Acta Biologica Sibirica

Introduction

In the study of past climate change and associated vegetation transformation, one of the main methods is the analysis of fossil pollen spectra (Xu et al. 2016). The central problem of such studies is the interpretation of the percentage of certain taxa pollen in the selected spectrum. In most cases, the interpretation of pollen spectra is based on the analysis of the presence and abundance of selected indicator taxa (Gaceur et al. 2017, Lee et al. 2024) and is presented as the concentration or percentage of pollen types. However, in some cases, indicator taxa are absent or, in contrast, abundantly represented, which significantly complicates the interpretation of the results obtained. Therefore, various pollen ratios have been developed, such as Artemisia/Cyperaceae (A/Cy) (Herzschuh 2007, Zhao & Herzschuh 2009), Pinus/Artemisia (Xu et al. 2007); Poaceae/Artemisia (Po/A) (Tang et al. 2009), Cyperaceae/Poaceae (Cy/Po) (Cour et al. 1999), Cyperaceae/(Poaceae + Artemisia) (Cy/(Po + A)) (Tang et al. 2009), Artemisia/Betula (A/B), Poaceae/(Artemisia + Chenopodiaceae) (Po/(A + C) (Fowell et al. 2003), Artemisia/Chenopodiaceae (A/C) as semiquantitative indicators (Zhang et al. 2018).

The Artemisia/Chenopodiaceae (A/C) ratio is used as an indicator of temperature and aridity variability in a number of regions: arid and semi-arid areas of China (Herzschuh 2007, Huang et al. 2009), southeastern Mongolia (Liu et al. 2006), and mountains. In addition to arid and semi-arid regions (most often desert and steppe areas, respectively), this ratio has been calculated for meadows and forests (Zhao et al. 2012). However, such studies are mainly local in nature, which limits the interpretation of data on fossil pollen spectra and reconstruction for other sites. For a more detailed distinction between steppe and forest-steppe landscapes, the ratio of pollen from the the sum of percentages of wormwood and goosefoot family divided by the percentage of grasses (Artemisia + Chenopodiaceae)/Poaceae (A+C)/P) is used, since with increasing humidity, the proportion of grass pollen in the composition of phytocenoses increases.

Several studies have been conducted on modern surface pollen spectra for the Republic of Altai and Altai Krai (Pelánková & Chytry 2009; Blyakharchuk et al. 2023, Kuryatnikova et al. 2024). However, insufficient attention has been paid to the study of semiquantitative indicators, especially for the plain part of Altai Krai located within the steppe and forest-steppe zones. It is worth noting that the results of the palynological analysis of lake sediments in the steppe zone are available for this territory (Rudaya et al. 2012, 2020), but the ratios of indicator taxa based on modern pollen spectra have not been calculated, which determines the relevance of this work. Furthermore, this area represents a large regional ecological environment. On the one hand, this is due to its location in the contact zone between the plains of Western Siberia and the mountain structures of the Altai-Sayan Mountain Country. On the other hand, it is crossed by an imaginary line that delimits humid and arid landscapes based on the balance of heat and moisture. The long-term and large-scale impact (primarily plowing) has also significantly influenced the composition and structure of vegetation. The results of these studies will expand our understanding of modern pollen-vegetation relationships and provide valuable information for reconstructing long-term climate change and vegetation dynamics.

The aim of the work is to test the effectiveness of the calculated pollen ratios A/C and (A + C)/P within the steppe and forest-steppe zones of Altai Krai, as well as to identify the conditions and limitations of their use to study modern pollen-vegetation relationships in steppe and forest-steppe landscapes.

