The paper is interested in the idea that global seed production might be partially driven by selection from seed predators and pathogens. This paper uses what appear to be opportunistic data (compiled by MASTIF project but don’t seem to be systematically collected, given large latitudinal biases in data) on seed production by trees (I think trees only? Not clear if non-tree seeds are excluded from seed-trap data), which are paired with data on tree basal area, so authors can look at overall seed production and seed production scaled by tree size (with the simplifying assumption, I think, that basal area reflects tree size comparably across ecosystemss). These seed production data are then compared to large-scale data on Net primary productivity, annual temperature, moisture, and soil cation activity ( a proxy for soil fertility). The authors reason that if changes in tree fecundity can be predicted by abiotic variables alone, then there is no evidence that biotic interactions are meaningfully contributing to geographic patterns in seed production, whereas if fecundity cannot be predicted by abiotic factors alone, it could be driven by selection from biotic interactions. They find the latter pattern.

The last global synthesis on seed production was >10 years ago, and the data and methods here are considerably more in depth, so the study and data definitely have merit. Interest in geographic gradients in biological patterns is having a dramatic renaissance currently, so I think the topic will be of great interest to readers of Ecology Letters. I see three weaknesses with the paper that would have to be addressed before it was suitable for publication, esp in Ecol Letters. Almost all of my detailed comments relate back to these three issues. It looks like the paper was originally submitted to a very-short-format journal, in which case the authors may have some word length to play with. I would encourage them to use this to more clearly outline the studies importance and logic for readers.

1) I feel that there is a disconnect between the pitch (based heavily on species interactions) and the data (seed production and large-scale abiotic data – ie no data on species interactions). The authors are inferring the importance of species interactions from how well abiotic data predict tree fecundity, but the robustness of this assumption is not clearly argued in the Intro. Even if that logic was clarified, I still think the prominence of interactions (title, abstract, intro) is overplayed given that there are no data on interactions and all the analyses are correlative. That is not a weakness of the data/analyses, so doesn’t detract from their strength or the papers merits, but I think the interactions angle needs to be tempered. Results do not suffer from this problem, as strong interactions are more widely interpreted as being a potential product (not just cause) of high seed fecundity. Might help to move away from ‘interactions’ when statements are more specific to competition vs granivory?

2) The contribution of the paper is not clear from the Introduction. Why do we need this study? How does it really change our understanding of Ecology beyond the smaller (but still good) studies that existed previously? What is the problem it is trying to solve? What are the alternate hypotheses (alternate hypotheses are in there but felt a bit burried

3) I found the logic hard to follow in many places in the paper, within individual sentences, within arguments across sentences, and sometimes in how different components of the paper matched up. I’ve pointed out specific instances in the minor comments below, but this is the issue that gives me the most pause about publication in Ecology Letters.

DETAILED COMMENTS

(I’m using author line numbers throughout)
L156 “This coincidence of high diversity with intense competition” – text has not yet made the case that these coincide (arguments have been about NPP & competition, previous sentence is hypothesis)

The writing was a bit more work to read than it needed to be as there were lots of vague pronouns – eg ‘this’ or ‘these’ or ‘in one’, where it wasn’t clear what the pronoun was referring back to. I’ve put some examples below from the first page, but encourage the authors to adjust not only these (ie swap vague pronouns for more specific ones) but others throughout.
L160 – depend on them – on who – tree offspring? Or trees?
L166 – depend on this demographic variable - which demographic variable?
L180 – as an increase in one attracts – in one what? One host?
L181 – the seed diversity available to consumers could differ from that of trees – from seed diversity of trees? Or just diversity of trees?

L160 – I know space is tight, but would be great if authors could clarify why this distinction matters (ie why do we care which mechanism drives the latitudinal gradient in seed production?). Right now the questions are simply stated (are these the primary questions of the manuscript?) but without context to understand why we should care about the answer. Also not clear how being embedded in more productive environments would increase seed production (mechanism 2) other than more seed production per tree size (mechanism 1)

L176 – not sure this makes sense. If selection for high seed production is to offset losses due to granivores, why would it intensify competition? Argument assumes selection makes even more seeds than needed to offset seed predation which isn’t intuitive

L179 – not following the logic that generalist seed consumers impose indirect competition among trees, as a consumer will always lower density and therefore alleviate competition

L186 – need citation for interactions being more intense in tropical forests (then in temperate forests presumably?)

