In the manuscript ‘Nutrient flows and biogeomorphic feedbacks: linking seabird guano to plant traits and morphological change on sandy islands’, F. van Rees and colleagues investigated the interactive effects of plant traits and seabird guano on shoreline geomorphology on small, dynamic, sandy islands in the Wadden Sea. The findings are highly interesting and novel, and complement ongoing research on the circular seabird economy system with new evidence from temperate region and on the cascading effects of seabird guano to island geomorpho-dynamisms. The manuscript is well written and summarises the context and key findings of the study well, but I miss some clarity around the specificities of island ecosystems, where this study was carried out. For example, the role of allochthonous nutrient inputs for terrestrial productivity is most pronounced on islands (see seminal work by Wendy Anderson, Gary Polis, in the 1990s-early 2000s) relative to mainland systems. Similarly, the geodynamisms of sandy islands are quite different to mainland sandy shorelines, as sediments can be reworked around islands 360°. Also, seabird colonies are nowadays often restricted to small islands because mainland shorelines are often too developed and/or disturbed, so the downstream implications are mostly relevant to island ecosystems and island conservation. Given that your study has major implications for seabird conservation and island ecosystem restoration (see also my comment further below to elaborate on these implications in the disucssion), I would recommend emphasising the context specificity of island systems in the abstract and throughout the main manuscript (in particular introduction, discussion). Below are specific comments that I suggest addressing before publication. Especially the methods section lacks some key information details to understand your work. All in all, I applaud the author team to a great and relevant study, and I look forward to seeing it published and citing it in my own work:

Specific comments:

l. 46–47: what kind of ‘physical processes’ are you referring to here?

l. 55–58: around here I would have wished to read more about the uniqueness of island systems, in which autochthonous production is small relative to allochthonous input. I think this is an important context specificity of your study that needs explicit attention and introduction.

l. 67: ‘guano fosters vegetation growth’ – this is a bit too oversimplified. Increasingly higher guano loads can cause plant community composition shifts, reduce species richness, and cause plant die-back (especially when paired with low rainfall; see Gulf of California island studies). As you are also speaking about plant trait-dependencies for soil stabilisation, it might be worth introducing that guano is not a universal ‘booster’ for any plant growth but can also shift plant communities to certain species, growth forms, or traits (which may or may not be beneficial to soil stabilisation in your study system).

l. 72: Island systems come out of the blue here (before: only ‘coastal landscapes’), so I recommend some more specific introduction earlier, see comments above.

l. 80–81: I wonder whether the relevance of your work can be packaged slightly differently? Testing whether seabirds exert a measurable, and ecologically meaningful influence on island shoreline stability would demonstrate that seabird conservation and active ecological restoration is not only a tool for species-centric conservation efforts but constitutes an ecosystem-based adaptation for small, uninhabited sandy islands.

l. 83: define ‘biogeomorphic islands’.

l. 84: ‘ideal laboratories’ but because of their unique dynamic nature, they constitute a specific system that does not necessarily inform about feedback loops between seabirds and shoreline dynamism on mainland coasts.

l. 87: Why is the Dutch Wadden Sea a model system? Model for what?

l. 101: I presume you mean uninhabited, not inhabited, islands?

l. 110: In the title, you refer to these birds as ‘seabirds’ but here (and also at a few other places throughout the manuscript) you use the term ‘coastal birds’. I suggest sticking to one term only, and I recommend using ‘seabird’ as the more widely recognised group of birds that drive cross-system energy flow. Along these lines, you may find the framework of the ‘Circular Seabird Economy’ also interesting to contextualise your study.

l. 110: what is a ‘classic’ barrier island?

l. 126: I would have liked to read total breeding pair numbers (summed across all species) per island here to get a rough understanding about the colony sizes and seabird influence. Are we talking 100s, 1000s, or millions of birds per island?

Figure 1: I find the green line visualising the vegetation border a bit hard to see with the colour-scale background. Consider changing to a more strongly contrasting colour.

