the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 May 2026
Bacterial Community Structure and Predicted Functional Potential Across Andean Lakes with Contrasting Preservation Status Based on 16S rRNA Gene Profiles
Abstract. Northern Andean highland lakes play a key role in supporting agriculture and local livelihoods, but they are increasingly affected by anthropogenic pressures such as agricultural expansion and urbanization. These pressures can alter water quality and ecosystem functioning, yet their effects on microbial communities remain poorly understood. Here, we analyze bacterial community structure and predicted functional potential across four freshwater lakes in the Eastern Cordillera of Colombia, representing contrasting levels of environmental preservation. Using 16S rRNA gene sequencing, we characterized microbial diversity and community composition in two lakes subject to higher anthropogenic influence (Fúquene and Tota) and two relatively well-preserved systems (Calderona and Colorado). Lakes with lower preservation status exhibited higher microbial richness and diversity, likely associated with increased nutrient inputs and environmental heterogeneity. In contrast, more preserved lakes showed lower diversity but more stable community composition. Spatial heterogeneity also played a key role, with larger systems displaying greater horizontal variability linked to tributary inflows and extensive littoral zones, as well as vertical stratification driven by gradients in light, temperature, oxygen, and nutrients. Co-occurrence patterns among dominant genera suggest environmentally structured microbial associations across lakes. Inferred functional profiles indicate differences in predicted metabolic potential among preservation statuses, particularly in pathways related to carbohydrate metabolism and xenobiotic degradation. However, these results represent inferred functional capacity based on taxonomic composition rather than direct measurements of microbial activity. Overall, our findings indicate that environmental preservation status and spatial heterogeneity shape both the composition and predicted metabolic potential of bacterial communities in Andean freshwater ecosystems. These results provide a baseline for understanding microbial responses to environmental change and highlight the need for future studies integrating metagenomic and experimental approaches to validate ecosystem processes.
- Preprint
(1389 KB) - Metadata XML
-
Supplement
(172 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2137', Anonymous Referee #1, 29 Jun 2026
-
RC2: 'Comment on egusphere-2026-2137', Anonymous Referee #2, 02 Jul 2026
This study explores the diversity of bacterial communities and their functional potential in several Andean lakes subject to varying levels of anthropogenic pressure. Using amplicon sequencing and functional prediction based on 16S rRNA gene taxonomy, the authors observed greater bacterial diversity in lakes exposed to greater anthropogenic influence than in better preserved environments, as well as differences in the representation of metabolic pathways related to carbohydrate metabolism and xenobiotics degradation. They also observe some spatial heterogeneity and vertical stratifications within the lakes.
This work may be useful as a preliminary study of the condition of the lakes and will be all the more valuable if sampling continues over the years. However, the present study has several significant weaknesses. Firstly, the number of samples is unfortunately too small to allow for statistical analysis: six in total for four lakes, and for Lakes Tota and Fúquene, the two samples taken were from different seasons or years. Secondly, most of the statistical tests presented in the study are not valid given the number of samples, in particular Spearman’s correlation. Thirdly, the authors make several claims regarding the results and their significance that the dataset and analyses cannot support.
I therefore recommend that the manuscript be thoroughly revised, that the analyses of the dataset be redone, and that the article be rewritten accordingly. The study would benefit greatly from increasing the size of the dataset, with at least true triplicates per lake, collected simultaneously. If more extensive sampling is not possible, a more descriptive preliminary study is possible, but the analyses of the dataset must be thoroughly revised, with appropriate use of the statistical tests, which will then be limited. Finally, I would have liked to have metadata concerning the samples, such as weather conditions and certain characteristics of the lakes (pH, temperature, etc.). With the exception of the 16S rRNA gene sequencing data, all metadata is missing from the study. These factors could influence bacterial communities and affect the interpretation of the results.
In its current form, the article doesn’t meet the journal’s publication criteria.
Major comments:
The abstract will need to be revised in accordance with the new analyses and the revision of several statements, such as L25: “vertical stratification driven by …” I have not found any analysis of this kind in the manuscript.
L59-61: If all these described processes have already been extensively documented, what novelties your manuscript is bringing? This sentence or the objective of the study needs to be reworded.
L120: In what way is the sample taken from Lake Tota in 2018 considered an internal control? How was it taken into account in the analyses? Further details are required.
L116-118: Unfortunately, the dataset is too limited to reveal any variation between the conditions (at least three samples per site, taken on the same date, would be required). Furthermore, the experimental protocol and the selection of samples need to be reconsidered. We have a mix of conditions: a single sample for two lakes, and two samples for two other lakes, but two of these are from the same year but a different season, and two others are from a different year. No trends can be observed with such experimental protocol.
L117: I found the term “semester” rather vague and would recommend specifying the date or month instead (or the season if there are several samples). This makes it even more difficult to identify a seasonal trend, or is this aspect relevant given the local climate?