L189 – I would reconsider the acronyms ISP and CSP – they are not common acronyms (I’ve never heard of them at least and I study seeds) and so require work from readers to remember. They are not used that often in the ms – are they needed? Even if authors spelled out individual vs community it would help (IndividualSP vs CommunitySP). If ISP is always standardized for tree size then its meaning is not intuitive from its name (I would assume ISP meant seeds/tree) and neither is its comparison to CSP, which is not scaled. Readers will need to be reminded of that nuance throughout. Consider changing the name of ISP to do so (eg size-standardized individual seed production – then you would definitely need an acronym! sIndivSP?).

L207 – logic not clear to me here. If fecundity gradients are steeper than predicted from size or NPP alone, why would that mean the climate was driving intense biotic interactions? I think that’s one leap of inference farther than the direct interpretation, which is that biotic interactions might be driving higher seed production. But even that interpretation needs better justification - there are surely other mechanisms that could drive higher seed production. More intense biotic interactions could also be driven by evolutionary history (e.g. higher speciation -> higher biodiversity -> more intense interactions) rather than climate. Maybe I’m misunderstanding the directionality of the logic, but either way it needs to be clarified throughout this paragraph

L221 – mortality of who? Seeds? Trees? Not following how this arms race would work (also not following proposed mechanism of arms race argument on L316-318)

L230-232 - not following the logic here. “A decline in predicted seed-mass density (per forest floor) with increasing latitude reported from a study that included only forests at low latitudes and mostly heath and grasslands at high latitudes [43] highlights the need to separate variation in tree fecundity from variation in tree abundance”
43 = (Moles et al 2009 Is there a latitudinal gradient in seed production). Beginning of sentence implied could be an additional advantage to doing the study within taxa (trees), which doesn’t seem to match end of sentence. Also its not obvious why the biome differences in Moles et al “highlights the need to separate variation in tree fecundity from variation in tree abundance”. Biome differences probably meaningfully contribute to latitudinal gradients in seed production – authors need to clarify their argument that looking within forests specifically (I think that’s what they are arguing with ‘tree abundance’?) adds something important to the larger question. Comes back to need to more clearly articulate the importance of the work. Also, given that Moles et al is THE previous review on latitudinal gradients in seed production, I would expect it to come up in the discussion as well to help put results into context (and more clearly acknowledge past literature that this study builds on)

L233 – I would add a few words here (before methods) to indicate why these variables matter (esp shade class and soil cation exchange capacity – what is that representing?)

L334 – typo, observations should be plural

L336 – spell out seeds/fruits (could mean seeds per fruit vs seeds or fruits)

L340 - I’m not familiar with ‘redistribution models’. Suggest a few additional words re what class of models these are (generalized linear models? Bayesian approach)

L348 –clarify how its possible that 98% of sites are temperate but 80% of observations are tropical? Is ‘observation’ being used only for crop-counts? If so this is confusing as all data presented are observational

L355- define crown class – preferred light-habitat of mature trees?

L363 – not clear why leaving out standardization enables comparisons among sites. Doesn’t standardization usually enables better comparisons? Clarify in a few words for readers unfamiliar with SPEI

L372 – need reference that cation exchange is a good measure of soil fertility

L379 – I don’t have a sense from this (main Methods) paragraph of how the data described above are used within this modelling framework – ie how data and model parameters line up

L401 – “We evaluated weighted mean fecundity’ for ISP and CSP – this is the only time in the paper that the term ‘weighted mean fecundity’ is used so its hard to see where its coming from / how its derived

L405 – equations are never my fav part of a manuscript but I’m not following this description. Might be easier to follow if authors clarified how do equations 4 for Individual mean fecundity relates to eq 1 for ISP.

L244 – ‘that it can play a central role’ – what is ‘it’ referring to? Amplification? NPP?

