General

This manuscript describes a new coupled Landscape Evolution Model (LEM) and eco-evolutionary model called AdaScape. The LEM is FastScape, a well-known model that simulates river erosion and hillslope erosion an uplifting the landscape. The authors additionally incorporate orographic precipitation into the LEM component of the model. They then couple FastScape with an eco-evolutionary model in which individuals experience trait evolution in response to selection, mutation, dispersal, and competition. The authors explore elevation and precipitation as organismal traits in the models presented in the paper. The results show that trait-mediated competition has a large influence on the number of taxa. The landscape dynamics also influence the number taxa, but most significantly the rate of uplift influences the timing of lineage diversification. I think that AdaScape will be a very useful addition to the growing field of coupled Earth-Life science models. The paper is generally well-written, and the model clearly communicated. I think the manuscript will be improved with a deeper consideration and discussion of some topics that I detail below. There are some minor grammatical and wording issues throughout that I’ve noted in the section “Technical Corrections.”

Specific Comments

Taxon Definition: If I understand the taxon definition correctly, some degree of trait divergence is necessary for speciation to be recognized in the model. However, speciation can occur (and as recent work has shown, may most frequently occur) in the absence of trait divergence. For example, Anderson and Weir (2022) compared sister species pairs of birds, mammals, and amphibians and find that the majority evolve under similar selective pressures (and show little trait divergence). I wonder if the authors have considered other definitions of taxon groups that would not require trait divergence, perhaps geographic distance and/or time. You could imagine, for example, two populations becoming isolated in lowlands across a mountain range – they may undergo allopatric speciation because of a geographic barrier even though their traits may not diverge substantially. While the current taxon definition may be useful for some specific questions, it may not be useful for replicating real-world patterns where allopatric speciation seems to be common.

Mutation: The authors implement a mutation process described as “an intergenerational stochastic variation of trait values”.  I think the implementation is very reasonable, but I think it may be important (especially for geoscience readers) to clarify that the mutation process here is more conceptually complicated than a mutation rate in units base pair/time. Here, the model assumes that mutation affects a specific trait optima, when, most mutations are likely deleterious. I don’t think these additional complications need to be added to the model and they are discussed in Irwin (2012), but it would be useful for readers to know if they embark on using the code as the specific understanding of what mutation means as it has implications for the values used for rate of that process.  

Line 180-181: The authors claim that the model “produces well known patterns of evolutionary branching under fewer 500 generations along a continuous gradient”. This claim requires more justification. What exactly is the pattern of branching that is observed and what aspect of it matches “well known” patterns?  

Line 190: The local carrying capacity is lowered to reach a similar total abundance of ~400 individuals. Is this done by trial and error or is there a way to a priori decide on these values needed to produce models with similar abundance?

Temporal scaling: The mechanism for linking the eco-evo component of the model to the LEM in time was not clear to me until the very end of the manuscript (Lines 323-331). It would be helpful to discuss how they are linked in time (e.g. one generation = one LEM time step) earlier in the manuscript, probably in section 3.2.

Effect of competition: One of the main results from the study is that taxon richness is much higher with trait-mediated competition as compared to without trait-mediated competition. In fact, this difference seems to be greater than the comparison between a static versus dynamic landscape. Does this mean that in the face of selection and competition the landscape dynamics are not very important for generating biodiversity? I think the manuscript would be improved with a deeper consideration of the implications of their results for understanding the importance of landscape dynamics.

Figure 4G: I think there are a grey line and a thin dashed line along the x-axis that can’t be seen easily. I suggest dropping the axis down slightly below 0 so the lines can be seen. I was confused about why there were only wo lines in Figure 4G when Figure 4N has three (min, mean, max).

Figure 5: It is hard to see the 2D histogram, can the lines be darkened?

Technical Corrections (grammar, wording, etc):

Line 3: “an ideal tool” --> “ideal tools” (referring to models which is plural)

Line 12: “observed” --> “observe” (present tense)

Line 25: “challenge to study” --> “challenge of studying”

Line 25: I think “recalled” is not the right word, perhaps “utilized”?

Line 30: The final sentence of the paragraph here is a bit hard to understand but is crucial for distinguishing the model here from previous efforts where eco-evo models run on static landscapes, suggest rephrasing to better emphasize the difference.

Line 31: You aren’t generating landforms themselves, suggest rephrasing as “Generating landforms in simulations…” or “Simulating landforms…” or “Understanding how landforms are generated…”

Line 51: “which is” --> “which are” (subject is plural)

Line 182: suggest deleting the word “exemplary” which would imply two really stand-out and impressive results (where I think the authors mean two typical examples of results). Same thing on line 203.

Line 210: “the simulations ultimately reach equivalent mean, maximum, and minimum values” (of what?)

Line 218: “To large differences” ….unsure what that means, typo?