L139: I would have liked a brief explanation as to why this particular protocol was chosen. Why was a home-made protocol chosen rather than a commercial kit, which would have allowed for a better comparison of the data with that from other publications? Is this due to technical reasons and the type of samples?
L138-149: Please elaborate a little further on the description of the methodology here. Was a pre-amplification PCR carried out prior to sequencing? Were the PCR and MiSeq sequencing steps carried out in-house or performed by a sequencing platform? If in-house, which kits were used?
L160: I think there’s a sentence missing here to explain how the ASVs were trained, and what parameters were used for the clustering?
L177: Spearman’s correlation test, like correlation tests in general, requires a considerable number of samples to produce reliable results. Please review your analysis.
L210: I am not entirely sure how the differential analysis carried out here with 2 samples per treatment group (or less?). Please revise your analysis to ensure you only use statistical tests that are appropriate for your dataset.
L233: Please could you explain and clarify in the ‘Methodology’ section how you filtered out ASVs ‘with no name available’? Did this filtering apply only to unclassified ASVs, or also to those with multiple affiliations at certain taxonomic levels?
L236-242: I don’t think it’s possible to observe a statistically significant difference between just two samples. This echoes my general comments on the proper use of statistical tests.
L271: I believe the authors are referring to ‘alpha diversity’ in this section? Please use standard terminology for diversity analysis. Once again, the authors must be very cautious when interpreting the results of statistical tests carried out on such a small sample.
L281-283: This is an overstatement that is not sufficiently supported by your results and statistical tests. It is not possible to demonstrate temporal or spatial heterogeneity by comparing just two samples. The same applies to several of the claims made in the discussion.
L215-295: As mentioned above, I recommend either omitting this part of the analysis or re-running it using a larger sample size. The Spearman’s correlation test has been misapplied here.
L305: once again, it is not possible to draw this conclusion from a sample of this size.
L319: Why this interest? Please provide a more detailed explanation of the interest in these taxa and mention it in the introduction.
L380: Please reformulate the sentence by removing the word “significant”. This applies to statistical tests, and your results do not support such a claim.
L381: I do not see how these results could be linked to the size of the lake and its spatial heterogeneity. Please rephrase your argument and be careful not to overstate your point.
L400: Please modify terms such as “stable”, “heterogeneity”, “significant”, etc. The results don’t support these claims.
L401-403: overstatement
L421-423: Based on what? Your dataset can’t support such statement.
L429-441: See my previous comments about co-occurrence and Spearman test. I suggest strongly to remove this part.
L474-476: overstatement
L492-498: I suggest removing this part too. I have serious doubt about the robustness of such differential abundance analyses. The size of the dataset doesn’t support it.
Figures:
Fig1. Panel a does not really provide any information and could be incorporated into a table, for example Table1. The captions for panels e and f are difficult to follow. I would recommend using a more common visualisation, such as the one in panel d.
Fig2. Panel a does not currently contribute anything to the analysis or discussion. I would recommend removing this figure or replacing it with a box plot containing a larger number of samples.
Fig4. The captions are not readable.
Minor comments:
L93-94: “the sea level, and is the largest remaining…”? This comment applies to several places in the manuscript. The text will need to be checked to ensure it does not contain errors of this kind.
L126: Please avoid giving the reference twice and write instead “Paver and coworkers (2020).
L138: Please correct the title by removing “Assembly”. I don’t think any assembly step was done here.
Citation: https://doi.org/10.5194/egusphere-2026-2137-RC2 -
RC3: 'Comment on egusphere-2026-2137', Anonymous Referee #3, 14 Jul 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2137/egusphere-2026-2137-RC3-supplement.pdf
Data sets
Supplementary information Andrés Gómez-Palacio https://doi.org/10.5281/zenodo.19571462
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 180 | 41 | 15 | 236 | 20 | 14 | 13 |
- HTML: 180
- PDF: 41
- XML: 15
- Total: 236
- Supplement: 20
- BibTeX: 14
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The authors used 16S rRNA gene amplicon sequencing combined with PICRUSt2 functional prediction to systematically analyze the bacterial community structure, co-occurrence patterns, and potential metabolic functions of four high-altitude lakes (F ú quene, Tota, Calderona, Colorado) with different conservation states in the eastern Andes Mountains of Colombia. Research has found that lakes with greater human influence, such as F ú quene, exhibit higher microbial richness and diversity, while lakes with better protection, such as Calderona and Colorado, have more stable communities but lower diversity; Spatial heterogeneity (such as inflow rivers and coastal zones) and vertical stratification further shape community composition. Functional prediction reveals potential differences in bacterial communities in pathways such as carbohydrate metabolism and exogenous substance degradation under different protection states. This study provides baseline data for understanding the response of microorganisms in Andean mountain lakes to anthropogenic stress, and has certain ecological and conservation biological value.
However, despite the practical significance of the research topic, there are several significant deficiencies in the rigor of experimental design, data analysis, and generalizability of the conclusions. Taking all factors into consideration, I believe that the manuscript is currently not sufficient for publication in this journal. I suggest making significant revisions and reconsidering it.