Materials and methods

The material for this study was based on recent samples collected using two schemes in the plain and foothill parts of Altai Krai belonging to the steppe and forest-steppe zones. The territory includes three large geomorphological regions: the Kulundinskaya Lowland Plain, the Priobskoye Upland Plato, and the Predaltaiskaya Piedmont Plain (foothills). Figure 1 shows the physiographic regionalization of the territory and the location of sampling points. The first scheme involved sampling using Tauber traps installed in accordance with the requirements of the international Pollen Monitoring Program (Hicks et al. 1996). Traps were installed from 3 May to September 20, 2022, in the steppe and forest steppe zones (Table 1). Although the installation dates did not coincide with the beginning of tree pollination, this did not affect the accumulation of pollen for Artemisia, Chenopodiaceae, and Poaceae, which bloom primarily in the second half of the growing season. According to the pollen calendar for adjacent territories (radius 100–500 km from sampling points), graminoids pollen grains are recorded in the air only in the third decade of May, while those of wormwoods and goosefoots in the second decade of June (Nenasheva 2013). The second scheme is topsoil sampling at the end of the 2025 growing season. The sampling according to this scheme was implemented only in the steppe zone. Sampling points were selected to cover the most representative sites that encompass the full diversity of zonal, intrazonal, and extrazonal landscapes. Despite the fact that the samples we obtained reflect the pollen spectra over different time periods (Tauber traps cover a single growing season, while soil samples represent the mean pollen spectrum over several last years) and were collected in different years (2022 and 2025), their comparison may well be correct. This is primarily due to the fact that the pollen contribution of each taxon to the spectrum fluctuates from year to year within certain limits determined by the natural zone, location, and vegetation of the sampling point. Therefore, their interannual variability is not stochastic (random) and does not lead to sharp discrete distortions, making the comparison of the 2022 and 2025 sample groups representative for calculating semiquantitative indices. However, to obtain more representative results and subsequently compare them with other studies, we performed separate analyzes of the data obtained for the traps and soil samples. Zonal comparisons were made only for the Tauber trap data.

Laboratory sample preparation of the collected samples was carried out according to a modified von Post method (Grichuk & Zaklinskaya 1948) including the following steps: 10% KOH, sieving, HF, 10% HCl. For soil samples with a large number of mineral particles, a heavy liquid KJ + CdJ2 + H2O was also used. Microscopic analysis of the prepared samples was performed on a Carl Zeiss Axio Lab A1 light microscope at a magnification of 400 times. Atlases were used for taxonomic identification (Kupriyanova & Aleshina 1972, 1978; Karpovich et al. 2013). For each sample, at least 450 pollen grains were counted, which is considered statistically significant in palynological studies. According to E.D. Zaklinskaya, 150 grains is the minimum number for which the error margins remain the same as for counting larger numbers of grains (Pollen analysis 1950). The sum of pollen from woody and herbaceous taxa was taken as 100%, and the percentage of taxa was calculated relative to this sum. The A/C ratio was calculated as Artemisia%/Chenopodiaceae%, and the (A + C)/P ratio – (Artemisia% + Chenopodiaceae%)/Poaceae%.

Due to the small sample group size, a combination of statistical methods was used to analyze the calculated ratios. To compare the calculated ratios and subsequently differentiate between zones using pollen trap data installed in the steppe and forest steppe zones, the Mann-Whitney U test was used, because it is optimal for small sample groups. To assess how well the calculated ratios indicate belonging to the steppe or forest-steppe zones, a ROC analysis with Leave-One-Out cross-validation was performed. To examine the relationships between the calculated ratios and climatic parameters (mean annual precipitation (MAP) and mean annual temperature (MAT), we computed the Pearson coefficient and linear regression.

MAP and MAT values were obtained from the WorldClim 2.1 dataset, which presents data for 1970–2000 (Fick & Hijmans 2017). The choice of this particular database is due to its widespread use in palynological research. This data set is a spline interpolation of weather observation data with a spatial resolution of 1 km. Data extraction for each point (based on the sampling location coordinates) was performed in ArcGIS using the Extract Multi Values to Points tool, which assigns the value of the central pixel of the raster layer to each point.

Results

A characteristic feature of the the pollen spectra of surface samples of the study area is the presence of pollen from Artemisia, Chenopodiaceae, and Poaceae in each sample. Additionally, Artemisia and Chenopodiaceae are overrepresented in surface samples compared to their proportion in phytocenoses (Table 2), which may be due to their high pollen productivity and the possibility of wind transfer from adjacent territories. The excessive amount of wormwood pollen has been discussed in other studies (Zhang et al. 2014, 2018). Given that geobotanical descriptions were not carried out at the locations where Tauber traps were set, such conclusions are impossible for these samples. The proportion of Artemisia pollen in surface samples from the steppe zone varies from 4.4 to 25.7%, and the proportion of Chenopodiaceae is 1.7 to 51.0%. The proportion of Poaceae pollen in the samples is underestimated compared to the share of graminoids in phytocenoses and fluctuates from 1.2 to 7.4%. In the samples of the Tauber traps installed in the steppe zone, wormwood was detected within 30.6–38.0%, goosefoots – 17.3–24.6%, and graminoids constituted approximately 11.0%. In samples from the forest-steppe zone, a significant increase in graminoids (9.0 to 29.0%) and a decrease in goosefoots (4.6 to 20.0%) were noted. Since the topsoil samples were collected only in the steppe zone, zonal comparisons were made based solely on the results obtained for the Tauber traps.

