Editing IPCC:AR6/WGII/Chapter-5 (section)

== 5.3 Methodologies and Associated Uncertainties ==

<div id="h1-4-siblings" class="h1-siblings"></div>

Chapter text draws on previous IPCC reports, other reports (i.e., High Level Panel of Experts (HLPE), Food and Agriculture Organization (FAO), and Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES)), and literature published since 2014. This section highlights key trends in research topics and methods since AR5.

<div id="5.3.1" class="h2-container"></div>

<span id="methodologies-for-assessing-impacts-and-risks"></span>
=== 5.3.1 Methodologies for Assessing Impacts and Risks ===

<div id="h2-6-siblings" class="h2-siblings"></div>

Since AR5, there are more examples of observed impacts from past climate change in cropping systems ( [[#5.4.1|Section 5.4.1]] ), pastoral systems ( [[#5.5.1|Section 5.5.1]] ), forests ( [[#5.6.1|Section 5.6.1]] ), fisheries ( [[#5.8.1|Section 5.8.1]] ) and mixed farming systems ( [[#5.10.1|Section 5.10.1]] ). These assessments of observed impacts make use of historical data on climate, production area and yield to attribute the role of climate in driving changes in suitability, production, yield, food quality or total factor productivity ( [[#Ortiz-Bobea--2021|Ortiz-Bobea et al., 2021]] ). Observations across the global food systems have been analysed ( [[#Cottrell--2019|Cottrell et al., 2019]] ), with the advantage that unexpected impacts due to changes in seasonality and biotic interactions can be detected. Quantitative analysis is only possible in places with adequate historical data; in many cases, studies rely on qualitative assessments, often drawing on farmers’ perceptions of climate impacts.

Projecting future climate impacts relies on modelling that combines climate data with data from experimental studies testing how species respond to each climate factor. In cropping and forest systems, a network of experimental studies with plants exposed to elevated CO 2 concentrations, ozone and elevated temperature provides data on the fundamental responses to climate and atmospheric conditions (i.e., free-air carbon dioxide enrichment (FACE) and temperature free-air controlled enhancement (T-FACE) systems). FACE results have been combined and assessed more extensively since AR5 ( [[#Bishop--2014|Bishop et al., 2014]] ; [[#Haworth--2016|Haworth et al., 2016]] ; [[#Kimball--2016|Kimball, 2016]] ; [[#Ainsworth--2021|Ainsworth and Long, 2021]] ). Field-based FACE studies have several advantages over more enclosed testing chambers, although results from more controlled experiments and coordination between different methods continue to give new insights into crop responses to climate change and variability ( [[#Drag--2020|Drag et al., 2020]] ; [[#Ainsworth--2021|Ainsworth and Long, 2021]] ; [[#Sun--2021|Sun et al., 2021]] ). Experimental results have limitations and can be difficult to scale up ( [[#Porter--2014|Porter et al., 2014]] ; [[#Haworth--2016|Haworth et al., 2016]] ), but generally the conclusions follow known plant responses ( [[#Lemonnier--2018|Lemonnier and Ainsworth, 2018]] ). As highlighted in AR5, there is a scarcity of FACE infrastructure in the tropics and subtropics ( [[#Leakey--2012|Leakey et al., 2012]] ; [[#Lemonnier--2018|Lemonnier and Ainsworth, 2018]] ; [[#Toreti--2020|Toreti et al., 2020]] ). One area that has been investigated further is the negative impact of elevated CO 2 on crop nutritional value, which has important implications for human nutrition ( [[#Scheelbeek--2018|Scheelbeek et al., 2018]] ; [[#Smith--2018|Smith and Myers, 2018]] ; [[#Toreti--2020|Toreti et al., 2020]] ; [[#Ainsworth--2021|Ainsworth and Long, 2021]] ). Increasingly, experimental studies seek to examine the interaction between climatic factors such as temperature, drought and ozone, or the responses of understudied food systems, crop species, cultivars and management interventions ( [[#Kimball--2016|Kimball, 2016]] ; [[#Ainsworth--2021|Ainsworth and Long, 2021]] ). The use of experimental data to improve projections has also expanded in other systems. There has been an increased focus on the impact of warming on livestock health and productivity (5.5.3) ''.'' Aquatic system studies have incorporated projected impacts on physiology, distribution, phenology and productivity (5.8.3).

