Talmy et al. “Killing the predator: impacts of top-predator mortality on global-ocean ecosystem structure”

This is a nice, concise paper analyzing the effects of two variants of a plankton food web structure, with two types of losses (linear and quadratic) for the top predator in the modeled food web. Here, the predators are microzooplankton, but I believe the results are extensible to a case where there is one microzooplankton and one mesozooplankton. In Talmy et al., the authors show that the “diamond” food web structure, where one zooplankton feeds on two phytoplankton, results in a marine ecosystem with less dynamic range between the gyres and the poles in phytoplankton carbon, as well as less co-existence in community composition. Simultaneously, the quadratic losses in the top predators results in a phytoplankton to zooplankton biomass relationship that better represents recent observations. I believe this

This paper is written in a clear and consise way, but one major criticism is its lack of robust engagement in other mechanisms that may contribute to variations in community co-existence, Z:P ratio, and phytoplankton carbon.  Further, while the authors do cite other publications (Ward et al. 2012, Dutkiewicz et al. 2020) that list parameter values, it was very difficult to evaluate the performance of the model without a list of parameter values. Therefore, my recommendations are:

• List out in the appendix all the ecosystem-relevant parameter values used in the model
• Can the authors engage a bit more robustly in the discussion, other strategies that modelers use to modify zooplankton grazing in a diamond food web structure that may also result in improvements in the two main metrics (dynamic range between gyres and poles in phytoplankton carbon, community co-existence):
  1. For example, many biogeochemical models utilize either different maximum grazing rates for zooplankton depending on the prey type, OR use a prey selectivity factor to modulate grazing. For the former, BEC/MARBL (Moore et al. 2004, Long et al. 2021) uses different maximum grazing rates for a single “adaptive” zooplankton to mimic the effect of multiple zooplankton in a single zooplankton class. For the latter, there are multiple examples of this strategy within the biogeochemical models, e.g., PISCES (Aumont et al. 2015), and COBALT (Stock et al. 2014), though those two models also have multiple zooplankton types so it may be slightly harder to compare with a single zooplankton type. However, mathematically, the effect of both these strategies would be similar.
  2. On grazing, values for the zooplankton maximum grazing rates and the grazing half-saturation constant are amongst the least well constrained parameters in food web models, and variations in these parameters have an enormous impact. Rohr et al. 2022 (Progress in Oceanography) shows this quite nicely in a robust analysis, along with evaluating differences in the grazing functional form itself (Holling type II or type III functional responses). It would be nice if the authors could engage a bit more in the discussion regarding whether modelers would be able to compensate for the lack of a second zooplankton (e.g., in the diamond food web model) by modulating maximum grazing rates and grazing half-saturation constants.
  3. Lastly – prey switching is a major issue that is only mentioned in passing in the discussion. It would be nice to see a more robust discussion – do the authors think that modifications in the switching form would result in substantial changes in the modeled ecosystem, and why? There are a lot of approaches towards switching, as laid out extensively in Gentleman et al. (2003), but models typically use just one or two forms (e.g., Stock et al. 2008 Journal of Marine Systems has addressed this quite nicely in a simple system). In my opinion, a more than cursory treatment of this topic would be important in this paper.
  4. Other parameters that may additionally modulate phytoplankton carbon in food web models that aren’t addressed include the fraction of phytoplankton and zooplankton losses that go to dissolved organic matter vs. particulate organic matter, which may influence the recycling rate and strength of the microbial loop. Lastly, variations in the relative nutrient uptake rate of the different phytoplankton may also result in more or less differences in the phytoplankton carbon between the gyres and the poles.

Other than these points, I found the manuscript written quite clearly, with compelling figures and nice presentation. With a more robust discussion addressing a range of these additional points listed above, this manuscript would make a nice addition to the literature.