The calculated A/C ratio in topsoil samples ranges from 0.26 to 8.86 (mean 2.03±2.46), while in steppe samples collected using Tauber traps it ranges from 1.27 to 2.19 (mean 1.67±0.43), and in forest steppe it ranges from 1.70 to 4.40 (mean 3.04±1.04). It should be noted that for samples of the same type taken in the steppe and forest-steppe zones, there is a significant overlap of the intervals (Zhao et al. 2012). This A/C ratio is not always indicative for transitional (ecotone) zones such as forest-steppe, where steppe and forest elements interpenetrate (Ma et al. 2017). Therefore, the (A+C)/P ratio was additionally calculated. For the forest steppe zone, the (A+C)/P ratio, according to Tauber trap data, varies from 1.74 to 3.09 (mean 2.38±0.61); in the steppe zone, this indicator increases on average to 5.24±0.17. For topsoil samples, this ratio ranges from 4.90 to 25.80 (mean 8.01±6.35), which corresponds to the typical values for steppe regions (Table 3, Fig. 2).

Although the A/C ratio for points located in forest steppe landscapes indicates steppe landscapes (>1), the (A+C)/P ratio still indicates forest-steppe status (A+C)/P < 5 (Fowell et al. 2003, Tables 3, 5). For a more precise zonal delimitation based on the calculated ratios (Tauber traps), we used the Mann–Whitney U test, which is resistant to outliers and optimal for small sample groups. For the A/C ratio, a statistically significant difference is observed (p=0.04, U=6), but there is significant overlap in the ranges (steppe – 1.27-2.19, forest-steppe – 1.70-4.40), significantly reduces its utility for zonal delimitation. For the ratio (A + C) / P, the calculated Mann-Whitney U test (p=0.002, U=0) indicates complete separation of the sample groups (all values for forest-steppe are lower than all values for the steppe), which makes it a potentially reliable classifier for the steppe and forest-steppe zones for this type of samples located in anthropogenically modified areas. The precision of this classification is also confirmed by the results of the ROC analysis with Leave-One-Out cross-validation based on the ratios for pollen traps (Table 4).

The (A+C)/P ratio excellently differentiates all samples according to their be-longing to the steppe and forest steppe zones (AUC = 1.0, precision 100%), but the accuracy for the A/C ratio is only 70%. Low specificity (0.83) and accuracy (70%) prevent the calculated A/C ratio from distinguishing zones and significantly limit its independent use as a zonal marker. All this further demonstrates that the (A+C)/P ratio calculated for Tauber traps has a potentially high classification accuracy and can serve as a zonal indicator for this type of sample collected in anthropogenically modified by humans.

Discussion

Analysis of previously calculated data showed that the range of A/C ratios varies between regions, even when samples were collected from the same type of land-scape (Table 5). For desert landscapes, this range varies from 0.3 in the Middle East to <1 in Northern and Northwestern China. Based on this, an A/C ratio of <1 is considered the threshold value for desert landscapes. A ratio of >1 is typical for steppe landscapes. For meadows and forests, the A/C ratio is less indicative, as there is a fairly wide range with overlapping values. The value of the ratio (A+C)/P >5 indicates arid conditions with steppe vegetation, and values <5 are interpreted as humid conditions in forest steppes (Fowell et al. 2003). It should be noted that some studies show that these ratios are not universal and their use is limited by a number of factors – anthropogenic load, landscape heterogeneity, and edaphic conditions (Yang & Scuderi 2010, Zhao et al. 2012), which is confirmed by the results of our calculations.

In general, the ratios for different sample types calculated in our studies agree fairly good agreement with previously obtained data for other areas (Table 5). This is particularly evident in steppe areas, especially in topsoil samples. In most cases, our calculated A/C ratio is >1, which is consistent with values for steppe landscapes.

