Review Reports - Reproductive Success Beyond Pollinators: Microhabitat Effects and Pollen Dynamics in <i>Epipactis bugacensis</i>, a Traditionally Obligately Autogamous Orchid

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. Summary and overall assessment
This manuscript investigates the reproductive biology of Epipactis bugacensis, combining (i) repeated flower-visitor observations with (ii) a pollinator-exclusion (net-covering) field experiment at two Hungarian sites (JF and HK). The authors report a potentially novel pollen-transfer pathway in which halictid bees fragment pollinia and transport fragments (including via scopal loads), which could enable geitonogamy and potentially xenogamy. They further find that net-covered plants show markedly higher fruit set at JF (53.4% vs 22.1%; odds ratio = 4.05, p < 0.001).
The work is interesting and potentially important for orchid reproductive ecology, but the central inference “pollinator exclusion increased reproductive success” is currently difficult to interpret causally because the netting plausibly alters microclimate and physical exposure in addition to excluding visitors, and the manuscript itself acknowledges this possibility.
I therefore recommend major revision, primarily to strengthen causal attribution, statistical treatment of hierarchical data, and evidentiary support for the proposed pollen-fragment transfer mechanism.
2. Major comments
Major comment 1. Causal interpretation: pollinator exclusion vs “netting effects” (microclimate/physical protection)
A key result is that net covering increases fruit set and other reproductive traits, and the authors note that the net covering may modify microclimatic conditions and provide physical protection.
The Discussion further argues that higher conversion under covering is unlikely to be pollinator-driven and may reflect indirect benefits (reduced mechanical damage, reduced herbivory/pollen robbing, enhanced humidity, reduced stress).
What is needed for rigor:
1) Treat “net covering” explicitly as a compound treatment (pollinator exclusion + microhabitat modification).
2) Provide direct measurements (or at least structured proxies) of the non-pollinator pathways: temperature/humidity within vs outside nets; wind exposure; evidence of herbivory/mechanical damage; and prevalence of pollen robbers.
3) If adding new measurements is impossible (e.g., protected species constraints), the manuscript should reframe the causal claim: the experiment tests the netting intervention, not pollinator exclusion in isolation.
A simple addition with high value would be a small microclimate sub-study (even a few loggers) or repeated spot measurements during the critical flowering-to-early-fruiting interval.
Major comment 2. Experimental design and treatment allocation: imbalance and potential site-specific confounding
The allocation approach (every third plant covered; others uncovered) is described as systematic and “unbiased,” but the final sample sizes are extremely unbalanced at JF (11 covered vs 91 uncovered), because uncovered individuals were intentionally increased for power.
This imbalance can be acceptable, but it materially affects analysis choices and interpretation.
Suggestions:
1) Clarify whether the transects traverse microhabitat gradients (light, litter depth, soil moisture). Systematic allocation can still align with gradients, especially in linear transects.
2) Provide a short assessment of baseline comparability between treatments (e.g., plant height, bud number, immediate microhabitat descriptors). You already counted bud number at JF before covering. Use that information to test balance or include it as a covariate if available.
3) Consider reporting how many transects were used, their lengths, and whether treatment assignment was restarted per transect or continuous across all transects (this matters for spatial autocorrelation).
Major comment 3. Statistical analysis: hierarchical structure and (likely) pseudo-replication
Several outcomes are measured at levels that are nested (capsules within plants; flowers within plants; plants within sites). For example, fruit set is evaluated at the flower level using totals for covered vs uncovered at JF.
Seed number and capsule volume are measured for multiple capsules per individual, and for a subset all capsules were collected to study position effects.
Why this matters: treating each flower/capsule as independent can inflate effective sample size and reduce p-values. Fisher’s exact test on pooled flower counts ignores within-plant clustering and plant-level heterogeneity.
Recommended approach (strongly):
1) For fruit set: use a binomial mixed model (GLMM) with plant random intercepts (and potentially an observation-level random effect if overdispersion is present), or a beta-binomial model. Report effect size with 95% CI (you already report OR; add CI).
2) For capsule volume/seed number: use mixed models with plant as a random effect and fixed effects for site, treatment, and their interaction. The manuscript already highlights site × treatment interaction qualitatively.
3) For capsule-position analyses: because regressions are performed per individual (Panels A–J), consider a hierarchical slope model to summarize population-level position effects while acknowledging between-plant variability.
Even if the authors retain nonparametric tests, they should justify independence assumptions and, ideally, complement them with plant-level aggregated analyses (e.g., per-plant mean seed number) as a sensitivity check.
Major comment 4. Evidence for the proposed “pollinium fragmentation transfer” mechanism
The reported behavior (pollinium disaggregation, fragments adhering to pistils and potentially transferred to other flowers/individuals) is central to novelty.
The Discussion acknowledges that the mechanism was documented in a limited number of visitation events but argues it is mechanistically sufficient.
To strengthen this claim:
1) Quantify observational effort and event frequency: total observation hours, number of halictid visits, number of fragmentation events, and how often fragments were observed on stigmas or exported to adjacent flowers. (Some observation protocol details are present; make the linkage to the key phenomenon explicit.)
2) Clarify whether halictid individuals were identified to species or morphospecies, and whether captured individuals carried E. bugacensis pollen fragments in scopa (microscopy evidence would
be compelling).
3) Distinguish clearly between (i) deposition of fragments within the same flower (autogamy facilitation) versus (ii) transfer to other flowers/individuals (geitonogamy/xenogamy). The abstract makes strong claims about both.
Consider tempering wording if direct evidence for cross-flower transfer is limited.
Major comment 5 . Non-target organisms (ants, aphids) and treatment integrity
The manuscript reports ants exploring flowers and that pollinia were absent in flowers with ants present.
It also states that two covered individuals were excluded after non-target organisms were detected beneath nets.
These points imply that arthropods other than pollinators could materially influence pollen availability and reproduction.
Recommendations:
1) Report how frequently non-target organisms were detected under nets and how often nets were inspected.
2) Provide a structured assessment of whether the 0.2 mm mesh plus leaf-litter sealing fully excludes small arthropods (or whether the treatment primarily excludes larger visitors while permitting some smaller taxa).
3) Consider including an explicit “treatment integrity” section, because “pollinator exclusion” is only valid if access is reliably prevented.
3. Minor comments / editorial suggestions
1) Report confidence intervals for key effect sizes (e.g., odds ratio for fruit set).
2) Clarify capsule-volume computation (provide the exact ovoid approximation formula and how radii/width map into parameters).
3) Data availability: “available upon reasonable request” is acceptable, but a public repository (even with sensitive metadata removed) would substantially improve reproducibility, given the multiple analytical steps (SciPy/NumPy/pandas).
4) Site descriptions: The climatic and habitat differences between JF and HK are detailed; consider a short paragraph connecting these site differences to the observed site × treatment interaction in reproductive traits.
5) Mortality analysis:Mortality differences at JF are reported as marginal (p = 0.0664). Consider presenting an effect size (risk ratio) with CI and discussing whether mortality could mediate observed reproductive outputs (survivorship bias).
6) Tone precision: where the manuscript states “previously unreported pollen-transfer mechanism,” ensure claims are narrowly phrased to what is directly evidenced in this dataset.
4. Recommendation
Major revision. The manuscript has a potentially valuable contribution (especially the mechanistic observation of pollinium fragmentation and the explicit recognition that netting can alter reproduction via microhabitat pathways).
However, the causal interpretation of treatment effects and the statistical handling of nested data structures need to be strengthened before the main conclusions can be considered robust.