Modelling approaches differ widely and serve different purposes (Table 5.2; [[#Porter--2014|Porter et al., 2014]] ; [[#Jones--2017a|Jones, 2017a]] ). The use of process-based and statistical modelling alongside remote sensing and other spatial data has grown. Projections increasingly draw on a combination of modelling approaches and coordinated efforts for model intercomparisons and ensemble techniques, using standardised emission scenarios (RCPs). For major crops, models of global yield impacts from CO 2 concentration, air temperature and precipitation have been refined and compared ( [[#Challinor--2014|Challinor et al., 2014]] ; [[#Iizumi--2017|Iizumi et al., 2017]] ; [[#Ruane--2017|Ruane et al., 2017]] ; [[#Zhao--2017|Zhao et al., 2017]] ; [[#Rojas--2019|Rojas et al., 2019]] ). Despite advances since AR5, modelling is still constrained by limited data from field experiments ( [[#Ruane--2017|Ruane et al., 2017]] ). Increasingly, studies attempt to incorporate effects of elevated CO 2 , ozone and climate extremes ( [[#Barlow--2015|Barlow et al., 2015]] ; [[#Schauberger--2019a|Schauberger et al., 2019a]] ; [[#Vogel--2019|Vogel et al., 2019]] ), as well as attempts to incorporate more complex interactions with soil and crop management ( [[#Basso--2018|Basso et al., 2018]] ; [[#Smith--2020b|Smith et al., 2020b]] ). However, only a few models consider crop protein content and other quality factors ( [[#Nuttall--2017|Nuttall et al., 2017]] ; [[#Asseng--2019|Asseng et al., 2019]] ). Some models take account of the impacts of climate on the timing of key biological events (phenology) in the target species; however, incorporating biotic interactions with pests, pathogens and pollinators remains a challenge (Table 5.2; Sections 5.4.1, 5.4.3).

'''Table 5.2 |''' A comparison of modelling approaches and their application in climate change impact projections. Model types are categorised by: food system, with labels representing the food systems from this chapter where each model type is used ( {CROP} , {TREE} , {LIVES} , {FISH} , {MIX} , '''{FOOD}''' ); scale over which each model type is usually applied local [()], regional [( )], global [( )], or a combination of these); and sensitivity to climate change where the colour intensity indicates the ability of each model type to incorporate each of the listed factors. After [[#Van%20Wijk--2014|Van Wijk et al. (2014)]] , [[#Kanter--2018|Kanter et al. (2018)]] and Thornton (2018). Integrated assessment models are discussed in the main text.

{| class="wikitable"
|-
|
| '''Description'''

| '''Applications for each food-system'''

| '''Scale'''

| colspan="5"| '''Sensitivity to climate change factors and responses'''

|-
| colspan="5"|
| Climate

| CO 2

| Biotic

| Adaptation

| System responses

|-
| rowspan="2"| '''Empirical'''

| Agroclimatic indices

| Use simple equations to link agricultural performance to key climate factors, such as drought or heat stress, or summarise agricultural requirements using multiple environmental descriptors.

| Comparing regions; matching crops to regions; early warning systems: e.g Agro-ecological zones, Ecocrop, Palmer Drought Severity Index {CROP} .

| (( ))

|
|-
| Statistical models

| Use quantitative associations between agricultural performance and climate, based on past observations. Can include projections for biotic factors such as pest and disease.

| Productivity and production area projections; annual climate variability; attribution: e.g. Traditional: regression, statistical emulators {CROP} {TREE} {LIVES} {FISH} ; e.g. Spatial suitability models /niche models: MaxEnt, CLIMEX, Ecocrop {CROP} {TREE} {FISH} .

