Evaluating climate-driven agricultural suitability models for adaptation planning: stakeholder insights from British Columbia

1 Introduction

Climate change is reshaping agricultural systems, landscapes, and production geographies. Warmer temperatures and longer frost-free periods are expanding the potential range of some crops, while water scarcity, heat stress, extreme events, and seasonal unpredictability threaten production in currently active agricultural regions (IPCC, 2022; Government of British Columbia, 2025a; Li et al., 2026). These changes affect where, when, and how crops can be grown, which presents a need for adaptation strategies that connect climate information with agricultural practice and policy (Agriculture and Agri-Food Canada, 2020; Grigorieva et al., 2023; Hu et al., 2024).

British Columbia’s agricultural sector is economically important and geographically diverse. The 2021 Census of Agriculture reported 15,841 farms occupying approximately 5.65 million acres in the province, while food and beverage manufacturing generates more than $13 billion in annual revenue and supports over 40,000 jobs (Statistics Canada, 2022; Government of British Columbia, 2025b). Climate-related changes in agricultural suitability therefore have implications beyond individual farms, including implications for rural development, food processing, transportation, labor, investment, land-use planning, and regional food security.

In British Columbia, agriculture ranges from intensive production in the south coast and interior valleys to livestock, forage, and other agricultural systems in northern and remote regions. Much of the province’s farmland is safeguarded within the Agricultural Land Reserve (ALR), but legal protection alone does not guarantee long-term agricultural viability. Shifts in temperature, precipitation patterns, and growing-season conditions may create opportunities in some regions while increasing heat and water stress in others (Government of British Columbia, 2025a).

Canada’s plant-hardiness mapping already combines several climate indicators, including temperature, precipitation, and frost duration, to describe conditions relevant to plant survival (Natural Resources Canada, 2026). Agriculture and Agri-Food Canada (2020) similarly identifies longer growing seasons, additional frost-free days, and possible northward expansion of warmer-weather crops where soil conditions permit. Such information is valuable for anticipating geographical change, but climatic opportunity does not by itself establish whether production will be technically, economically, or institutionally viable.

Spatial crop-suitability models integrate climate projections with land, soil, and other variables to identify areas that may become more or less suitable under alternative emissions scenarios. Their outputs can support scenario exploration and long-term planning; however, a model is only a useful decision-support resource when its outputs are understandable, credible, and connected to the decisions users actually face (Newell et al., 2021; Rose et al., 2018; Ikendi et al., 2025). Stakeholder evaluation is therefore important for understanding not only whether users can operate a model interface, but also how they interpret its assumptions, uncertainties, applications, and risks.

Although studies increasingly emphasize user-centered climate services and agricultural decision support, less attention has been given to how diverse end users interpret regional crop-suitability projections after model development. Using the Canadian province of British Columbia as a question, the present study addresses the question: how do stakeholders perceive the strengths, limitations, and potential uses of regional agricultural suitability models for supporting climate-adaptation planning, policy, and practice in British Columbia?

2 The model

The study evaluates the practical and decision-support value of a set of future agricultural suitability models developed for British Columbia. The modelling methods and predictive results are reported in detail by Vakhshoori et al. (2025). This section summarizes the model features to provide context for the stakeholder focus group outcomes and analysis.

The models assess the future suitability of agricultural land for seven economically and dietarily important field-grown crops such as cabbage, cauliflower, lettuce, strawberry, kale, broccoli, and celery. Baseline suitability was estimated using historical crop planting and production, local and regional climate conditions, land-capability classifications, and an economic indicator based on regional property values. Machine-learning models were then used to project suitability from 2021 to the end of the 21st century under different climate change scenarios. The climate change scenarios were based on the Intergovernmental Panel on Climate Change’s (IPCC) Shared Socioeconomic Pathways (SSP) framework, and data from climate projections for the SSP2, SSP3, and SSP5 scenarios were used in the modelling work (Figure 1).

The model results revealed future crop-specific spatial shifts. Cabbage, cauliflower, and lettuce showed northward expansion under warmer conditions, particularly under SSP3 and SSP5. Strawberry, kale, and broccoli showed expansion under moderate-to-high warming but losses in total suitable agricultural land later in the century under SSP5. Celery showed a consistent decline across scenarios, although the reliability of this result was affected by limited data on current provincial cultivation.

