Overall, this paper only utilized simple ordination analysis and multiple regression analysis to briefly examine the indicative significance of surface pollen on vegetation types and climate in the Tibetan Plateau. It did not carefully analyze the relationship between pollen assemblages of different vegetation types and the actual vegetation. In the western and northern Tibetan Plateau, the vegetation is sparse, dominated by steppe-desert, but the surface pollen contains an uncertain amount of arboreal pollen, which is evidently transported from distant sources. The proportion and types of long-distance transported pollen may vary across different locations, and the sources could also be different - these are issues that need to be cautiously considered when reconstructing past vegetation and climate based on surface pollen (quantitative) data. Therefore, this paper only provides some surface pollen data, without clear implications for other disciplines.

Response: As you may know that this paper is a data paper, the main purpose of this paper is to provide original pollen count data. We believe that these data will be beneficial for future researchers in conducting diverse studies related to vegetation/paleovegetation, climate change, and other related fields. Revolving around this purpose, we have provided comprehensive details on sample collection and pretreatment, methods for pollen identification, frequency and spatial distributions of pollen count for both samples and sites, the spatial distributions of the main pollen types, and the frequency distribution of the number of identified pollen types from different vegetation types. Through the application of statistical analyses, we have proposed potential uses for the dataset and cautionary notes for its utilization. Based on the results obtained, we aim to show the readers that whether or not the pollen data from this dataset can be used to distinguish vegetation types or can be reliably used to distinguish which vegetation types, and which climatic parameters are most promising to reconstruct based on the pollen data from this dataset. As for your comment that “It did not carefully analyze the relationship between pollen assemblages of different vegetation types and the actual vegetation.”, we understand that you think we have not processed or selected the pollen data to ensure that it accurately reflects the vegetation types. However, regarding the question of how to effectively process the data to enhance the reliability of the pollen data in representing vegetation types or climatic parameters—such as excluding arboreal taxa from data derived from non-forested area, or using representation value or relative pollen production to adjust the percentages of specific pollen taxa—we consider this a potential research area of interest for future researchers who will utilize this dataset, potentially leading to the publication of further research papers.

For the second question, we completely agree with you that pollen data collected from open landscape contains an uncertain amount of long-distance transported pollen types, with sources potentially located as far as over 50 kilometers away. Undoubtedly, this issue must be meticulously considered when reconstructing past vegetation and climate using surface pollen data. We will emphasize this concern in the section “5 Potential usage of the dataset”. Besides, we will add another result from the ordination analysis conducted on a processed pollen dataset that excludes samples from forested areas and removes arboreal pollen types. By doing so, we aim to show readers whether excluding arboreal pollen types from data collected in open landscape can enhance the indicative significance of the pollen data to vegetation types. You can check the result from the updated Figure 5 attached.

Q1 (Line 10): This dataset fills a geographic gap in the published pollen count data for this region. To allow the reader to clearly see which locations are covered by this new dataset, the geographic distribution of the previous data should be indicated in Figure 1.

Response: Thanks very much for this suggestion. We have revised Figure 1 to include the geographical locations of the original pollen count data from our study (indicated by circles) and from other publicly available sources (indicated by triangles, including Cao et al., 2021 and Ma et al., 2024). Please check the updated Figure 1 attached.

Q2 (Lines 23-24): Italic

Response: Thanks for pointing out this problem. We will make correction.

Q3 (Line 73): The References for the Tibetan Plateau boundary.

Response: thanks for this suggestion: The reference related to the boundary of the Tibetan Plateau is: Zhang, Y., Li, B., Liu, L., and Zheng, D. (2021) Redetermine the region and boundaries of Tibetan Plateau, Geographical Research, 40, 1543-1553, https://doi.org/10.11821/dlyj020210138, 2021.

Q4 (Line 79): In the transition zone around 3,000 meters of elevation, how can the delineation of distinct vegetation types be ensured

Response: We did not attempt to distinguish different vegetation types based on altitude. We just want to state a fact that, the samples we collected from temperate vegetation generally distributed between 800-3000 m.a.s.l. and those from alpine vegetation above 3000 m a.s.l, but not to classify the vegetation below 800-3000 m a.s.l as temperate vegetation types and those above 3000 m a.s.l. as alpine vegetation types. In fact, as we mentioned on Line 80, we determined the vegetation types for the sampling sites based on the local vegetation types.

