly predicts yield responses to climate variation.
Dissemination of findings has included journal publications and a consistent presence at international meetings, most importantly the annual General Assembly of the European Geosciences Union. We have reached out to the land-surface modelling community in various other ways, including a published article showing that GPP and plant respiration as formulated by REALM can beneficially be incorporated in the Noah-MP model framework.
Two projects have built on the success of REALM. Schmidt Sciences now funds the multinational project Land Ecosystem Models based On New Theory, obseRvations and ExperimeEnts (LEMONTREE: 2021–2026: Colin Prentice is science strategy leader). LEMONTREE involves working with weather and climate modelling groups to trial the implementation of REALM-derived model components in their land-surface frameworks. The European Horizon programme funds the project Improved CarbOn cycle represeNtation through multi-sCale models and Earth observation for Terrestrial ecOsystems (CONCERTO: 2025–2029). One goal is the implementation of REALM process representations into several global land-surface models. Wildfire risk modelling within the Leverhulme Centre for Wildfires, Environment and Society (Colin Prentice is Director) is also using REALM model components.
We envisage that research conducted within REALM and its successor projects will transform the practice of global vegetation and land-surface modelling, including ecohydrological and agricultural aspects. Thereby, it will establish the foundations for a more robust understanding of the role of terrestrial ecosystems in Earth System dynamics and climate change.