The suitability models produce regional spatial projections, which were displayed using the Agrilyze platform.¹ Agrilyze provides an interactive interface through which spatial data can be explored as a decision-support resource. Users can compare crops, locations, years, and climate scenarios rather than treating model outputs as a single forecast (Figure 2). It is important to clearly articulate that the model estimates climatic and land-based suitability, whereas the decision-support potential emerges from how model outputs are communicated, interpreted, and combined with other information. This study specifically focuses on the decision-support potential of the model data and outcomes as it is displayed on the Agrilyze platform (see www.triaslab.ca/agrigenomics#model).

For the projections to inform practice, model outputs must be presented in ways that are accessible to farmers, planners, governments, consultants, and other potential users. The focus groups examined both the substantive implications of projected suitability change and participants’ experiences interpreting and applying the outputs through Agrilyze.

3 Methods

3.1 Study design, recruitment, and participants

Two online focus groups were conducted on October 15 and 16, 2025, to collect stakeholder feedback on the strengths, weaknesses, and potential applications of the agricultural suitability models and maps developed by Vakhshoori et al. (2025). Participants were purposively sampled, and recruitment was done through targeted email invitations. Eligibility required participants to be based in British Columbia and actively involved in farming or an agriculture-related role, including research, planning, policy, consulting, community work, agri-technology, or private industry.

A total of 19 participants took part across the two sessions. The sample represented a range of professional sectors, including farming, provincial and regional government, academic and research institutions, agricultural and environmental consulting, community organizations, agri-technology and innovation organizations, and private industry. Note that multiple participants held more than one role and affiliation; thus, the participant affiliations reported in Table 1 sum to n = 22, rather than n = 19. This diversity supported discussion among subject-matter experts and potential end users involved in agricultural production, research, land-use planning, and policy development.

Table 1. Number and sectoral affiliations of participants.

The total number of affiliations reported is 22, reflecting overlapping affiliations. Two participants categorized as farmers were also affiliated with government and public administration, and one farmer was also employed in the private sector.

The study received institutional research ethics approval. Participants reviewed and signed a letter of informed consent at the beginning of each session. All discussions were recorded with participant consent.

3.2 Focus-group procedure and data generation

Each session lasted approximately 2 h and were held by Zoom. Each focus group began with a presentation on the project’s objectives and model framework, followed by a demonstration of the crop-suitability outputs in Agrilyze. Participants were shown how to explore mapped suitability, layers, locations, years, and climate scenarios, and then, they engaged in two activities. Activity 1 asked participants to examine a region with which they were familiar and identify projected changes for selected crops. This exercise assessed whether the model communicated future trends clearly and whether participants could develop insights relevant to their experience. In Activity 2, participants discussed the implications of projected opportunities and threats.

Each session followed a set agenda and schedule of activities. The sessions consisted of a five-minute welcome, 5 min for the agenda and consent process, a 15-min introduction to the report and research purpose, a 15-min explanation of the modelling methodology and Agrilyze demonstration, 20 min for Activity 1, a 5-min break, 20 min for Activity 2 in two breakout rooms, a 10-min plenary discussion, and 10 min for closing reflections. The data-generation components of the sessions (i.e., Activities 1 and 2, the plenary discussion, and closing reflections) accounted for approximately 60 min of the total session.

Four project-team members facilitated the sessions. During the breakout activity, two facilitators joined each room. Facilitators used open-ended questions and prompts to promote discussion, and follow-up questions were used when clarification was needed. Participants were provided with multiple opportunities to share thoughts, ideas, and perspectives during the activities, breakout discussion, plenary discussion, and closing reflections. The breakout discussions were organized around four guiding questions:

• What forms of support would farmers need in areas where agricultural suitability may emerge or increase?
• What infrastructure, resources, and planning would communities need in or near areas of emerging suitability?
• What roles could agri-technologies play in supporting farmers in areas of declining agricultural suitability?
• What forms of agri-technology capacity could support communities located in or near areas of declining suitability?

After the breakout sessions, participants reconvened to discuss the model’s usability, interface, communication of results, limitations, and decision-support value. As done with the breakout groups, guiding questions were used for the full-group discussion. Participants provided open-ended reflections, which generated insights that both related to the guiding discussion questions and extended beyond the discussion questions. The questions used to guide the full-group discussion were as follows:

• How could the maps and models inform the policies or programs you work on?
• What information would be most useful to add, or what do you feel is missing?
• Which features of the data platform did you find most useful, and what new features would you like to see?