Author Response

Major comment 1.: Causal interpretation: pollinator exclusion vs “netting effects” (microclimate/physical protection)
A key result is that net covering increases fruit set and other reproductive traits, and the authors note that the net covering may modify microclimatic conditions and provide physical protection.
The Discussion further argues that higher conversion under covering is unlikely to be pollinator-driven and may reflect indirect benefits (reduced mechanical damage, reduced herbivory/pollen robbing, enhanced humidity, reduced stress).
What is needed for rigor:
1) Treat “net covering” explicitly as a compound treatment (pollinator exclusion + microhabitat modification).
2) Provide direct measurements (or at least structured proxies) of the non-pollinator pathways: temperature/humidity within vs outside nets; wind exposure; evidence of herbivory/mechanical damage; and prevalence of pollen robbers.
3) If adding new measurements is impossible (e.g., protected species constraints), the manuscript should reframe the causal claim: the experiment tests the netting intervention, not pollinator exclusion in isolation.

Response 1:

We thank the reviewer for this important and well-founded comment. We fully agree that the net-covering treatment cannot be interpreted just as pollinator exclusion in isolation. Accordingly, we have revised the manuscript to explicitly treat net covering as a compound intervention, combining (i) exclusion of larger flower-visiting insects with (ii) modification of the local microenvironment and physical protection of developing buds and flowers.

Although the experiment was originally designed to limit access of animal visitors, we acknowledge that under in situ field conditions such an intervention inevitably tests the netting treatment itself, rather than pollinator exclusion alone. We have therefore reframed causal statements throughout the manuscript to avoid attributing treatment effects exclusively to pollinator absence and to emphasize that the observed increases in reproductive success likely reflect the integrated outcome of pollinator exclusion, physical protection, and microhabitat buffering.