| (( ))

|
|-
| rowspan="2"| '''Process-based (dynamic simulation models)'''

| Vegetation focussed

| Use combinations of land-surface energy and soil water balance models to simulate the growth of crop species along with natural vegetation, typically using plant and crop functional types.

| Productivity projections; interactions with non-climate variables (e.g. CO '''2''' ): e.g. PEGASUS, Agro-IBIS, DayCent, LPJmL, LPJ-GUESS, ORCHIDEE {CROP} {TREE} .

| ( )

|
|-
| Species focussed

| Use mechanistic models based on the known responses of species to key environmental descriptors over time. Typically based on detailed information for a particular species within a region, but also applied to mixed systems such as agroforestry and globally.

| Productivity projections; matching tree species to locations; species interactions; interactions with non-climate variable s (e.g.CO 2 ); adaptation projections: e.g. point-based versions: APSIM, AquaCrop, DayCent, DSSAT, EPIC, Infocrop, SARRA-H, STICS {CROP} IBIS {TREE} LIVSIM, RUMINANT {LIVES} Fish-MIP {FISH} Yield-SAFE, WaNuLCAS, Hi-sAFe {MIX} ; e.g. global gridded version: pDSSAT, pAPSIM, GEPIC, GLAM, MCWLA, PEGASUS, SARRA-O {CROP} .

| (( ))

|
|-
| rowspan="3"| '''Integrated Models'''

| Optimization methods

| Mathematical representations of systems with regard to key indicators, constraints, and objectives. Allows prioritisation of different climate change response options using the defined indicators.

| Adaptation projections; food security projections; livelihood projections; trade-offs; live cycle assessment: e.g. Global Timber Model {TREE} CSAP toolkit, FarmDESIGN {CROP} {MIX} '''{FOOD}'''

| ()

|
|-
| Economic (Econometric, Economic surplus)

| Used to integrate the broad impacts of climate change with other economic drivers, to quantify the economic costs and assess the value of adaptation/mitigation interventions.

| Adaptation projections; food security projections; livelihood projections: e.g. GFPM {TREE} FUND 3.8, DICE 2010, IMPACT '''{FOOD}'''

| ( )

|
|-
| Household and village models

| Use detailed site-specific data to generate rules that describe the current behaviour of stakeholders such as households or villages. Can be integrated with other model approaches to consider climate response and adaptation interventions.

| Adaptation projections (case specific); behavioural responses; trade-offs; participatory monitoring: e.g. DECUMA, PALM, MPMAS, MIDAS, TOA-MD {LIVES} {MIX} '''{FOOD}'''

| (())

|
|}

In addition to productivity projections, research also draws on climate suitability estimates (Table 5.2). These compare the known climate suitability of species and habitats with projected climate conditions across different locations. Such projections are useful especially for incorporating movement of pests and pathogens but cannot be applied in isolation if non-climate constraints are not considered. As different research groups use different assumptions and data inputs, more coordination is needed if suitability projections are to be compared globally (SM5.3).

Increasingly, projections look across different disciplines and across multiple components of the food system, including livestock, fisheries and mixed farming systems ( [[#Campbell--2016|Campbell et al., 2016]] ; [[#Mbow--2019|Mbow et al., 2019]] ). Major timber species have been modelled, with projected impacts on productivity, duration of rotation and distribution (i.e., climate suitability) ( [[#Albert--2018|Albert et al., 2018]] ). Livestock systems are influenced by plant productivity projections via their feedstock, for example, rangeland cattle impacted by changes in net primary production (NPP) ( [[#Boone--2018|Boone et al., 2018]] ). Direct climate impacts on animals are also projected, using indices based on direct observations ( [[#5.5.3|Section 5.5.3]] ). Since AR5, Fisheries and Marine Ecosystem Model Intercomparison Project (Fish-MIP) has allowed for global intercomparisons and ensemble projections of marine fisheries, and projections capturing interactions from multiple food systems (e.g., Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP); Sections 5.8, 5.10).