Figure 2: This is a very useful schematic figure, thank you! My only suggestion is to directly link each block in the figure to the corresponding section in the methods (e.g. I presume ‘field approach’ is section 2.3; ‘NDVI’ corresponds to section 2.4.1., etc). I think this would make your very extensive methods and data types even more accessible.

l. 148: How did you handle the differing transect lengths in downstream modelling?

l. 148: How did you spread your plots along each transect?

l. 151: explain in more detail what this ‘haphazard’ sampling looked like?

l. 153: 4m² plots – I assume these were 2x2 meter, but can you state this explicitly?

l. 153: how did you conduct the percent cover estimation? Was this based on %-increment bins?

l. 155: where within your plots did you place the soil profile measurements?

l. 160: Was this your fixed threshold (d15N>10) to determine seabird influence on plant communities in downstream modelling?

l. 161: For what kinds of measurements did you do this procedure?

l. 184: Did you center and scale your variables prior to PCA computation? Given that your variables are all on different scales with probably greatly varying ranges and means, I think you should center-scale them for proper PCA representation.

l. 206: Delete double full-stop.

l. 211: Please provide some more detail on the field methodology for determining location and density of seabird nesting on the islands. Were these protocols harmonised across your five islands?

l. 216: Since nesting in your temperate system is highly seasonal, does it make sense to calculate guano deposition per year?

l. 227: You set the limit of your guano dispersion model to 300 m (l. 218), then tested dispersion length up to 300 m (l. 224), and then report that your dispersion model detects best fit at exactly this limit (l. 227). Isn’t there some circularity in this approach?

l. 240: I am not too familiar with the phenology of Wadden/ temperate seabirds but is there also relevant overwintering (of potentially different seabird species) that may further increase guano input in autumn and winter months?

l. 267: how did you handle the spatial replication across five islands? Were these captured with group-level intercepts? From your model equations, it looks as if ‘island’ was treated as main effect – why not as group-level effect (“random intercept/slope” in frequentist language)

l. 274: Normality and homoscedasticity are not obligate requirements for Bayesian models (e.g. you could capture heteroscedastic data with a distributional modelling framework). However, I would have liked to read some more information on the Bayesian model evaluation, incl. prior sensitivity analysis, convergence, effective sample size, and posterior retrodictive checks? (e.g. see the WAMBS checklist: https://doi.org/10.1037/met0000065)

Figure 4: in panel (b), scientific names should be italicised.

l. 304: You report that the model explains 9% more variance when adding guano as an additional parameter. However, mathematically, adding parameters to a model almost always increases R² because the model gets more degrees of freedom to explain variation, so reporting the 9% alone is not necessarily evidence for an ecological effect. I would like to see more evidence that this 9% increase is indeed an ecological effect attributable to guano, rather than a mathematical byproduct of increasing the number of predictors.

l. 325: is this posterior effect size reported on the z-standardised scale? Also, as per Bayesian convention, I suggest reporting posterior credible intervals, not standard deviation.

l. 330: The terms “fixed effect” (and “random effect”) come from frequentist terminology but do not reflect the Bayesian thinking (e.g. all modelled parameters are probabilistic and not fixed), so I recommend using “population-level effect” or “main effect”

l. 331: For better interpretability, could you un-z-standardise the posterior slope estimates?

l. 350: interaction with what?

l. 371–372: ‘lower areas experiencing more erosion or less accretion’ – this statement contradicts what you write earlier in the results section, line 358: ‘lower areas were more prone to accretion or less erosion’. Please check carefully for consistency.

l. 400: Vegetation tends to allocate biomass and productivity primarily in the compartment that is most limited (e.g. canopy-development in highly competitive, light-limited tropical rainforest conditions, root development under nutrient limitation). Couldn’t you also argue that seabird guano input would ease the nutrient limitation for plant species, therefore reducing their allocation in root biomass (sensu ‘optimal partitioning theory’)?

l. 408: missing word: ‘might also HAVE limited us’?

l. 445: Rather than a discussion on model development, I would have wished to read a few sentences on the implications of your findings for seabird conservation / restorations. How can your findings be used to promote seabird conservation as a tool for enhancing sandy island ecosystem resilience? How should your findings be used and framed by conservation practitioners that seek to promote seabird protection and active restoration (e.g. through translocation, decoys, acoustic luring) as an ecosystem-based adaptation?