Direct measurements of microclimatic variables (e.g., temperature, humidity, wind exposure) were not conducted, as repeated instrumentation or logger deployment was not feasible due to the strictly protected status of Epipactis bugacensis and site-access constraints. We note that there is currently no direct literature on orchid pollinator-exclusion netting that jointly quantifies microclimate modification and pest protection effects. However, experimental and agricultural studies in other plant systems consistently demonstrate that comparable netting interventions reduce solar radiation and wind exposure, buffer temperature fluctuations, slightly increase relative humidity, and provide an effective physical barrier against herbivores and pests. These findings strongly support the plausibility of similar mechanisms contributing to the observed effects in E. bugacensis.

To improve transparency regarding treatment integrity, we have added explicit information on inspection frequency. At the Jászfényszaru site, nets were checked daily during the flowering period. At the Harkakötöny site, daily inspection was not possible due to logistical constraints; this limitation is now clearly stated in the Methods and Limitations sections.

Overall, these revisions clarify that our experiment evaluates the ecological consequences of a netting intervention under natural conditions, rather than a pure test of pollinator dependence, and our conclusions have been adjusted accordingly.

Major Comment 2.: Experimental design and treatment allocation: imbalance and potential site-specific confounding.
The allocation approach (every third plant covered; others uncovered) is described as systematic and “unbiased,” but the final sample sizes are extremely unbalanced at JF (11 covered vs 91 uncovered), because uncovered individuals were intentionally increased for power.
This imbalance can be acceptable, but it materially affects analysis choices and interpretation.

Major Comment 2.1) Clarify whether the transects traverse microhabitat gradients (light, litter depth, soil moisture). Systematic allocation can still align with gradients, especially in linear transects.

Response 2.1:
Both study sites (JF and HK) consist of even-aged Populus × euramericana plantations established on flat calcareous sandy soils (planted in 1997 at JF and 1998 at HK; see Section 4.2)

The stands are structurally homogeneous and lack pronounced topographic variation.

Transects were laid out within these uniformly managed plantation compartments, where no visible microtopographic gradients (e.g., slope, depressions, elevation differences) were present. During fieldwork, we recorded no systematic spatial pattern in estimated canopy closure, litter depth, or soil surface conditions along transects.

Canopy cover was consistently high (approximately 90–95%) at both sites, resulting in similarly shaded understory conditions throughout the sampled areas. Litter depth typically ranged between 3–5 cm and appeared spatially uniform within each site. Soil surface conditions (calcareous sand) also showed no visually detectable moisture gradients along transects at the time of plant selection. The herb layer was sparse and structurally simple (Section 4.2), further suggesting limited small-scale environmental heterogeneity.

Based on these observations, we consider it unlikely that systematic treatment allocation (every third individual covered) aligned with any consistent microhabitat gradient. Therefore, the systematic assignment method is not expected to have introduced site-specific confounding related to light, litter depth, or soil moisture gradients within transects.

Major Comment 2.2) Provide a short assessment of baseline comparability between treatments (e.g., plant height, bud number, immediate microhabitat descriptors). You already counted bud number at JF before covering. Use that information to test balance or include it as a covariate if available.

Response 2.2:

At the JF site, we recorded the number of buds per individual immediately before applying the exclusion nets (Methods, Section 4.3.2). To assess baseline comparability between treatments, we compared pre-treatment bud/flower numbers between individuals assigned to the covered vs. uncovered groups using a non-parametric Mann–Whitney U test (given the unbalanced sample sizes).

Baseline bud/flower numbers did not differ significantly between treatments (covered: median = 9, IQR = 4.5–15.0, n = 11; uncovered: median = 7, IQR = 4.0–11.0, n = 91; Mann–Whitney U = 603.5; p = 0.268).

Therefore, the observed treatment differences in fruit set and reproductive output are unlikely to be driven by pre-existing differences in initial reproductive potential.

Major Comment 2.3) Consider reporting how many transects were used, their lengths, and whether treatment assignment was restarted per transect or continuous across all transects (this matters for spatial autocorrelation).

Response 2.3:

At the JF site, sampling was conducted along five spatially separated transects located 20–50 m apart within the same poplar plantation compartment. At the HK site, four transects were used. All of them were situated in separate plantation member, more than 50 m apart. Individuals were selected sequentially within each transect, and treatment allocation (every third plant designated as covered) was applied within transects.

The use of multiple spatially separated transects substantially reduces the likelihood that treatment categories are aligned with consistent environmental gradients. Particularly at the HK site, the multi-kilometer separation between transects further limits the potential influence of spatial autocorrelation across the study area.

Given the absence of visible microtopographic gradients and the scattered distribution of individuals within homogeneous plantation stands, spatial confounding between treatment categories is considered unlikely to have materially affected the results.