L245 - ? “If individual fecundity scales with tree basal area, then ISP (seed mass per tree basal area) would be flat in Fig. 1b.” not following. Xaxis of Fig 1b is climate – ISP would be flat if there was no relationship between ISP and climate. Mean would be a straight line? (that argument’s not great (as simply written) either as curves on back-transformed scale can result just from non-normal error distributions). Also what is xaxis on Fig 1 b? surely moisture (top x axis) and temp (bottom xaxis ) don’t always correspond directly?

L290 – logic not clear. What is take home argument re small seeds? Why are vertebrate seed predators particularly important (or are they)? Did this study capture or exclude those small seeds?

L295 – concentrating consumption means consumption by specialist consumers? Generalists would be expected to still consume seeds/tree species proportionately to seed production/tree species, no?

The current submission is definitely improved. The main results, that latitudinal gradients in tree fecundity exist, are dramatic, and cannot be fully explained by climate or NPP, are likely enough to ensure the paper is well read and cited. My main comments largely echo my main comments from before.

1. The data are really cool and really interesting. But the authors rely heavily on undermining work that has gone before in order to set up their current study. I don’t know why they do this as it is totally unnecessary re the importance of the current study (again, current study already great). I strongly encourage the authors to repitch – focus on telling us why your work is awesome (there is almost none of this in the Intro – readers are given no sense of how big the current data set is geographically or temporally. Why not provide a map of where data come from in Fig 1?), while also embracing the advances that have gone before. Stand on the shoulders of giants rather than throwing previous work under the bus

2. The writing still suffers from vagueness that often muddies the meaning.

Specific comments
I feel like the Intro currently falls into the writing trap of overstating mixed past results to justify new work, rather than accurately portraying the current state of affairs (lots of great science has already been done) and then clearly building from there.

L160: ‘Whether or how seed production might vary with latitude is not resolved by current evidence [22, 23, 24]”. ? 22 (Peco) & 24 (Chen) are about seed predation, contain no data on seed production, & should be removed. To not bring Moles et al 2009 (global synthesis on latitudinal gradients in seed production) at this point is misleading. This is the definitive previous work that the current manuscript is aiming to replace and should be mentioned right up front (as mentioned in my last review). Further, Moles et al find a strong latitudinal gradient in seed production – where is the equally strong counter evidence that justifies this statement? The only remaining ref (23, Garcia) is about 1 tree species restricted to Europe, and even they find latitudinal gradients in some components of seed production. Rather than implying we don’t have the data, which I don’t think is true, just say that previous evidence suggests there is a strong latitudinal gradient in seed production, and then (quickly here then can elaborate in 2nd last paragraph of Intro as done currently) more clearly describe what your study brings to the table that the Moles synthesis does not (e.g. standardized data collation).

L178 – “Seed removal studies are too limited in time and space to address the multi-order of magnitude variation in both seed production [39, 40, 41, 42, 43, 44] and losses to consumers [20, 45, 46, 47]; they offer conflicting evidence of geographic trends [23, 22, 24, 5]”.
First, seed removal studies are not designed to address variation in seed production, so not clear how first part of sentence is relevant. Second, yes lots of seed removal studies are limited in time and space, but why bother citing all the small studies? Why not just look to large-scale studies with the power to detect temperate to tropical gradients? Hargreaves et al 2019 spanned 64 degrees of latitude with decent temporal replication and found strong latitudinal patterns in seed predation. Chen et al spanned 28 degrees of latitude (ok not tones but did go temperate to tropics) and also found significant latitudinal gradients in seed predation once standardized seeds were used. Are there equivalently large studies that did not show latitudinal gradients to justify the ‘conflicting evidence’ angle (which is not needed to justify the current study on seed production)?

The results / Discussion really rains on small scale studies, but readers have not been given convincing evidence that current study does any better. (would do if methods were not at the end). Make this clear in last intro paragraph that introduces MASTIF

172 – think you mean ‘first put forward’ rather than ‘first recognized’?

I encourage the authors to have another go at avoiding vague pronouns – there are still a lot and they really do cloud the meaning of the writing. Eg
176 – ‘while at same time intensifying them’ – is ‘them’ about seed predation or other interactions (eg competition)?
L180 – one what? one host? One consumer?
L183 – arms race between which species? trees (as in sentence before)? Trees vs their enemies?
Other examples:
L186 – clarify this paragraph is about tree fecundity
L208 – variation in what?
L250 – which ‘both’ variables?
L308 – means that values of what are over-estimates?