Global simulations have uncovered important differences between regions ( [[#Deryng--2016|Deryng et al., 2016]] ; [[#Blanchard--2017|Blanchard et al., 2017]] ). Efforts to coordinate and combine regional and global modelling studies allow for greater insight into regional differences in climate change impacts, such as the Coordinated Global and Regional Assessments (CGRA) performed by the Agricultural Model Intercomparison and Improvement Project (AgMIP) ( [[#Blanchard--2017|Blanchard et al., 2017]] ; [[#Müller--2017|Müller et al., 2017]] ; [[#Rosenzweig--2018|Rosenzweig et al., 2018]] ; [[#Ruane--2018|Ruane et al., 2018]] ; [[#Lotze--2019|Lotze et al., 2019]] ). Increasingly, multi-model intercomparisons are used to evaluate global gridded crop models’ performance and sensitivity to temperature, water, nitrogen and CO 2 within AgMIP, with the focus mostly on major annual crops ( [[#Valdivia--2015|Valdivia et al., 2015]] ; [[#Ruane--2017|Ruane et al., 2017]] ; [[#Müller--2021a|Müller et al., 2021a]] ). Differences in model type, structures and input data can result in large variation in projections, particularly for the response of crops to elevated CO 2 and temperature (5.4.3.1); methods for quantifying and minimising this uncertainty have been developed, but improvement is still needed ( [[#Asseng--2015|Asseng et al., 2015]] ; [[#Li--2015|Li et al., 2015]] ; [[#Zhao--2017|Zhao et al., 2017]] ; [[#Folberth--2019|Folberth et al., 2019]] ; [[#Tao--2020|Tao et al., 2020]] ; [[#Müller--2021a|Müller et al., 2021a]] ; [[#Ruane--2021|Ruane et al., 2021]] ). The use of multi-model intercomparisons has widened the range of uncertainties but has increased the robustness of impact assessments ( [[#Asseng--2013|Asseng et al., 2013]] ; [[#Challinor--2014|Challinor et al., 2014]] ; [[#Zhao--2017|Zhao et al., 2017]] ). Model outputs are strongly influenced by decisions over which factors to include; for example, including drought impacts can result in positive yield projections switching to neutral or negative values ( [[#Gray--2016|Gray et al., 2016]] ; [[#Jin--2018|Jin et al., 2018]] ). Models are also limited in their ability to incorporate socioeconomic drivers and extreme events ( [[#Porter--2014|Porter et al., 2014]] ; [[#Campbell--2016|Campbell et al., 2016]] ; [[#Ruane--2017|Ruane et al., 2017]] ; [[#Jagermeyr--2018|Jagermeyr and Frieler, 2018]] ; [[#Webber--2018|Webber et al., 2018]] ; [[#Schewe--2019|Schewe et al., 2019]] ).

For long-term projections and integrated assessments, a large component of uncertainty remains in the ability to represent socioeconomic responses to climate change and the degree to which these will mitigate or exacerbate climatic changes ( [[#Valdivia--2015|Valdivia et al., 2015]] ; [[#Prestele--2016|Prestele et al., 2016]] ; [[#Arneth--2019|Arneth et al., 2019]] ). This includes the potential adaptation responses of food producers. Models that incorporate alternative socioeconomic responses offer one solution (e.g., AgMIP) ( [[#Nelson--2014|Nelson et al., 2014]] ; [[#Von%20Lampe--2014|Von Lampe et al., 2014]] ; [[#Wiebe--2015|Wiebe et al., 2015]] ; [[#Rosenzweig--2018|Rosenzweig et al., 2018]] ; [[#van%20Zeist--2020|van Zeist et al., 2020]] ). Another approach is the use of solution-oriented scenarios to compare the effectiveness of adaptation options ( [[#Le%20Mouël--2017|Le Mouël and Forslund, 2017]] ; [[#Arneth--2019|Arneth et al., 2019]] ), or to quantify the time period in which adaptation responses will become essential ( [[#Challinor--2016|Challinor et al., 2016]] ; [[#Rojas--2019|Rojas et al., 2019]] ). Others point to the necessity of managing food systems within the context of uncertainty ( [[#Campbell--2016|Campbell et al., 2016]] ).