Major comment 3.: Statistical analysis: hierarchical structure and (likely) pseudo-replication.
Several outcomes are measured at levels that are nested (capsules within plants; flowers within plants; plants within sites). For example, fruit set is evaluated at the flower level using totals for covered vs uncovered at JF.
Seed number and capsule volume are measured for multiple capsules per individual, and for a subset all capsules were collected to study position effects.
Why this matters: treating each flower/capsule as independent can inflate effective sample size and reduce p-values. Fisher’s exact test on pooled flower counts ignores within-plant clustering and plant-level heterogeneity.
Recommended approach (strongly):
Major comment 3.1) For fruit set: use a binomial mixed model (GLMM) with plant random intercepts (and potentially an observation-level random effect if overdispersion is present), or a beta-binomial model. Report effect size with 95% CI (you already report OR; add CI).

Response 3.1:

We thank the reviewer for highlighting the hierarchical structure of the fruit-set data. We agree that flowers are nested within plants and that pooled contingency analyses do not account for within-plant clustering.

In the revised manuscript, we therefore complemented the pooled Fisher’s exact test with a binomial generalized linear mixed model (GLMM) fitted at the flower level, including treatment as a fixed effect and plant identity as a random intercept. This model explicitly accounts for the non-independence of flowers within individuals.

The mixed-model analysis supported the same qualitative and quantitative conclusion as the pooled analysis (covered vs. uncovered OR = 6.01, 95% CI: 2.99–12.06). For transparency, we also report the pooled odds ratio with its 95% confidence interval (OR = 4.05, 95% CI: 2.69–6.09). Overdispersion was assessed and was not substantial.

These results indicate that accounting for plant-level clustering does not alter the inference that net covering is associated with a substantially higher probability of flower-to-capsule conversion at the JF site.

Major comment 3.2) For capsule volume/seed number: use mixed models with plant as a random effect and fixed effects for site, treatment, and their interaction. The manuscript already highlights site × treatment interaction qualitatively.

Response 3.2:

We agree that capsule-level traits are nested within plants. To address the potential inflation of effective sample size, we complemented the original capsule-level nonparametric comparisons with a plant-level re-analysis using aggregated values as independent units. For the standard dataset (two lowest capsules per individual at both sites), we calculated per-plant mean capsule volume, mean seed number per capsule, and mean seed density, and compared these plant-level values across the four site × treatment groups. This plant-level approach supported the same qualitative conclusions as the capsule-level analyses: reproductive output (capsule volume and seed number) was highest at the JF site, particularly under covered conditions, whereas seed density remained broadly similar across groups. As a robustness check, we also fitted capsule-level mixed-effects models with plant identity as a random intercept; these analyses were consistent with the plant-level interpretation.

Major comment 3.3) For capsule-position analyses: because regressions are performed per individual (Panels A–J), consider a hierarchical slope model to summarize population-level position effects while acknowledging between-plant variability.
Even if the authors retain nonparametric tests, they should justify independence assumptions and, ideally, complement them with plant-level aggregated analyses (e.g., per-plant mean seed number) as a sensitivity check.

Response 3.3:

We appreciate the reviewer’s suggestion regarding hierarchical slope modeling. In addition to the per-individual regression analyses presented in Figures 4–6, we fitted linear mixed-effects models including capsule position as a fixed effect and plant identity as a random intercept and random slope.

These population-level models confirmed the overall basal-to-apical decline in seed number and capsule volume, while also revealing substantial between-plant variability in slope magnitude. In contrast, seed density showed no consistent population-level position effect, supporting our original interpretation that seed packing efficiency remains relatively stable along the inflorescence.

Thus, the hierarchical modeling approach corroborates the qualitative conclusions drawn from the individual regression panels.

Major comment 4.: Evidence for the proposed “pollinium fragmentation transfer” mechanism.
The reported behavior (pollinium disaggregation, fragments adhering to pistils and potentially transferred to other flowers/individuals) is central to novelty.
The Discussion acknowledges that the mechanism was documented in a limited number of visitation events but argues it is mechanistically sufficient.
To strengthen this claim:
Major comment 4.1.) Quantify observational effort and event frequency: total observation hours, number of halictid visits, number of fragmentation events, and how often fragments were observed on stigmas or exported to adjacent flowers. (Some observation protocol details are present; make the linkage to the key phenomenon explicit.)

Response 4.1:

We thank the reviewer for requesting clarification of the observational basis of the proposed pollinium fragmentation mechanism.