Q5 (Line 83): Whether the pollen assemblage results obtained are representative of the different sampling densities in each vegetation area, which vary widely. If there are only three samples in a coniferous forest area, are they representative of the entire coniferous forest area?

Response: Thanks very much for this question. We acknowledge that three samples from only three sites can hardly show the entire picture of pollen assemblages from a vegetation covering a large geographic space. We will point out this problem in the section “5 Potential usage of the dataset” and emphasize the uncertainty may arise when using these data.

Q6 (Line 88): What is the weight of each sample?

Response: The dry weights of the samples in our dataset range from 5.25 to 32.7 g. Approximately 77% (427 samples) of the total samples have dry weights exceeding 20 g, 14.8% (82 samples) weigh between 10 and 20 g, and the remaining 8.2% (46 samples) weigh between 5 and 10 g. The dry weights of the samples used for pollen extraction were determined by the soil properties, particularly the grain size and the local vegetation surrounding the sampling sites. As mentioned in the manuscript, the weight of each sample has been included alongside the pollen data in the dataset.

Q7 (Line 110): Italic

Response: Thanks for pointing out this problem. We will make correction.

Q8 (Line 119): Does such a large gap in pollen grain counts all have an impact on analyses such as NMDS?

Response: Intuitively, we believe that a large gap in pollen counts would have some impact on the analyses, but the question on how significant the impact needs to be further studied. As you surely know, theoretically, 300 grains is the low limitation for pollen identification. Through communications with other researchers engaged in pollen studies on the Tibetan Plateau, we learned that for areas with sparse vegetation and few plant species, 200 grains may be enough to capture the compositional characteristics of the pollen assemblages. There is a detail that we did not emphasize in the manuscript but may be important information for readers that, the pollen data we used for analyses are the combined pollen data from multiple samples at each site. In the combined dataset, there are 9 pollen data have pollen counts less than 200 grains. Therefore, we re-analyzed the pollen dataset excluding these 9 data points and updated Figures 5 and 6 accordingly (please see updated Figures 5 & 6 attached). We will add these details on the data and data selection in the section “3.3 Numerical analyses”. In the updated Figure 5, the upper panel shows the results including samples from all vegetation types, while the lower panel shows the results including samples from non-forested area.

Q9 (Lines 122-125): Figure 2 shows that 31% of the samples contain between 200 and 300 pollen grains, and 44% contain between 400 and 600 grains. In other words, 75% of the samples have more than 200 grains. However, the key question is what percentage of samples contain fewer than 100 grains. While Figure 2 indicates the sample points with under 100 grains, it does not specify the overall percentage. If more than 20% of the samples have under 100 grains, the dataset should be reorganized to remove those low-grain samples.

Response: In our dataset, there are 310 samples have pollen count more than 300 grains, which occupy 55.86% of the total samples. 1etween 168 samples (30.27%) have pollen count between 200- 300 grains, 43 samples (7.75%) between 100-200 grains, and 34 samples (6.12%) have pollen count less than 100 grains. When considering pollen count for each site (as you may know from the manuscript that, we generally collected multiple samples from each site), there are only 3 sites (TP2019071102, TP2019071103, TP2019071202) have pollen count fewer than 100 grains. We kept these data in the dataset because we believe it might be useful to reflect some of the characteristics of vegetation. Since we not only provide pollen count data, but also the dry weight of each sample used for pollen extraction, Lycopodium grains per tablet, and the identified number of Lycopodium spores for each sample, pollen concentration of each sample can be calculated, and maybe some researchers might be interested in testing the reliability of using pollen concentration to reflect the sparseness of vegetation or can use pollen concentration to explore other scientific questions.

We realized from this question you raised that it is necessary to provide more details on the pollen counts for samples and sites, and we will add the information above into the manuscript.

Q10 (Line 161): Is it possible to consider the effect of altitude on pollen assemblages?

Response: If we have understood you question correctly, you want to know whether the pollen assemblage from our dataset can serve as indicators of changes along the altitude gradient. The answer to that is yes, as evidenced by the pollen diagram attached. However, it is important to note that the relationship between pollen assemblages and altitude is not "linear"; in other words, pollen assemblages vary across different altitude ranges. Within a specific altitude range, such as between 4065-4520 meters, it becomes challenging to discern changes in altitude based solely on the pollen assemblages.

