Optimising Regional Land Use to Enhance Water Productivity Under Climate Uncertainty: The Role of Perennial Crops

Comments 1: The authors should include quantitative validation metrics where possible; compare modeled crop allocations with observed agricultural census data; assess model performance using historical land-use records; discuss limitations arising from limited validation

Response 1: The authors acknowledge validation metrics assist the reader in assessing the research impact. However, as this research is looking forward to an uncertain future validation retrospectively to past land use would not be beneficial; past performance is not a reliable indicator of future performance. For this reason, the economic quantification of the region for 2020 at the current 23% perennial land use was used to validate the model’s prediction power. The authors believe this is a like-for-like comparison. Further, it is not reasonable to expect that optimising at the regional level will match actual farmer and agribusiness behaviour.

A comparison of irrigated land use could be possible using Murrumbidgee Irrigation Compliance Report 2023/24 and 2024/25 WY. However, this data is by crop only and does not consider the crop x soil allocation; unlike this research. Furthermore, the suggested census data does not disclose the details to allow a comparison between the four production systems studied (dryland, irrigated, annual, and perennial).  

Again, the authors are of the belief that conducting such an evaluation between model and historic data for the 2020s would not be useful due to the unprecedented triple la Nina events of 2020-2023.

The authors adopted last comment and have amended Section 4.3 Limitations and Uncertainty

Comments 2: Deficit irrigation is central to many conclusions of the study, particularly the economic advantage of perennial crops under future water scarcity. The selected target watering levels appear highly simplified. The manuscript assumes that yield reductions follow fixed percentages, responses are uniform across all perennial species, soil effects and cultivar-specific responses are negligible. The authors should provide stronger empirical justification for selected Deficit Irrigation thresholds; explain why species-specific Deficit Irrigation responses were not incorporated; include a sensitivity analysis demonstrating how conclusions change under alternative assumptions.

Response 2: The authors agree that the interaction of crop and deficit watering is this current work is an abbreviated approach. As noted in the Future Work section, the authors plan to connect the optimisation model to a biosimulation model such as the Agricultural Production Systems Simulator – APSIM. This marriage of STALS to APSIM would refine the yield potential for DI under future climates.

The authors have noted the reviewer’s comment regarding stronger empirical justification for the selected deficit irrigation thresholds and have amended the manuscript accordingly. See Section 2.2 Water Usage

As noted in the manuscript, researchers’ understanding of DI impact on yield and quality is still evolving (section 1.1.4 Economic Water Productivity). The STALS model is customisable, allowing new DI water targets and market response data to be easily incorporated when addition insights become available.

The authors are of the view that a sensitivity analysis would not reveal any deeper insight than what has been presented. Such an undertaking could be considered in Future Work, however, the authors believe the current list of identified enhancements are more beneficial to the understanding and answering of the research question.

Comments 3: The study assumes that all perennial crops are already established and mature. This assumption excludes establishment costs, juvenile non-productive periods, orchard replacement & removal costs. Consequently, the optimization framework effectively evaluates an idealized landscape rather than realistic transition pathways. The discussion should more explicitly acknowledge that results represent steady-state conditions, expansion of perennial crops may be economically less attractive when establishment costs are considered, land-use transition pathways remain unresolved. This limitation should be highlighted more prominently.

Response 3: The authors do disclose in Section 2.7 Assumptions that this iteration of the model assumes a ‘steady-state’. This assumption represents the decision that the region has moved to that reality and illuminates what a 10-year production cycle would look like. In addition, future work would include economic realities of land use transition to better assess land use opportunity costs.

The authors have addressed the reviewer’s comments and have amended Section 4.1 Economic Water Productivity and Section 4.2 Temporal Shifts of Land Use in the Regional Landscape.

Comment 4. The study uses only two climate projections. The conclusions rely heavily on these two models. The study does not consider climate model ensembles, uncertainty propagation, variability among GCMs. The authors should discuss why these two models were selected; how representative they are of broader climate uncertainty; whether conclusions are robust across a wider range of climate futures.

Response 4: An ensemble could be employed with a mixture of climate projections, but if there are extremes, an ensemble presents an average, so no benefit is gained from this approach. A more appropriate approach is the one adopted in this research, where two dissimilar GCMs (CanESM2 – warmer and wetter and ACCESS1.3 – hotter and drier) climate projections are chosen to capture the spectrum of future possibilities.

Prior work by the authors (Randall, M., Montgomery, J., & Lewis, A. (2022). Robust temporal optimisation for a crop planning problem under climate change uncertainty. Operations Research Perspectives, 9, 100219) found that attempting to simultaneously optimise across multiple climate change projections was impractical.

The inclusion of climate models with more extreme projections may be considered in future work, although the authors note that a growing number of climate models are converging on a single point, a hotter and drier climate; ACCESS1.3.

The authors accept the comment and have revised Section 2.4 Climate’s Economic Impact

Comment 5. The market-behavior mechanism assumes price increases under water scarcity. The manuscript would benefit from a clearer explanation of the economic model, justification for price multipliers, discussion of market uncertainty. A sensitivity analysis on price assumptions would further strengthen the study.

Response 5: A sensitivity analysis on price assumption is not within the scope of this work, where the focus is development of a generic framework which incorporates food and fibre production realities to optimize regional agricultural economic water productivity, along with demonstration of its usefulness. As previously noted, the model is customisable which enables parameter updates as new insights are available.

That said, the authors have modified the manuscript (Section 2.3 Economic) to clarify the economic component of the model. Furthermore, the authors acknowledge that this suggestion should be included in the next iteration of the model. However, as the horizon investigated is so large that future circumstances and events are impossible to forecast, the usefulness of the exercise is limited.

Comment 6 The manuscript acknowledges that water trading is excluded despite its major role in Australian irrigation systems. The authors should discuss more thoroughly how inclusion of water markets might change results, whether perennial crop dominance would persist under active water trading scenarios. 

Response 6: The authors note that unless water is traded at prices different from those paid by farmers for allocations, intra-regional water trading and its impact on regional revenue cannot be identified within the model. The STALS model provides insight into the redistribution of water across land uses and species to maximise regional economic water productivity. Within this framework, fallow land is implicitly assumed to transfer its water allocation to other productive uses. A further example is the reality of broadacre enterprises such as rice, sell their water allocation when the purchase price exceeds $350/ML to higher-value enterprises such as nut production. This important component of the realities of food and fibre production is a possibility for future work.

Comments 1:

"In all experimental instances (2020s, 2050s, 2090s), the addition of perennial crop land use to the landscape of the study region improved economic water productivity. The inclusion of the operational tactic of deficit irrigation improved the feasibility of perennial crops, particularly in the hotter and drier climate." The authors could note that Australia is already the driest continent and with climate change, dry places are expected to become drier.

Response 1: Thank you for pointing this out. Section 6. Conclusion has been updated to reflect this comment.

Comments 2:

The authors' findings are transferable to other major agricultural regions which feature a mix of perennial and annual crops, such as the Central Valley of California. Deficit irrigation is an interesting practice, but the authors should explain the difference in cost and water use between drip irrigation and flood irrigation systems

Response 2. Thank you for your comment. Section 2.7 Assumptions has been updated to accommodate your suggestion.

Comment 3.

The discussion of "self-mulching clay" seems a bit extraneous to the thrust of the paper, comparing perennial and annual crops.

Response 3 To illustrate the model’s data-rich outputs and utility as a decision-support tool, a representative land-use example was selected. Self-mulching clays were chosen due to their extensive spatial coverage within the study area and their high versatility in production capability.