Direct behavioral observations were conducted for approximately two hours (10:00–12:00) on 4 June 2018 at the JF site. During this period, three halictid visitation events were documented and photographed in detail. The revised manuscript now explicitly states that we have seen pollinium fragments two times on the stigmas due to the activity of sweat bees, and that we have seen three times pollinia fragment transportation by them: first at 11:06 when the halictid bee arrived at the flower already carrying visible pollen fragments on its legs; second, when this individual moved from that flower to the adjacent upper flower with pollinium fragments adhering to its body hairs; and third, when it departed from the upper flower bearing visibly increased amounts of fragmented pollinium material.

In all observed cases, pollinium disaggregation by chewing and by the first pair leg scraping was clearly visible.

Major comment 4.2.) Clarify whether halictid individuals were identified to species or morphospecies, and whether captured individuals carried E. bugacensis pollen fragments in scopa (microscopy evidence would be compelling).

Response 4.2:

The visiting bees were identified as halictid bees (Hymenoptera: Halictidae) based on morphological characteristics observed in the field and in photographs. Individuals were not captured for species-level identification, as capturing them would have risked disturbance to the highly protected plant species. Consequently, no microscopic analysis of pollen carried in the scopa was performed. This limitation has been clarified in the revised manuscript.

However, visible pollinium fragments adhering to the scopal and other body hairs were clearly documented photographically. The presence of these fragments on pollen-transporting structures during flower-to-flower movement is explicitly described in the revised text.

Major comment 4.3.) Distinguish clearly between (i) deposition of fragments within the same flower (autogamy facilitation) versus (ii) transfer to other flowers/individuals (geitonogamy/xenogamy). The abstract makes strong claims about both.
Consider tempering wording if direct evidence for cross-flower transfer is limited.

Response 4.3:

We agree with the reviewer that it is essential to distinguish between directly observed and inferred processes. The observations provide direct evidence for:

mechanical pollinium disaggregation;
intra-floral (autogamous) deposition of fragmented pollen onto the stigma;
adhesion of pollinium fragments to pollen-transporting body hairs;
transport of pollinium fragments between flowers on the same individual (geitonogamous transport); and
repeated contact between pollen-bearing body hairs and stigmatic surfaces during such movements.

While pollen-bearing hairs repeatedly contacted stigmas during movement among flowers on the same plant, direct experimental confirmation of successful inter-individual (xenogamous) pollen transfer was not obtained. The manuscript has therefore been revised to distinguish clearly between documented autogamous and geitonogamous pollen deposition events and the potential for subsequent cross-individual transfer during continued visitation.

Statements in the Abstract and Discussion have been carefully tempered to avoid overstating cross-flower or cross-individual transfer beyond the observational evidence. The novelty of the study lies in documenting pollinium disaggregation behavior and its immediate reproductive consequences, while extended pollen transfer remains a plausible - but not experimentally verified - outcome.

Major comment 5.:Non-target organisms (ants, aphids) and treatment integrity
The manuscript reports ants exploring flowers and that pollinia were absent in flowers with ants present.
It also states that two covered individuals were excluded after non-target organisms were detected beneath nets.
These points imply that arthropods other than pollinators could materially influence pollen availability and reproduction.
Recommendations:
Major comment 5.1) Report how frequently non-target organisms were detected under nets and how often nets were inspected.

Response 5.1:

We thank the reviewer for raising important concerns regarding treatment integrity and the potential influence of non-target arthropods.

As stated in response to Major Comment 1, nets were inspected daily at the Jászfényszaru (JF) site throughout the flowering period. At the Harkakötöny (HK) site, nets were inspected every 3–4 days due to logistical constraints. No non-target arthropods were detected beneath nets at the HK site during inspections. At the JF site, two covered individuals were found to harbor non-target organisms beneath the netting and were excluded from subsequent analyses to maintain treatment integrity.

Major comment 5.2) Provide a structured assessment of whether the 0.2 mm mesh plus leaf-litter sealing fully excludes small arthropods (or whether the treatment primarily excludes larger visitors while permitting some smaller taxa).

Response 5.2:

The 0.2 mm mesh size was designed to exclude larger flower-visiting pollinators. While this mesh likely prevents the entry of most pollinators, complete exclusion of extremely small arthropods – such as thrips, as discussed in Section 3.4 (Limitations of our results) – cannot be guaranteed. Leaf-litter sealing around the base of the nets was used to minimize ground-level entry; however, we interpret the treatment primarily as exclusion of larger flower visitors rather than total arthropod exclusion.

Major comment 5.3) ) Consider including an explicit “treatment integrity” section, because “pollinator exclusion” is only valid if access is reliably prevented.

Response 5.3:

The rarity of non-target detections and the exclusion of the two affected individuals ensure that the observed treatment effects reflect effective restriction of access by flower-visiting organisms exceeding the 0.2 mm mesh aperture size. Thus, differences between covered and uncovered plants can be attributed primarily to limitation of access by larger flower visitors.