3.3 Transcription and thematic analysis

Audio recordings were transcribed using Trint transcription software and analyzed in NVivo 15. The first author checked the transcripts and led the coding. Other project-team members subsequently reviewed the coding and emerging category structure through research team meetings. Analysis followed an iterative thematic approach informed by Braun and Clarke (2021), combining deductive organization with inductive coding. As a form of member checking, a report summarizing the workshop findings was circulated to all participants for their review and feedback. Participants were given a three-week review period to comment on whether the report accurately represented their views and to provide corrections or additional feedback.

The focus group guiding questions were used to develop the deductive coding framework. Specifically, breakout question 1 and 2 informed an ‘emerging agricultural suitability and opportunity’ code, breakout question 3 and 4 informed an ‘agricultural vulnerability, decline, and adaptation capacity’ code, and the full-group questions informed a ‘model use, limitations, and governance’ code. Inductive coding was then used to identify specific issues, participant perspectives, and contextual factors within and beyond the deductive codes. During revision, related emergent codes were compared and combined into higher-order interpretive themes that directly address the study’s research question. The final presentation of results is organized using the three deductive categories: emerging agricultural suitability and opportunity; agricultural vulnerability, decline, and adaptation capacity; and model use, limitations, and governance.

3.4 Researcher roles and credibility

Multiple procedures were used to support the credibility of the analysis. Transcripts were double-checked before coding. The code structure was reviewed by the research team through multiple meetings. Codes were revised through an iterative coding process to ensure thematic outputs accurately captured trends and patterns in the data.

The first author conducted the primary analysis, while other project-team members checked the coding and reviewed the relationships among categories, themes, and conclusions. This collaborative review was used to prevent overly narrow interpretations of the data and to ensure that the analysis remained grounded in the transcript data.

Participants are identified in the supplementary quotations using focus-group and participant codes (for example, FG1-P2). Sector affiliations are reported in aggregate by professional sector because several participants held overlapping roles, and due to the relatively interconnected agricultural community in British Columbia, specifically reporting a participant’s multiple affiliations could make them identifiable. Additionally, as this is a qualitative study with a small sample size, participant/stakeholder differences are interpreted in terms of the professional and/or lived experience context in which they are presented (for example, direct farming experience, planning practice, or rural and remote experience), rather than analyzed as statistically comparable subgroups.

4 Results

The findings are organized into three deductive coding categories and eight interpretive themes (Table 2). The interpretive themes are produced through a synthesis of the emergent (i.e., inductive) codes, and the themes represent the analytical outputs of the study that address the research question. Coded-reference counts are provided as descriptive metrics for the analysis, but they are not presented as statistical measures of concept prevalence or importance. Illustrative quotations from the data that support the analysis are presented in Supplementary material S1.

Table 2. Analytical structure of the focus-group findings.

Coded-reference counts describe the distribution of discussion in the transcripts and should not be interpreted as statistical measures of importance.

4.1 Emerging agricultural suitability and opportunity

4.1.1 Projected climatic suitability becomes meaningful only when coupled with crop, water, and production feasibility

Participants recognized that warmer conditions and longer growing seasons could enable crop diversification, new crop varieties, and expanded production in some regions. However, they consistently distinguished climatic suitability from practical viability. A location shown as suitable for a crop could remain unsuitable in practice if soil, irrigation, production costs, or market conditions were unfavorable. The model was viewed as an initial indication of opportunity rather than a comprehensive agricultural feasibility assessment.

Those speaking from direct farming experience emphasized the importance of considering crop-specific and farm-level constraints. Water access was highlighted in this discussion, with participants identifying that increased annual precipitation would not necessarily support production if water arrived outside the growing season or could not be stored and legally allocated. Participants also noted that successful crop or variety changes require experimentation, technical support, and funding that reflects future climate conditions. This was considered to be particularly important for perennial crops, where investment horizons are long and unsuitable crop choices may create losses for many years.

Economic feasibility further qualified the model results. Participants noted that crops may be imported because of production costs and supply-chain advantages, not only because local climate conditions are unsuitable. They therefore sought regional information on expected yields, costs, water requirements, and market access. Taken together, these perspectives show that suitability projections reduce uncertainty most effectively when they are embedded within a broader assessment of agronomy, resources, and farm business conditions.

4.1.2 Harnessing emerging opportunities reveal through the model requires coordinated regional infrastructure, markets, and policy over long decision horizons

Participants indicated that the model elucidated potential opportunities in northern and interior regions in British Columbia, including the Peace Region, but they also stressed that regional agricultural expansion would require more than changes at individual farms. Irrigation and water-storage systems, transportation, energy, packing, processing, grading, freezing, and access to markets would all shape whether emerging climatic suitability could support a durable regional agri-food economy. Long distances, cold-weather energy costs, and limited processing capacity were identified as particular barriers in northern and remote contexts.

