In this manuscript Delattre et al explore the effect of organic farming expansion on pest and predator densities in virtual landscapes. The topic is timely given the upcoming expansion of organic farming and the uncertainty for practitionners regarding its potential drawbacks. The manuscript is generally well written and the results are interesting. I have the following major issues with the manuscript as it currently stands:
- parts of the discussion need to be expanded by comparing the results presented here with published empirical and modelling work. I am also missing a section in the discussion discussing the potential limitations of the simulated results presented here given the assumptions made in the model. For instance the results show situations with predator to prey ratios above one, this is contradictory to most empirical settings. Another example is the assumption that pesticide will affect pest but not predator.
- many results are presented: two different type of agricultural fields, within three different landscape context, for three different scenario of organic farming expansion, with four different pesticide effect for two different dispersal intensities ... I would argue to add at least one schematic overview figure summarizing the results. I also wonder whether to streamline the results, based on the aim of the manuscript, by focusing on the interaction between organic farming expansion (three scenarios) and landscape context (two level of fragmentation) and put the rest in appendices.
- the manuscript uses quite some abbreviations, I would recommend at the start of each section to fully name all terms before using the abbreviations.
In addition I have the following minor comments:
line 25: remove "the"
line 47: cropping system sounds a bit too restrictive here, I would replace it with farming system.
line 71-72: by investigating the effect of OF on what? Please precise.
line 88: parenthesis around Petit can go away.
line 88: damage on what? Do you mean yield loss?
line 100-102: I would re-phrase the second part, something like: "and may support higher spill-over of predators from seminatural habitats into agricultural fields"
line 121: I would re-phrase something like: "lack of CBC data in the context of OF expansion"
line 142: de-capitalize intensity. Maybe better to use or add the word pesticide here?
line 152: please provide in the text or in appendices information on this model. Readers should not have to read another paper to know the basics of the model used here.
line 156: what is HSN?
line 169: error in formatting of the equation
line 165-182: I am not an expert of predator-prey LV models, but I find it curious that the effect of the prey on the predator growth is modelled separately from the predator growth function fp. I would have assumed that the function fp also takes as parameters the prey population and the interaction term. I am also not so sure what this interaction term represent, is this another way to specify the conversion efficiency of prey biomass into predator biomass? Could you cite some previous work that used a similar mathematical formulation of spatial predator-prey LV model? I also note that intrisic death rates as controlled by the parameters p are assumed to be solely driven by pesticides.
line 200-202: how could a pesticide having the same impact on pest species in conventional and OF fields have less or even no impact on predator in OF compared to conventional fields? These two scenarios looks like wishful thinking for me. Could you maybe back this up with empirical data showing that pesticides in OF fields have impacts on pests but not on predators?
Table 2: What is CF and BIO?
Figure 1: please use color-blind friendly color palette.
line 340-394: given that the aim of this manuscript is in studying predator-prey dynamics in the context of orgnaic farming extension, I find this paragraph to be beyond the scope. Consider reducing it or removing it.
line 406: this is interesting since my limited understanding of LV models is that they lead to a variety of dynamics from system collapse to stable periodicity. Please precise what you mean by stabilized (ie do you mean temporally constant densities?), at what spatio-temporal scale and provide some figures showing these stable dynamics.
Figure 4, 7 and 8: These figures are hard to interprete, they have multiple panels, different colors and line types. I would simplify and re-organize these figures to make the main results more readily stand out. For instance the text associated with figure 8 focus on the interaction between fragmentation and dispersion rates, figure 8 should then have one of these two variables in the x-axis with coloring for the second, patch area is not mentionned in the text associated with this figure so this variable can be averaged over.
line 516: which quartiles are represented in Figure 5?
Figure 5: predator to prey ratio seems extremely large, in organic fields with many SNH there were more predator individuals than preys. How do you explain this? And do you think that such a situation can occur in empirical systems?
line 519-521: convulated sentence, please re-formulate or split in two sentences.
Figure 7: please use the same typology of pesticide effects as in the method section.
line 569-570: the interactive effect of pest dispersal and fragmentation on pest densities appears marginal, the differences between fr=0.1 and fr=0.9 are low. These effects are also hard to see on figure 8, see my other comment.
line 590: at the start of each sections I would write again the full names of the abreviations, readers do not always read through from start to end but jump around.
line 601-602: Based on what results is this conclusion based?
line 602: To what does the this refers to? To your results that IP provide most benefits for biological control or to the fact that organic farming extension mostly happens in patterns similar to your IP scenario? I would rephrase this sentence and the next to make it clear.
line 612: densities of what?
line 614: SNH can be dropped.
line 621: predator to pest ratio
line 622: Aha, if the benefits that you mentionned at the beginning of the discussion section refers to the predator to pest ratio as a proxy for biological control then this should be stated also there.
line 625: using random development as a baseline seems odd especially since you mentionned a few lines above that in real landscapes organic farming expansion rather follows a clustered pattern. So either use IP scenario as baseline (what would happen if OF expansion follows the current trend) or discuss why you think that using random development is better suited as a baseline for this exercice.
line 629: use throughout the same wording for the predator to pest ratio.
line 631-647: A complex paragraph again focusing on land-use development a bit beyond the focus of this manuscript. I would recommend to remove it.
line 645-646: Do you mean that you assume to be able to predict how pest and predator dynamic will be in a given landscapes based on the results presented here? This sounds overconfident, especially given that your results showed interactive effects with landscape contexts.
line 664: What strategies?
line 669-673: I don't get these sentences first you write that IP is best for predator to pest ratio and then you write that under this strategy it declines. Harmonize the two sentences.
line 682-684: A very clear sentence, I would start the paragraph (line 669) with this one and develop from it.
line 716: What is meant by intensive organic farming?
line 662-754: There is little comparison of the results presented here with already published work. Please reflect on how the results presented here complement, contradict or present new results compared ot past empirical and modelling work.
line 755: In the conclusion I would use a term less cold than IP strategy, maybe replacing it by "clustered organic farming expansion".
Online repository: The model code on the OSF repository would profit from a fully fledged readme explaining how to run the models and either describing the main features of the models or citing the relevant literature. The zenodo repository is very large (1Gb), I would only put on the repository the post-processed datasets used to draw the figures, the raw data files of 500Mb are not needed there. The zenodo repository would also benefit from a readme file explaining what the code is doing.