Minor comments / editorial suggestions

1) Report confidence intervals for key effect sizes (e.g., odds ratio for fruit set).

Response: We agree and have added 95% confidence intervals for key effect sizes. Specifically, for fruit set at the JF site we now report the odds ratio from Fisher’s exact test (OR = 4.05, 95% CI: 2.69–6.09) and, to account for flowers nested within plants, the odds ratio from the complementary binomial GLMM with plant random intercept (OR = 6.01, 95% CI: 2.99–12.06). These additions are included in the Results section.

2) Clarify capsule-volume computation (provide the exact ovoid approximation formula and how radii/width map into parameters).

Response: We thank the reviewer for this helpful suggestion. We have clarified the capsule-volume computation in the Methods section. Capsule volume was approximated as a prolate spheroid (ovoid) using the formula V = (π/6) × L × W², where L denotes total capsule length and W the maximum capsule width. We also specify its equivalence to the ellipsoid equation V = (4/3)πa²b, where a = W/2 and b = L/2. This clarification ensures full reproducibility of the volume calculations.

3) Data availability: “available upon reasonable request” is acceptable, but a public repository (even with sensitive metadata removed) would substantially improve reproducibility, given the multiple analytical steps (SciPy/NumPy/pandas).

Response:
We appreciate the reviewer’s suggestion to enhance reproducibility through public data deposition. The dataset underlying this study includes detailed locality information for a strictly protected orchid species, as well as extensive primary measurements derived from long-term field and laboratory work. For conservation and stewardship reasons, and in line with responsible data management practices for sensitive biological datasets, we have opted not to deposit the complete raw dataset in an unrestricted public repository.

However, all data supporting the conclusions of this study are available from the corresponding author upon reasonable request. We are committed to providing access to researchers with legitimate scientific interest, and all analytical procedures are fully described in the manuscript to ensure reproducibility of the reported results.

4) Site descriptions: The climatic and habitat differences between JF and HK are detailed; consider a short paragraph connecting these site differences to the observed site × treatment interaction in reproductive traits.

We agree and note that the Discussion (Section 3.2.2) already explicitly links the documented habitat differences between JF and HK to the observed site × treatment interaction in capsule volume and seed number. We have reviewed this section to ensure that this connection is clearly articulated.

5) Mortality differences at JF are reported as marginal (p = 0.0664). Consider presenting an effect size (risk ratio) with CI and discussing whether mortality could mediate observed reproductive outputs.

Response: We thank the reviewer for this helpful suggestion. In addition to the reported p-value, we now provide the relative risk (RR = 1.67, 95% CI: 0.83–3.34) for mortality at the JF site. Although the difference did not reach conventional statistical significance, uncovered individuals exhibited a higher mortality trend. We also note in the Discussion that differential survival could potentially influence estimates of reproductive output through survivorship filtering.

6) Tone precision: where the manuscript states “previously unreported pollen-transfer mechanism,” ensure claims are narrowly phrased to what is directly evidenced in this dataset.
Response:
We thank the reviewer for this important suggestion. We have revised the phrasing throughout the manuscript to ensure that novelty claims are strictly limited to what is directly evidenced in our dataset. Specifically, we now refer to a “previously undocumented instance of pollinium fragmentation–mediated pollen movement in E. bugacensis,” thereby avoiding broader generalization beyond the documented observations.

THE UPLOADED FILE CONTAINS ONLY THE REVISIONS MADE IN RESPONSE TO THIS SPECIFIC REVIEW, HIGHLIGHTED IN RED WITHIN THE MANUSCRIPT.

THE FINAL REVISED VERSION, INCORPORATING THE MODIFICATIONS SUGGESTED BY BOTH REVIEWERS, WILL BE UPLOADED TO THE MAIN SUBMISSION PAGE SHORTLY.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

My comments are shown below.

L.40-55: here you should definitely cite Claessens & Kleynen (2011)

L.61: “In addition”

L.73-74: According to Sramkó et al. (2019), supports the possible treatment of these four taxa as a single species but later you refer that the two of them, namely E. bugacensis and E. dunensis, are treated as distinct. I think that it would be good to refer to the researchers/authors that treat them as distinct.

L.78: in treating as what?

L.83: non-functional instead of dysfunctional

L.97: I would suggest using lip instead of labellum.

L.320: taxa is plural. You may wanted to write taxon as you wrote “one flower visitor”.