Line 253: “as the rate of uplift is getting slower” --> “as the rate of uplift slows”

Line 257: “show also” --> “also show”

Line 269: “varies” --> “vary”

1. General comments

The manuscript presents a landscape evolution and individual based model integration. This is not new per se, but the integration the authors provide is a very valuable asset for studying the links between geo and bio dynamics. Throughout the manuscript the authors give the impression that the other previous models (e.g. Rangel’s and Hagen’s) cannot be used to study this link. The authors should make clear that this is not the case, while reinforcing the advantages of their tool since the integration of these two different models is provided in one software. Additionally, further description and discussion on assumptions taken, especially for the evolutionary component, should be clarified (e.g. individual based, speciation, taxon definition, mutation, gene flow, landscape residence, “generations vs timesteps” and runtime). Furthermore, deeper discussions and clarifications on the limitations such as computational time and limited biological processes implementation and uncertainties related to individual based modelling approach and their taxon definition and mutation process should be added. Deep-time eco-evolutionary models have been harshly criticized due to their large number of parameters leading to identifiability and structural error problems. I would like to see the authors defend this further and justify the advantages of adding these two models together while increasing inevitably the number of parameters. Specific comment suggestions point out what could be changed and what would benefit from further clarification.

2. Specific comments

I found the use of elevation, rather than a proxy of temperature rather unusual. For example, do you expect elevation to act similarly on closely related groups and higher and lower latitudes? Macro-ecologists would benefit from a more direct proxy for biological processes and I wonder if this could be implemented or at least discussed upon.

The authors present an individual based model of biodiversity to investigate deep-time questions, which is in strong difference with most current approaches investigating deep-time which use population based models to tackle geodynamics [https://doi.org/10.1111/ecog.06132]. I wonder what the computational constrains and advantages of this approach are. Specifically, it would be handy for readers to have an idea of computational requirements for a given spatial and temporal extent with some average parameter setting (or at least the spread of CPU time for different scenarios). 10’000 generations in about 80min seem very time consuming to simulate anything global and deep-time where plate tectonics are relevant. Consider providing more information of spatial and temporal extent coverage and resolution as well as possible processes that can be considered

Line 25: “In these types of models (e.g.: Rangel et al., 2018; Hagen et al., 2021a), individuals or species interact and evolve, generally, in a grid-based environment, where environmental fields (e.g. topography, temperature, and precipitation) are static representations of past or present features. Consequently, this offers little control over the processes that generate landforms while emphasising the ecological and evolutionary drivers of diversity.” This affirmation is wrong in two ways. First, both models consider dynamic environments, with the only difference being that input landscapes are generated prior to the biological model run. This can be a problem for pipelining multiple LEM parameters, but still gives flexibility in choosing or inputting landscapes. I see the biggest difference in these methods that the landscapes are created on the go by your toolbox, or did I get this wrong? This also has consequences that should be emphasized. Secondly, these two models do not consider individuals or species explicitly. These models are population based in order to cope with the large computing requirements when considering deep-time.

I was confused at fist on how generation time and discreate time-steps are considered? The authors hint on disadvantages of discrete models, but end up using the same method, sometimes referring to the discretization of input and processes as time steps or generation time. I recommend defining this more clearly and only using time steps. 10 kyr is of an unheard generation time, and using generation time specific for AdaScape would only cause confusion.

Another aspect that was not clear is: How does the taxon definition link with gene flow and geographic distance, or is this process not considered at all?

Are you considering landscape resistance? I have trouble in putting dispersal as an evolutionary process. Can you explain better how dispersal is working?

I missed some more explanation about the local neighborhood. Does it also influence trait homogenization?  Do you consider this at all?

How does the spatial component of the competition work? Seems like it can jump sights as presented on Figure 1  top right.

I did not understand if you have any reproductive isolation? When things diverge on traits but randomly get similar below the threshold, can they again become the same species? Can you please clarify this?

3. Technical corrections

Lines 8-9: “Particularly, we investigate how changes in tectonic uplift rate and ecoevolutionary parameters (i.e. competition, dispersal, and mutation) influence the temporal and spatial patterns of biodiversity”. I wonder why you did not add speciation here, since you focus on deep-time and mention speciation as the only process on Line 39: “Here we present AdaScape, a coupled speciation and landscape evolution model”.

Line 115. I could not follow what you meant by: “all individuals have been updated”.

Line 123. Something is off here “then branching does not occur all individuals will be assigned to a single taxon-id”.

Line 127-128. You can reword this to “only considering binary splits”, or “artificially transforming polytomy into simple splits”. This could resonate better with biologists.

Line 130. Close “)”.

Line 326-327. “The most adequate scaling is to consider that the mutation and dispersal rate need to be scaled by the square root of the number of real generations in a LEM time step.” Citation for the scaling proportion is missing. Consider only mentioning time step throughout since these are not only exclusive to LEM.

Line 328-331. You could add some further discussion here on the comparison of individual and population based models, or just provide citations to this other body of work.

I could not follow on section 2.1 how individual traits change, or if they are only selected. If only selected, I would rather call this an ecological, rather than an evolutionary process, making mutation the first evolutionary process (Line 78). The same applies to dispersal.

Could you justify the selection of sigma_u=2? I would expect a value of 1 for the experiment where trait proximity is irrelevant for competition.