<div id="5.3.2" class="h2-container"></div>

<span id="methodologies-for-assessing-vulnerabilities-and-adaptation"></span>
=== 5.3.2 Methodologies for Assessing Vulnerabilities and Adaptation ===

<div id="h2-7-siblings" class="h2-siblings"></div>

Methods for monitoring vulnerability and adaptation are under-researched but have increased since AR5. Increasingly, projections move from individual crops to assessing risks across the food systems and the relative vulnerability of different systems ( [[#Campbell--2016|Campbell et al., 2016]] ; [[#Gil--2017|Gil et al., 2017]] ; [[#Lipper--2017|Lipper et al., 2017]] ; [[#Richardson--2018|Richardson et al., 2018]] ). Adaptation options can be considered as parameters in integrated models, such as those used in ISI-MIP, while others use systematic assessments of case studies, such as the application of agent-based household models to assessments of adaptation in livestock systems ( [[#5.5.4|Section 5.5.4]] ). Quantitative studies are less common than qualitative assessments, and there is a need to combine modelling and qualitative approaches more effectively ( [[#Beveridge--2018a|Beveridge et al., 2018a]] ; [[#Vermeulen--2018|Vermeulen et al., 2018]] ).

The food system is dynamic, with changes in management practices driven by many factors, including climate adaptation ( [[#Iizumi--2019|Iizumi, 2019]] ; [[#Iizumi--2021a|Iizumi et al., 2021a]] ). Adaptation potential, such as expected advances in crop breeding, are often not explicitly accounted for in modelling studies, but more recent studies do quantify the potential for adaptation ( [[#Iizumi--2017|Iizumi et al., 2017]] ; [[#Tao--2017|Tao et al., 2017]] ; [[#Aggarwal--2019|Aggarwal et al., 2019]] ; [[#Minoli--2019|Minoli et al., 2019]] ). To account for this complexity, case studies rely on data derived from the perception and practices of stakeholders who are engaged in adaptation (usually autonomous adaptation) ( [[#Hussain--2016|Hussain et al., 2016]] ; [[#Lipper--2017|Lipper et al., 2017]] ; [[#Ankrah--2018|Ankrah, 2018]] ; [[#Sousa-Silva--2018|Sousa-Silva et al., 2018]] ). Case studies use a range of different indicators to monitor climate response options, making quantitative comparisons more difficult ( [[#Gil--2017|Gil et al., 2017]] ; [[#Vermeulen--2018|Vermeulen et al., 2018]] ). However, systematic comparisons have provided valuable insights ( [[#Descheemaeker--2018|Descheemaeker et al., 2018]] ; [[#Shaffril--2018|Shaffril et al., 2018]] ; [[#Aggarwal--2019|Aggarwal et al., 2019]] ; [[#Bene--2019|Bene et al., 2019]] ); for example, the sustainable livelihood framework has been applied widely to diverse aquatic systems ( [[#Bueno--2017|Bueno and Soto, 2017]] ; [[#Barange--2018|Barange and Cochrane, 2018]] ) and the Livelihood Vulnerability Index is well used across systems ( [[#5.1|Section 5.1]] 4). Coordinated efforts such as the AgMIP also provide systematic assessments ( [[#Blanchard--2017|Blanchard et al., 2017]] ; [[#Lipper--2017|Lipper et al., 2017]] ; [[#Antle--2018|Antle et al., 2018]] ). Nonetheless, the full effectiveness of different adaptation options is difficult to assess given that many impacts have not yet occurred (due to the cumulative nature of impacts and the inertia in the climate system) ( [[#Stocker--2013|Stocker et al., 2013]] ; [[#Zickfeld--2013|Zickfeld et al., 2013]] ).

Transformation of the food system that addresses all dimensions of ecosystem services is discussed in this chapter, including risk management and the communication of uncertainties ( [[#5.1|Section 5.1]] 4). The focus is on flexible approaches to risk and uncertainty, assessing trends, drivers and trade-offs under different future scenarios ( [[#Campbell--2016|Campbell et al., 2016]] ).

<div id="5.4" class="h1-container"></div>

<span id="crop-based-systems"></span>