L.442: This is just a hypothesis and is not supported by the results of the study. The fact that at the JF site both capsule size and seed number were larger compared to the HK site, cannot be definitely attributed to the local conditions. To support something like that with confidence you need to apply an experimental design with several sites having varying environmental conditions. IN L. 443 you refer to the higher humidity at the JK site. However, in L.620 you say that the climate in HK is more humid than in JK. Moreover, the humidity at the microsite level is also influences by the vegetation. In your case, the vegetation was consisted by poplar plantations. Here the query is related to the age and the plantar ligament of the plantation in both sites. Was that the same? Moreover, was the age of the plantations the same? These are factors that might have affected local microclimatic conditions but nothing was written about them.

L.447-449: again this is an assumption.

L.458: add references

L.566: The value used present the conservation value of the studied species cannot be understood, unless you are Hungarian. Either delete it, or clarify it.

L.570-586: if the description was not based on morphological traits measured by the authors, they should add references.

L.597: “In Hungary, it …..”

L.639: what are these 13 individuals? Were these the ones you monitored? If yes, refer to these before this line.

L.640: although the term tepala refers to monocotyledons, in Epipactis, and in general in orchids, is not used. Instead of this term, the terms sepals and petals are used.

L.660: clarify what you mean with the word “specimens”. Do you mean flowers, insects or what?

L.662: What is written in the small-sized paragraph and in the way it is written gives the impression that is not connected to the main process of the visitors’ observation.

L.671: along which transects?

L.670-673: this part needs to be rewritten. In its current form, one can hardly understand what you did and why.

L.674-675: why only at the JF site? What the number of covered/uncovered individuals would be without increasing it?

L.685: why only at JF?

L.687-689: if by that, you mean that a number of individuals was covered, then why it is written here and not before?

L.718-720: It is unclear how did you count the seeds per capsule. Did you count them one by one (almost impossible I would say) or did you use any kind of statistical/photographical method? How did you assess seeds density?

Figs. 4, 5 & 6: write on the graphs what X- and Y-axes represent. Maybe it would be better to present them all individuals together per treatment.

Additional references

Claessens, J., & Kleynen, J. (2011). The Flower of the European Orchid. Form and Function. Jean Claessens and Jacques Kleynen: Schrijen-Lippertz, the Netherlands.

Author Response

Comment 1: L.40-55: here you should definitely cite Claessens & Kleynen (2011)

Response 1: We have added the requested citation in the Introduction and included the full reference in the References section.

Comment 2: L.61: “In addition”

Response 2: The typo has been corrected (“I addition” → “In addition”).

Comment 3: L.73-74: According to Sramkó et al. (2019), supports the possible treatment of these four taxa as a single species but later you refer that the two of them, namely E. bugacensis and E. dunensis, are treated as distinct. I think that it would be good to refer to the researchers/authors that treat them as distinct.

Response 3: We have added references to the POWO database entries [31,32] which treat E. bugacensis and E. dunensis as distinct taxa.

Comment 4: L.78: in treating as what?

Response 4: The sentence has been clarified to explicitly state that we follow the taxonomic treatment of E. bugacensis as a distinct species according to Plants of the World Online [31,32]

Comment 5: L.83: non-functional instead of dysfunctional

Response 5: We have replaced “dysfunctional” with “non-functional”.

Comment 6: L.97: I would suggest using lip instead of labellum.

Response 6: We have adopted the term “lip” throughout the manuscript and clarified its first occurrence as “lip (labellum)”.

Comment 7: L.320: taxa is plural. You may wanted to write taxon as you wrote “one flower visitor”.

Response 7: We have corrected “taxa” to “taxon” where a singular form was required. The remaining occurrences refer to multiple taxa and therefore remain unchanged.

Comment 8: L.442: This is just a hypothesis and is not supported by the results of the study. The fact that at the JF site both capsule size and seed number were larger compared to the HK site, cannot be definitely attributed to the local conditions. To support something like that with confidence you need to apply an experimental design with several sites having varying environmental conditions. IN L. 443 you refer to the higher humidity at the JK site. However, in L.620 you say that the climate in HK is more humid than in JK. Moreover, the humidity at the microsite level is also influences by the vegetation. In your case, the vegetation was consisted by poplar plantations. Here the query is related to the age and the plantar ligament of the plantation in both sites. Was that the same? Moreover, was the age of the plantations the same? These are factors that might have affected local microclimatic conditions but nothing was written about them.

Response 8: We appreciate the reviewer’s important methodological remark. We have revised L.442–444 to avoid definitive attribution of reproductive differences to local environmental conditions. The sentence now reads: “This pattern may reflect site-specific microenvironmental differences within the poplar plantations,” thereby clearly framing the interpretation as tentative rather than causal.