Q11 (Line 173): Why does the A/Am also show high values in the alpine desert. The pollen assemblage should be dominated by Amaranthaceae.

Response: As you can observe in the figure 4, Amaranthaceae comprises approximately 30~50% of the pollen taxa in alpine deserts, whereas Artemisia is even more abundant, reaching up to 50~70% or even higher. This explains why we obtain higher values of the A/Am ratio for this vegetation type. According to the plant survey data collected from the same sample sites, Amaranthaceae indeed dominates in temperate desert, but this is not the case in alpine deserts. For more detail, you may refer to the published plant community dataset in the Chinese Journal of Plant Ecology (https://www.plant-ecology.com/CN/10.17521/cjpe.2022.0174). Besides, as you surely know that the representativeness of Amaranthaceae and Artemisia to their parent plants vary significantly across different vegetation types, which is another reason why the A/Am ratio is high in alpine deserts on the plateau.

Thanks very much for your positive comment on our manuscript.

Regarding your question on data screening, we actually did not screen the data but provided all the pollen data identified from the collected samples. The reason why we collected 664 samples but only obtain 555 pollen data is that when we pretreated samples collected during 2018 (a total of 172 samples from 63 sites) for pollen identification, we mixed samples collected from different plots within the same site. For samples collected during 2019-2022 (a total of 492 samples from 244 sites), we did not mix any sample but pretreat each of them separately. We have mentioned this issue in the section “3.2 Pollen analysis” but perhaps it was not explained clearly enough. We will elaborate on this in more detail in the revised manuscript.

Thanks again for your comments.

General Comments

1. Regarding the impact of climate on pollen assemblages: while the results at the plateau scale suggest precipitation is the most influential factor, considering the spatial heterogeneity of the Tibetan Plateau, would region-specific statistical analyses yield different outcomes? Additionally, only temperature and precipitation were considered as climate variables—could other climatic factors, such as solar radiation or wind speed, also play a role?

Reply: Thanks for your suggestions. We agree and admit that regional-specific statistical analyses must be yield different results. However, we must address that our work takes the plateau as whole. Thus, there is no necessary to do regional-specific analysis. Furthermore, we also agree that other climatic and topographic factors must cast considerable impacts on vegetation and pollen compositions. We have data for 23 bioclimatic variables. However, we consider it inappropriate to include all these variables in the ordination analysis due to collinearity issues among them and the potential for poor readability of the resulting biplots. So, we firstly conducted Pearson correlation tests between each pair of variables and we removed those with a correlation coefficient exceeding 0.8. Ultimately, seven bioclimatic variables were retained for the ordination analysis: BIO1 (Annual Mean Temperature), BIO2 (Mean Diurnal Range), BIO3 (Temperature Seasonality), BIO4 (Isothermality), BIO12 (Annual Precipitation), BIO14 (Precipitation of Driest Month), and BIO23 (Moisture Index) were retained for ordination analysis. Additionally, we included the intensity of human disturbance (HD) as a predictor in the updated ordination analysis. HD shows low to moderate correlations (r < 0.5) with the 23 bioclimatic variables. We have added details regarding the selection of bioclimatic variables in the 'Materials and methods' section and revised the results pertaining to the ordination analysis in the “Potential usage of the dataset” section.

1. Based on local vegetation data, only a small proportion of the samples/sites are classified as forest, yet the abundance of forest taxa ranks second only to temperate desert taxa. This discrepancy may be influenced by pollen productivity and dispersal A brief discussion on this aspect would be valuable.

Reply: Yes, only several forest quadrats are investigated at eastern Tibetan Plateau and most of the quadrats are located in the middle and western Tibetan Plateau with vegetation of meadow, steppe and desert. We admit that most of the pollen data contains arboreal pollen and must be transported from regional forest within and also beyond the plateau. It is clear that this issue roots in pollen productivity and dispersal distance. We have added a brief discussion on this aspect in the second paragraph of the section 4.1 as “It should be noted that, due to the limited sample size (n = 3) collected from the coniferous forest, the number of pollen taxa for this vegetation type is likely underestimated. Additionally, issues related to differences in pollen production and dispersal capacity may also influence the number of identified pollen taxa. Pollen types with a high pollen production and wide dispersal commonly dominate a pollen assemblage, thereby reducing the probability of detecting pollen types with a poorer representation and/or a low production in the pollen source area (Odgaard, 1999; Weng et al., 2006).”.