The timing of projected change also affected the perceived usefulness of the model. Farmers may make decisions within personal, business, or intergenerational horizons that do not align with end-of-century projections. Participants therefore wanted clearer information on the pace of change and near-term regional trends. For communities and governments, however, longer time horizons were considered valuable because water systems, processing infrastructure, land-use policy, and workforce development require proactive, long-term planning.

Planning-oriented comments from the participants framed the model as a basis for anticipating geographic shifts in production, population, investment, and rural economic activity. Participants considered such foresight useful for aligning infrastructure and policy before opportunities become urgent. In this interpretation, the model’s value lies less in directing an immediate crop choice than in helping regional actors coordinate long-term transitions and avoid costly, reactive responses.

4.2 Agricultural vulnerability, decline, and adaptation capacity

4.2.1 Regional projections align with participants’ observations of climate pressures but cannot represent the full range of farm-level adaptation

Participants in areas projected to lose suitability described the maps of model outcomes as consistent with experienced heat, prolonged summer dryness, and growing dependence on irrigation. For some, the projections reinforced concern about the future of local agriculture. At the same time, participants resisted interpreting regional decline as the end of production. They described climate change adaptation in terms of altered planting schedules, overwintering crops, extended shoulder-season production, shade cloth, tunnels, frost protection, soil management, and night-interruption lighting.

These participant comments reveal a difference between adaptation responses to regional climatic patterns and the adaptive capacity and responses occurring at the individual farm level. Local topography, frost drainage, moisture, day length, management skill, and access to protective infrastructure can create growing conditions and opportunities that differ from what regional scale maps present. Participants therefore regarded the model as capable of confirming broad pressure while remaining unable to represent the full range of localized practices through which producers respond.

Agri-genomics and cultivar breeding were discussed as longer-term approaches to temperature, variability, and day-length constraints. Participants nevertheless placed equal or greater emphasis on practical, low-energy technologies that could be used in open-field or protected outdoor production. This suggests that adaptation pathways should not be reduced to high-technology systems, and their relevance depends on crop, farm scale, local skill, energy requirements, and the nature of the climate constraint.

4.2.2 The model does not capture unevenness in climate adaptation capacity due to economics, land use, and infrastructure considerations

Participants described climate vulnerability as inseparable from land values, farm income, infrastructure, and access to capital. Smaller producers may have limited ability to absorb repeated climate losses or experiment with unfamiliar crops and technologies, while larger operations and investors may be better positioned to act on long-term projections. In high-growth regions, participants viewed high land values and broader development pressures as factors shaping agricultural viability alongside climate suitability.

Several comments connected these economic pressures to the structure of the food system. Participants argued that storage, processing, and distribution infrastructure often serves large and global supply chains rather than local and regional production. Without coordinated investment, farms in vulnerable areas may have few pathways to diversify, process, or market products locally. Infrastructure and policy therefore shape not only whether adaptation is technically possible, but also who can afford to pursue it and which communities benefit.

Participants also linked urban development to hydrological change, soil degradation, and the loss of ecosystem services on adjacent agricultural land. Riparian restoration, hedgerows, and other green infrastructure were proposed as possible buffers, although participants noted that evidence of effectiveness is uneven. These findings frame adaptation as an equity and governance challenge. Modelled exposure may be regional, but the resources required to respond are distributed unevenly among producers and places.

4.2.3 The model does not capture how agricultural transitions require place-specific technologies and coordinated long-term planning

The future agricultural suitability model can help identify where transition pressures may emerge, but it cannot determine which technological or development pathway will be appropriate. In a discussion on agricultural technologies, participants did not view vertical farming, controlled-environment agriculture, or advanced sensing as universally applicable climate adaptation solutions. Their viability was considered strongly dependent on location, energy, population density, infrastructure, and markets. Urban and institutional settings were described as potentially favorable for such technologies because production can be located close to consumers and may connect with waste heat, wastewater, or existing energy systems. Rural and remote participants were more cautious, emphasizing limited energy access, small markets, and high operating costs.

Long-term adaptation was consequently understood as a coordinated, place-based process rather than a simple substitution of technology for declining climate suitability. Participants discussed the possible movement of farms, processing systems, and agricultural investment between regions, but noted that relocation and transition require institutional support, land-use planning, infrastructure, and long-term policy continuity.