In addition, we have expanded the Materials and Methods section to clarify that transects were established within structurally homogeneous, even-aged (21–25 years old) poplar plantation stands located on flat terrain, with comparable canopy closure (approx. 90–95%), litter depth (3–5 cm), and no visible microtopographic variation. This addition addresses the reviewer’s concern regarding potential plantation-age or structural differences between sites.

Comment 9: L.447-449: again this is an assumption.

Response 9: We thank the reviewer for this comment. We agree that the original wording could be interpreted as overly definitive. We have therefore revised this section to adopt a more tentative formulation, indicating that the large effect sizes suggest biological relevance while avoiding mechanistic interpretation. The revised text now explicitly frames statements regarding potential developmental constraint as speculative.

Comment 10: L.458: add references

Response 10: We have added the appropriate references [78,79] to support this statement.

Comment 11: L.566: The value used present the conservation value of the studied species cannot be understood, unless you are Hungarian. Either delete it, or clarify it.

Response 11: We have removed this statement.

Comment 12: L.570-586: if the description was not based on morphological traits measured by the authors, they should add references.

Response 12: We have added reference [19].

Comment 13: L.597: “In Hungary, it …..”

Response 13: The sentence has been corrected by inserting the missing comma after “In Hungary”.

Comment 14: L.639: what are these 13 individuals? Were these the ones you monitored? If yes, refer to these before this line.

Response 14: We have clarified that the “13 individuals” refer to the number of simultaneously visible flowering shoots of E. bugacensis from a fixed observation point at the JF site used for pollinator monitoring.

Comment 15: L.640: although the term tepala refers to monocotyledons, in Epipactis, and in general in orchids, is not used. Instead of this term, the terms sepals and petals are used.

Response 15: The term “tepals” has been replaced with “sepals” and/or “petals” throughout the relevant sections.

Comment 16: L.660: clarify what you mean with the word “specimens”. Do you mean flowers, insects or what?

Response 16: The term “specimens” has been clarified and replaced with “flower visitors” and “insects,” as appropriate.

Comment 17: L.662: What is written in the small-sized paragraph and in the way it is written gives the impression that is not connected to the main process of the visitors’ observation.

Response 17: We thank the reviewer for this observation. We have clarified that these records represent supplementary opportunistic observations conducted independently of the systematic monitoring of the 13 focal individuals. The paragraph has been revised to explicitly distinguish these observations from the main observation protocol.

Comment 18: L.671: along which transects?

Response 18: These observations were independent of the systematic monitoring of the 13 focal individuals. The manuscript has been revised to clarify this distinction.

Comment 19: L.670-673: this part needs to be rewritten. In its current form, one can hardly understand what you did and why.

Response 19: We thank the reviewer for this suggestion. The relevant section has been rewritten to more clearly describe the experimental design and its rationale.

Comment 20: L.674-675: why only at the JF site? What the number of covered/uncovered individuals would be without increasing it?

Response 20: We thank the reviewer for this question. At the HK site, increasing the number of covered individuals was not feasible due to logistical constraints associated with regular inspection requirements. The pollinator-exclusion treatment required frequent monitoring to ensure treatment integrity, which limited the number of individuals that could be reliably maintained at the more distant site. Without increasing sample size at JF, the number of covered and uncovered individuals would have been 11 and 22, respectively.

Comment 21: L.685: why only at JF?

Response 21: The description refers to methodological aspects of treatment allocation and has therefore been retained in the Materials and Methods section in accordance with the journal’s structural guidelines.

Comment 22: L.687-689: if by that, you mean that a number of individuals was covered, then why it is written here and not before?

Response 22: The description refers to methodological aspects of treatment allocation and has therefore been retained in the Materials and Methods section in accordance with the journal’s structural guidelines.

Comment 23: L.718-720: It is unclear how did you count the seeds per capsule. Did you count them one by one (almost impossible I would say) or did you use any kind of statistical/photographical method? How did you assess seeds density?

Response 23: We thank the reviewer for this question. Seeds were counted individually (one by one) under a stereomicroscope for each capsule. Counting was repeated until consistent totals were obtained. Seed density was calculated as the number of seeds per capsule divided by capsule volume. The Methods section has been revised to clarify this procedure. Alternative estimation approaches [92–95] were explored but did not provide sufficient accuracy.

Comment 24: The X- and Y-axes have now been clearly labeled on all relevant figures (Figs. 4–6) to improve clarity and interpretability.

Response 24: The X- and Y-axes have now been clearly labeled on all relevant figures (Figs. 4–6) to improve clarity and interpretability.

In the annotated manuscript accompanying this response, only the revisions made in response to Reviewer 2 are highlighted. Revisions addressing the comments of Reviewer 1 are highlighted in the separate annotated version submitted with that response. A fully integrated version incorporating all revisions will be provided as the final revised manuscript.