1. The manuscript frequently compares the presented dataset with other published datasets. Beyond differences in data format (percentages vs. counts), please elaborate on any differences, such as differences in nomenclature and classification systems. If researchers aim to integrate your dataset with others, what considerations should be taken into account?

Reply: This is really important comments that will benefit us and other researchers. We have included detailed information on the nomenclature and classification systems used for pollen types in our dataset. Additionally, we propose specific considerations for merging this dataset with others, as outlined below:

“The nomenclature of pollen types in our dataset generally follows the conventions outlined in the Flora of China (https://www.iplant.cn/). However, we have retained the pollen type Chenopodiaceae despite its taxonomic inclusion within the family Amaranthaceae as per the Angiosperm Phylogeny Group (APG) classification. Therefore, if researchers intend to integrate our dataset with others, they should pay particular attention to standardizing the nomenclature of pollen types.”

Specific Comments

1. Figure 1: For visual clarity, please remove the minutes and seconds from the latitude and longitude labels.

Reply: revised already.

1. Line 105: Given the use of 1-km resolution climate data, does this resolution accurately represent the spatial scale inferred by modern topsoil pollen data? Please clarify the spatial representativeness of modern pollen assemblages.

Reply: To be frank, we are unable to clarify this issue due to vast spatial scale covered by our dataset. Atmospheric convection regimes on the plateau and its surrounding area are complex, with different atmospheric circulations prevail in different regions. Consequently, the dispersibility of a specific taxon may vary noticeably across variation parts of the study area. We admit that, in open landscapes, the pollen source area can be extensive, and the 1-km resolution employed in this study may not accurately represent the spatial scale inferred from modern topsoil pollen data. Since our statistical analyses focused on the entire Tibetan Plateau and the southern margin of Xinjiang adjacent to the plateau, this vast area exhibits pronounced environmental gradients. Consequently, we argue that 1-km climate data can adequately capture the overall climate gradient, as exemplified by the Mean Annual Precipitation (MAP) gradient depicted in Figure 4, and thus provide reliable results.

1. The manuscript repeatedly refers to human disturbance. How do the levels of human impact vary across different sites in this dataset? Has human activity altered the composition of pollen assemblages?

Reply: According to the published plot-level vegetation data (Jin et al., 2022, http://doi.org/10.17521/cjpe.2022.0174), the intensity of human disturbance was classified into to 3 levels (1 to 3), representing “weak”, “moderate”, and “strong”, respectively. From the figure in the attached file, you can find that strong human disturbance is concentrated along the southern margin of Xinjiang adjacent to the plateau and in the northeastern part of the Tibetan Plateau. In contrast, weak human disturbance is predominantly observed in the southwestern part of the plateau, while moderate disturbance is primarily distributed in the central or south-central region of the plateau.To figure out whether human activity has altered the composition of pollen assemblages in our dataset, we have included human disturbance as a predictor variable when conducting the ordination analysis. The result indicates that human disturbance did have some impact on the composition of pollen assemblages, though this impact was not crucial. Please refer to the updated Figure 5 in the attached file.

1. Line 215: The integration of pollen data with plant functional traits offers deeper insights into long-term vegetation responses to past climatic disturbances. Please briefly explain how data of differing resolutions were integrated in this analysis.

Reply: If we understand corerctly, you are referring to discrepancies in spatial resolution between the pollen source area and the size of the plant quadrat. We acknowledge that the pollen source area in open vegetation is significantly larger than the spatial scale of the vegetation surveys (1 m×1 m for meadow, steppe, and desert, and 2 m×2 m or 5 m×5 m for shrubland). The distinct difference in spatial scale between the pollen source area and the vegetation survey may result in a mismatch between the relative abundance of pollen taxa and the relative importance of their parent plants. Consequently, this discrepancy will inevitably introduce bias when calculating indices of plant functional traits using pollen data. We have mentioned this issue in another paper comparing functional traits derived from pollen and plant assemblages, which is currently under review. Frankly, resolving such a discrepancy remains challenging.

1. Line 246: The dataset's continued expansion is anticipated, particularly in the southeastern and northwestern regions of the Tibetan Plateau.

Reply: Thanks for this comment. We have already collected hundreds of samples from the southeast Tibetan Plateau, and these samples are currently undergoing pretreatment.