L184 – ‘if ecologists preoccupation with them’ – phrasing doesn’t reflect very kindly on ecologists

I don’t personally find Fig 1a useful. Maybe as its not clear what some of the terms or arrows mean from the figure itself (eg gradient, amplification).
Fig 1b is better, but yaxis is confusing. Is it about fecundity? If the units are about change as labelled, shouldn’t zero change be zero? (if its just a ratio call it a ratio rather than a change)

Fig legend: 50 – is dash supposed to be a minus sign, in which case I’m confused, or functioning as a colon, in which case switch to colon or other punctuation to avoid confusion)

L212 – ‘If so, climate gradients would be a driver of intense biotic interactions…’. this connection is still not clear. I think you mean ‘This could mean that climate drives…’, but again I think you’re missing a logical leap. If fecundity gradients cannot be explained by climate, this implies that stronger interactions in the tropics drive increased seed production. But what drives stronger interactions is not clear (climate is not the only driver of interaction strength explored in the references listed). Also doesn’t last sentence of paragraph offer an alternative explanation for why gradients in seed production might be stronger than gradients in NPP?

L221 – isn’t this the hypothesis invoked in 2nd sentence of paragraph before (rather than a new mechanism as phrased)?

L233 – not clear how gradients in variation in seed production are relevant to gradients in seed production itself – elaborate

L236 – if study includes non-forest areas it cant be presenting data as per forest area

L241 – to my mind ‘determine’ is too strong for the broad correlations with an equally broad climate variable (annual temperature). ‘Assess’ or ‘quantify’ would be fine

L251 – stronger trends in fecundity than NPP is not ‘providing the first evidence that amplification influences species interactions’ – this is too causal but also the wrong direction of inference (amplification -> interactions). There are still no data on interactions. You could maybe say ‘strongly suggesting that species interactions contribute to latitudinal gradients in tree fecundity’

Paragraph starting L259 was pretty hard to parse.
L259 – mean ‘dwarf’? ‘dominate’ often implies ‘contribute to’
L260 – not following – what is within-tree (over time) variation and between tree variation? First time these terms are used. Are these seen within your data? (Supp figures can support a claim but readers shouldnt have to go to the supp mat to understand the claim)
L261 – storage of what? what does pollen supply have to do with arguments?

L264 – ‘The average seed production for 95% of trees of a given size varies…’ – is this within a site or across latitudes?

L280 suggest changing ‘report on’ to ‘synthesis / analysis of’. Think ‘report’ underplays the work that goes into a global meta-analysis

Fig 2 – suggest removing ‘but either was high fecundity stimulates those interactions’ – this is not shown in the figure and confuses the last point

Paragraph starting 311 – the point of this paragraph seems to be to undermine current data on latitudinal patterns in seed predation, but the logic is not clear and nor is the point of doing so. Arguments focus on the ‘short’ time frame of these studies, but don’t clarify why this matters. If the implication is that these studies are not capturing true gradients in seed predation come out and say so and justify why. The short duration is not a self-evident justification – why would you expect a different pattern if seeds were left out longer? In many of these studies 100% of seeds are eaten within 24 hours at some sites, so its not clear what longer times would add. As in Intro, I’m not sure why authors of current study (on seed production) need to undermine studies on seed predation. Claiming that the seed predation literature is inconclusive doesn’t add to the novelty of the present study, so why put so much emphasis on the perceived shortcomings of the seed predation literature vs the results?

Again in the next paragraph L320 The present study does NOT add to literature on actual predation rates, it complements it by asking a different question using a different approach. It DOES add to the literature on seed production so make that comparison clear instead. Also haven’t told readers at this point how big in spatial or temporal scale current study is so points fall a bit flat

L312 – does ‘host’ here mean seed species? That is an odd use of ‘host’
L317 – I agree that leaf damage is irrelevant, don’t think it adds anything here (suggest removing everything about leaf damage)