4.3 Model use, limitations, and governance

4.3.1 The model is most useful for exploration, communication, and regional planning rather than site-specific prescriptions

Participants identified the value of using the model to compare scenarios, years, crops, and regions. This temporal and spatial exploration made gradual changes visible and supported discussion of emerging opportunities, declining conditions, and possible research needs. Agricultural consultant and planner participants indicated that the model is useful for early-stage resource assessment, regional feasibility work, community planning, and place-based communication about climate futures. The same participants also cautioned against applying regional outputs directly to individual properties. Microclimates, frost drainage, elevation, hydrology, salinity, nutrient availability, pH, and farm-management practices can create substantial variation within a mapped area.

Participants identified discrepancies where crops were visibly grown outside the ALR (as per their own observations) but did not appear in the model’s baseline data. This observation reflects data limitations of the model, that is, the model only covered ALR areas. Such observations support using the model to formulate questions and identify broad trends in agricultural suitability, which must then be followed by local, place-based assessments and producer knowledge.

The model was noted to value as a knowledge mobilization tool. Participants described how mapped scenarios could support place-based storytelling, communicate that climate change creates both risks and opportunities, and help identify research needs, coordinate data collection, and improve the model’s predictive capacity. The model’s usefulness therefore extends beyond making a single decision. It can provide a shared object around which producers, planners, researchers, and communities examine alternative futures.

4.3.2 Model credibility depends on transparent uncertainty and iterative, locally informed improvement

Participants expressed a desire for clearer communication of model assumptions, confidence, and spatially varying accuracy. A high-level province-wide performance measure was not considered sufficient in terms of communicating accuracy, as such a measure does not reveal the reliability of the model at a particular location. Suggested improvements included visual indicators of model performance and uncertainty, concise explanations of what the model does and does not include, and regional summaries of key agricultural suitability highlights for the next five to 10 years.

Participants proposed expanding the model to include irrigation demand, sea-level rise, flooding, saltwater intrusion, ecosystem services, protective technologies, land-use change, genomics, perennial crops, and region-specific production systems. They also wanted flexible suitability settings for crops and varieties. These additions cannot all be incorporated into a single model without increasing complexity, but the suggestions identify priorities for complementary layers, linked tools, and future data collection.

Model improvement was described as an iterative and collaborative process. Participants saw opportunities for producers, researchers, planners, and institutions to contribute local observations and data, allowing the model to evolve with changing agricultural systems. This approach to model improvement shifts the modelling process from developing a finished product to develop a product with assumptions, evidence, and interface that can be revised in response to user experience and feedback.

4.3.3 Governance safeguards are needed to prevent exploratory projections and model outcomes from being used as deterministic evidence

A particularly important concern was the potential use of projected suitability decline to support the conversion of agricultural land to urban or other non-agricultural uses. Participants noted that a regional model could be interpreted as evidence that future farming is impossible, even where rehabilitation, adaptation, alternative crops, or localized conditions could sustain production. Such use would frame what is a conditional projection into a deterministic outcome that can be (mis)used for land use development purposes.

Responsible application therefore requires explicit statements about scale, uncertainty, excluded variables, and appropriate uses. The participant concerns were not meant to be taken as an argument against making the model public. Rather, it showed that transparency must be accompanied by governance-related communications, and model outputs should be combined with local evidence, planning safeguards, and professional judgement when they inform consequential land-use or investment decisions.

5 Discussion

This study examined how stakeholders perceive the strengths, limitations, and potential uses of regional agricultural suitability models for climate-adaptation planning, policy, and practice. The findings show that the model’s principal value is in developing high-level strategies and policy directions, rather than in prescribing specific farming practices. It can help users explore scenarios, identify spatial and temporal change, and initiate discussion about emerging risks and opportunities. Its outputs become actionable only when interpreted alongside water, soils, infrastructure, markets, land use, governance, and local knowledge.

The findings of this study are consistent with broader work on climate decision support. Findlater et al. (2021) argue that climate services often concentrate on improving data while paying less attention to the institutional and decision contexts in which information is used. Similarly, Ikendi et al. (2025) found that stakeholders identified data translation, visualization, community engagement, and training as being critical for the usefulness of agricultural climate support tools. Similarly, research using interactive food-system visualizations in Revelstoke demonstrated that visualizations can help stakeholders explore possible futures, while the representation of local infrastructure, the completeness of the visualized environment, and the plausibility of scenarios influence how those futures are interpreted (Newell et al., 2024). The findings from the current study extend these observations to crop-suitability modelling. A technically credible regional projection does not become decision relevant through visualization alone. Users also need explanations of scale and uncertainty and connections to the practical constraints that determine whether adaptation is feasible.

