Dear authors

Thank you for your responses to the three independent reviews of your originally submitted manuscript. All three reviews were positive with regards the overall motivation and approach and that there is clear merit in your proposed framework.

I invite you to submit a revised version of your manuscript that will be sent for another round of independent peer review. In preparing your manuscript please focus on tightening its structure to clearly identify the knowledge gap your framework will address. As has already been identified, your manuscript is well written and scholarly, but content is not always entirely germane to the issues that the framework is aimed at. It is perfectly acceptable to propose a way towards closing a knowledge gap, but we need a very clear identification of current abilities, gaps, and so opportunities, and how your framework addresses them.

Best wishes
James Dyke

Dear Authors

Thank you for your revised manuscript. I have decided to accept your manuscript subject to minor revisions that I will review.

Both reviewer reports are positive. Neither require corrections, but R2 provides a list of suggestions which I strongly recommend you consider. I have captured some of these and made additional recommendations below.

The Sustainable Development Goals may provide a useful framing for your methodology. How can we reduce hunger, provide clean water and reduce climate change impacts for example? This challenge seems to suggest an intergrated modelling approach. You briefly mention this at the end of the article, but it may serve as a more important framing near the start.

P2 lin 25 Rockström et al., 2009, I suggest you use more recent Steffen et al (2015) as this also includes regional rather than global water use boundary.

P3 L15. You use climate modelling as a metaphor for model complexity. But to avoid any possible confusion, I suggest you mention that socioeconomic processes have come to be linked to biophysical models that include climate dynamics: IAMs. More generally, you are arguing for the need to develop socio-ecological modelling. That is a reasonably well established discipline or approach. It’s a minor issue, but you are setting the reader up in this introduction section so you need to be very clear how your approach is situated with existing work - and where and how it goes beyond it. Previous reviewer reports slightly misunderstood your motivations in places.

P5 L33 “However, the pathway to capturing these dynamics within global models is not obvious (Müller-Hansen et al., 2017).” this is a citation to a regional study. Verburg et al (2015) may offer a more productive global perspective. But it may not include sufficient discussion on water-energy-climate nexus.

P5 L34 “Thus, the second knowledge gap our framework aims to fill is to incorporate the complex dynamics associated with socioenvironmental systems within the IAM framework, without adding unwanted complexity to those models. We aim to achieve this by focusing on cities.” You argue that cities represent a useful aggregation of the agents of interest, that You don't need to capture dynamics of individual farmers or farms for example. This is a centrally important modelling structure decision. It needs to be defended or supported. You provide some high level data that suggests that a focus on cities would represent sufficient capture of the relevant dynamics. Some readers may not be very convinced by that. How else can you effectively argue that resolving agents as cities is the best approach?

P8 L1 “However, we recommend tight coupling between dynamic vegetation models such as Lund-Potsdam-Jena managed Land (LPJ-ML) model and a complex hydrological model that captures water fluxes between the soil layer and groundwater reservoir, which is key for computing groundwater recharge rates (Bierkens and van Beek., 2009; Bondeau et al., 2007; Hanasaki et al., 2008b).” It is not clear if you have implemented this method, or have a route towards this method, or if this is an aspiration or some sort of necessary condition for your method to be effective. In any event it is a large piece of work. I think the reader needs to get some understanding of what is involved here.

P9 L7 “The upper level network consists of bilateral trade links among countries. Link weights among countries are calculated using a CGE model, which is coupled to certain IAMs (Stehfest et al., 2013).” More specificity here. Which “certain” IAMs?

P9 L34 “This is a novel and potentially important step in capturing non-equilibrium dynamics within pre-existing equilibrium approaches without adding unwanted complexity to CGEs or IAMs (Farmer and Geanakoplos, 2009).” I appreciate this is an important point. But, if we consider your multi-model framework to be a model (or meta-model), then this represents a very large increase in model complexity. On the one hand you argue that you don’t want to add “unwanted complexity” to some kinds of models, on the other your entire approach is about capturing more dynamics and coupling them within a much more complex modelling methodology. Can you be a bit clearer about your motivations here? I think you want to argue that your approach can fit in with the existing modelling 'ecosystem'. You won't be able to take models 'off the shelf' but you don't need to engineer entirely new IAMs. This is a very important point.

P11 The discussion on how intra city dynamics can impact inter city dynamics - up to the global scale - could be framed in terms of teleconnections. You are arguing that you can only capture such teleconnections in a complex social-ecological system by explicitly resolving agent dynamics and interactions.

L9 L21 “The main step required to make this framework a reality is the construction of a multi-agent network of city nodes and trade links.” I agree! Which makes the lack of detail about such a model quite important. I think we need some framework for this element of the methodology. I would recommend you briefly review the ODD +D protocol. You may find that you are able to able to quickly produce a schematic of your proposed multi-agent model. It’s not much more than a list of entities and descriptions, but it will allow you to produce a concise table with which a reader will be able to quickly understand the nature - if not the details - of this central modelling component.

References
Verburg, P.H., Dearing, J.A., Dyke, J.G., van der Leeuw, S., Seitzinger, S., Steffen, W. and Syvitski, J., 2016. Methods and approaches to modelling the Anthropocene. Global Environmental Change, 39, pp.328-340.

Steffen, W., Richardson, K., Rockström, J., Cornell, S.E., Fetzer, I., Bennett, E.M., Biggs, R., Carpenter, S.R., de Vries, W., de Wit, C.A. and Folke, C., 2015. Planetary boundaries: Guiding human development on a changing planet. Science, 347(6223), p.1259855.

Dear Authors

I have reviewed your corrected manuscript and response letter. It gives me great pleasure to accept your manuscript for publication. I caught two very minor errors, which I list below. You will be able to address these during the publication preparation. I do not want to delay publication any further.

Thank you once again for submitting your work to Earth System Dynamics. Your paper represents an exciting vision of future modelling and I hope it serves as a useful contribution to sustainability science discussions.

Minor issues:

P6 L36 " cities are agents and comprise of an urban area and associated hinterland" suggest change to: cities are agents and are comprised of an urban area and associated hinterland.

Dermody, B.J., Tanner, C.J. and Jackson, A.L., 2011. The evolutionary pathway to obligate scavenging in Gyps vultures. PLoS One, 6(9), p.e24635. (incomplete reference in current/revised manuscript).