The main issue with this manuscript is the severe lack of sample size and sampling design, which greatly undermines the reliability of all statistical inferences and ecological conclusions. In addition, the explanation of functional prediction results is too bold and lacks necessary cautious wording.
1.Rows 91-130, Table 1: The sample size is extremely small and the statistical power is seriously insufficient. This study only involved 4 lakes, and the number of sampling points for each lake was extremely small (2 samples from F ú quene, 2 samples from Tota, and 1 sample each from Calderona and Colorado). Such a small sample size cannot support any meaningful inter group statistical tests (such as comparisons between different conservation states), nor can it distinguish individual differences in lakes from conservation state effects. The so-called 'significant differences' (such as lines 237-241) are likely driven by a single outlier.
Suggest the author to significantly increase the sampling intensity. Each lake should have multiple sampling stations (at least 3-5) and repeat sampling in different seasons. If it cannot be achieved, this study should be honestly positioned as a "preliminary exploratory case study" and avoid making generalizations in the title, abstract, and conclusion.
2.Rows 116-120, Table 1: Pseudo repetition and time confusion. Two samples from F ú quene come from two different periods in 2019, one sample from Tota comes from 2018 and the other from 2019, while Calderona and Colorado only have one sample from 2019. This uneven and mixed time factor makes it difficult to distinguish the impact of "conservation status", "lake characteristics", and "sampling time" on community variation. For example, the high diversity of F ú quene may be due to its two samplings capturing seasonal fluctuations, rather than its low conservation state.
It is recommended that all lakes be sampled at the same time. If historical data (such as Tota's 2018 data) must be used, it should be analyzed separately as a control, rather than mixed with other 2019 data for comparison.
3.Lines 28-29, 85-88, 197-208, 354-370, 467-50: Overinterpretation of functional prediction (PICRUSt2) results. Although the author pointed out the limitations of PICRUSt2 in the methods section (lines 85-88), deterministic language such as "predicted metabolic potential variables" and "pathways related to... were identified" is frequently used in the results and discussion section. Especially when directly associating KEGG pathway annotations such as "infectious diseases" with potential pathogenic risks in environmental samples (lines 368-370), it can easily lead to misunderstandings. These are only computer predictions based on genome homology, not evidence.
Suggest revising this type of expression throughout the entire text. The word 'identified' should be changed to 'predicted to be present based on 16S rRNA gene induced metagenomics'; Change 'variation in... pathways' to' variation in the predicted abundance of genes associated with... pathways'. For the "infectious diseases" pathway, it must be clearly stated that this is only an artificial product annotated in the database and does not represent any pathogenic activity.
4.Lines 189-196, 319-338: Species level identification is unreliable. The author constructed a phylogenetic tree based on the V3-V4 16S rRNA gene fragment (approximately 460bp) and conducted species level identification of Mycobacteria. As is well known, the resolution of 16S rRNA genes at the species level is limited, especially for highly similar NTM groups. The author himself acknowledges this (lines 326-328), but still provides a detailed species level discussion below (lines 444-456).
Suggest either deleting all species level identification and discussions based on this fragment, and keeping the analysis at the genus level; Alternatively, it should be clarified that these are only "tentative species level assignments" and it is recommended to confirm them through whole genome sequencing or specific marker genes in the future.
5.The article discusses multiple driving factors such as nutrients, oxygen, and temperature, but lacks environmental physicochemical data.
During the discussion, it was repeatedly mentioned that factors such as "nutrient inputs," "temperature," and "oxygen gradients" shape microbial communities, but the entire text did not provide any synchronously measured environmental variable data (such as total phosphorus, total nitrogen, dissolved oxygen, chlorophyll-a, pH, etc.). This makes all inferences about environmental drivers unfounded. This is a fatal flaw.
It is recommended to supplement at least basic on-site hydrochemical parameters. If the data has been lost, it should be explicitly stated in the discussion that environmental microbiological correlation analysis cannot be conducted and the relevant inferences should be significantly weakened.
6.Lines 175-188, 286-296: The limitations of co-occurrence network analysis have not been fully discussed. The network is constructed based on only 6 samples (N=6), with 20 nodes and 226 edges. Calculating correlation on such extremely sparse datasets results in extremely unstable results, almost entirely driven by individual samples. Although the author mentioned in lines 186-188 and 526-527 that statistical associations do not equal ecological interactions, they did not emphasize the fundamental issues brought about by sample size.
It is not recommended to conduct network analysis under the current sample size. Alternatively, the author should explicitly downgrade this section to 'exploratory visualization' and warn readers that the results cannot be extrapolated.
The statement "Temporary stability" in lines 297-306 is not valid.
Only two time points of data (F ú quene and Tota twice each) claim 'temporary stability in dominant taxa'. Two time points can only describe changes and cannot prove stability.
Suggest deleting the word 'stability' and replacing it with 'temporary variation between two sampling events'.
7.It is recommended to conduct detailed language polishing and formatting proofreading throughout the entire text.