The findings also highlight the timing of stakeholder involvement. Participants evaluated the model after its initial development, which provided useful evidence about interpretation, missing variables, interface design, and possible misuses. However, several identified limitations (particularly those involving local conditions, decision horizons, and regional summaries) might have been addressed earlier through co-design. Longman et al. (2026) demonstrate how user-informed development can connect climate tools with site-specific producer needs, while Jacob et al. (2025) emphasize that co-production depends on sustained relationships, translation, and institutional capacity. Retrospective evaluation remains valuable, but an iterative process that engages intended users before, during, and after model development is more likely to produce decision-relevant outputs.

A second contribution of this study centers on the relationship between spatial scale and responsibility. Participants accepted that regional models cannot reproduce every farm-level condition, yet they express a need for communicating clear boundaries (and limitations) around how the outputs should be interpreted. This is especially important when model results could influence perceptions of agricultural viability or be cited in infrastructure planning and ALR-related land-use applications. Participants noted concerns that projections might be used to support the conversion of agricultural land demonstrates that uncertainty communication is not only a technical matter but also a governance requirement. Statements of appropriate use, visible uncertainty, local validation, and safeguards against deterministic interpretation should therefore be treated as part of model design.

The participant comments complicate an otherwise simple narrative that can be derived from the model, that is, agricultural suitability will shift from declining southern areas to newly suitable northern regions. As noted by participants, new climatic opportunities do not automatically result in farms, agricultural supply chains, water systems, labor, and/or markets. Additionally, declining climatic suitability does not eliminate producers’ capacity to adapt through seasonal shifts, crop changes, protective systems, restoration, or breeding. Regional adaptation planning should therefore consider emerging opportunities and vulnerabilities presented through the model within a rich context of whether (and which) producers and communities have the resources to act and adapt.

In terms of contributions to practice, the research findings support a ‘layered’ decision-support approach. The regional model can serve as an exploratory first layer. The second (more detailed) layer involves regional summaries, uncertainty information, and infrastructure or water overlays can support planning. The third layer involves site-level assessments and producer knowledge for informing farm decision and practices. This approach positions the model as useful tool for informing agricultural adaptation, but not the only tool (or technique). The layer approach can also be used to inform incorporation of future data into the model, and it supports an argument that the model is best viewed as an evolving tool (i.e., not a finished or comprehensive representation of agricultural viability).

Several study limitations should be considered. The sample was relatively small and recruited through targeted email invitations, creating potential self-selection and network bias. Participants held overlapping affiliations, and the analysis was not designed as a formal comparison among stakeholder sectors. The focus group dynamics may also have influenced which participants contributed most to the discussion. Additionally, the evaluation was retrospective and focused on one model and one provincial context, limiting transferability to other tools and jurisdictions. Finally, coding was led by one researcher (while reviewed by project team members). While transcript checking, iterative team review, and participant review of the workshop report supported the credibility of the findings, member checking was conducted at the level of the synthesized findings, rather than through participant review of the complete transcripts.

6 Conclusion

This study evaluated how British Columbia stakeholders interpreted a climate-driven agricultural suitability model and its potential role in adaptation planning. Participants valued the model as a regional, exploratory resource for comparing scenarios, identifying emerging patterns, and supporting long-term discussion. They did not regard climatic suitability alone as a sufficient basis for farm or land-use decisions.

Three implications follow. First, suitability projections should be interpreted with information on water, soils, infrastructure, markets, production economics, and local conditions. Second, model interfaces should communicate scale, assumptions, and uncertainty clearly and provide regional summaries that match practical decision horizons. Third, stakeholder engagement should be iterative and accompanied by governance safeguards where model outputs may influence consequential investment or land-use decisions.

Used in this way, regional agricultural suitability models can provide a shared evidence base for anticipating change without claiming to determine local agricultural futures. Their contribution lies in supporting coordinated inquiry and planning showing where further assessment is needed, where adaptation resources may be required, and where emerging opportunities warrant careful, place-based investigation.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fclim.2026.1928659/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Stewart Jennings, University of Leeds, United Kingdom

Reviewed by: Bwalya Mutale, Kapasa Makasa Univeristy, Zambia

Samuel Ikendi, University of California Agriculture and Natural Resources, United States

Footnotes

https://www.agrilyze.ca