However, in some topsoil samples, the A/C ratio differs significantly from the majority of points. This further confirms the fact that this relationship is not universal and should be used with caution with respect to heterogeneous or anthropogenically modified landscapes. Therefore, the maximum value was obtained for point Kul-07-25 and corresponds to the characteristic values of forest areas – 8.9. It should be noted that at Kul-07-25 the Artemisia pollen content (25.7%) is a little lower than the proportion of wormwoods (Artemisia campestris L., A. frigida Willd., A. glauca Pall. ex Willd.) in this community (41%), and in neighboring points the wormwood pollen content also has close values: Kul-06-25 – 25.2%, Kul-08-25 – 20.5%, Kul-09-25 – 18.0%, Kul-10-25 – 12.5%. At the same time, wormwoods have a significant share in the composition of the Kul-06-25 (22%) and Kul-08-25 (⁓38%) communities, are found sporadically in Kul-09-25, and are absent in Kul-10-25 (Table 2). Therefore, the high Artemisia pollen content in soil samples is primarily due to the regional background (component) and the activity of wind transport from adjacent areas. Furthermore, in the latter two communities, the total projective cover reaches 100%, which may hinder the penetration of airborne pollen into the soil. On the other hand, the proportion of Chenopodiaceae pollen at point Kul-07-25 (2.9%) is 4.2 and more times lower than at the other 4 points under consideration, where it is comparatively stable at points Kul-06-25, Kul-08-25 and Kul-10-25 fluctuating within the range of 11.9 to 12.9%. Only at Kul-09-25 does the Chenopodiaceae pollen content increase to 17.8%, where the licorice-gramineous meadow serves as its unique reservoir, being located in the first strip of closed vegetation behind a sand dune in the path of the prevailing southwest winds, which carry the pollen of goosefoots from the vast salt marshes (solonchaks) of Lake Kulundinskoye. It is likely that the combination of all the factors determines the patterns of pollen deposition patterns in the soil including the anomalously high A/C ratio for the steppe zone at the point Kul-07-25. Despite the fact that this site is located on a fallow land and represents a secondary steppe, we do not have sufficient grounds to assume that the anomalously high A/C ratio for the steppe zone is related to the consequences of anthropogenic impact. The fact is that we do not have an understanding of the mechanism of this impact on the A/C ratio for the reasons discussed above. On the other hand, there is evidence that the A/C ratio should be used with a caution in areas with strong anthropogenic load (Yang & Scuderi 2010, Zhao et al. 2012).

At the point Kul-01-25 located on the boundary between the dry and droughty steppes, the A/C ratio is 0.26, which corresponds to desert areas according to other studies. At this point, the pollen content (13.5%) is typical for the 10 points of the steppe zone, although its projected cover in the community is greater than 30% (Table 2). Such an anomalously low A/C ratio for the steppe zone is due to the extremely high proportion of Chenopodiaceae pollen (51.1%). Firstly, this is the only community of the 10 steppe points where representatives of goosefoots were noted. Secondly, Atriplex verrucifera M. Bieb. is the dominant species here with a projective cover 38%, which corresponds to the Chenopodiaceae pollen content of Chenopodiaceae in the soil sample. These communities of salt marshes dominated by goosefoots are the main source of their pollen in the steppe zone and they are even more prevalent in the desert zone. Given the aforementioned lower volatility of Chenopodiaceae pollen compared to Artemisia, it can be assumed that most of the Chenopodiaceae pollen is of local origin and was deposited locally. Therefore, the anomalously low A/C ratio for the steppe zone at this point is due to the characteristics of the local community that mimic the vegetation conditions of the desert zone.

Of particular importance when analyzing pollen spectra, including calculated ratios, is the comparison with temperature and mean annual precipitation data, which are key parameters for interpreting pollen spectra. We calculated the Pearson correlation between the calculated A/C and (A+C)/P ratios on the one side, and mean annual precipitation (MAP, mm) and mean annual temperature (MAT, °С) on the other side.

The calculations were performed in two versions according to the sample types: separately for topsoil samples and traps. The correlation analysis for the topsoil samples revealed a weak positive correlation, which did not reach the statistical significance for A/C and MAT (R2=0.298, p>0.001), A/C and MAP (R2=0.039, p>0.001).

The (A+C)/P ratio also correlates quite poorly with the temperature parameters (A+C)/P and MAT (R2=0.253, p=0.138). However, the correlation between (A+C)/P and MAP reaches a moderate level (R2=0.523, p=0.018) due to outliers at the Kul-01-25 point, while all other values are grouped on the right side of the graph. Thus, insignificant correlations indicate that the calculated ratios in topsoil samples are not influenced by one specific parameter, but by a number of factors (for example, the taphonomy of pollen grains, the influence of melt water, etc.), which must be taken into account when calculating such ratios specifically in topsoil samples.

For the ratios calculated for the Tauber traps, there is a significant positive relationship with the mean annual precipitation. A/C and MAP (R2=0.804, p<0.001) (Fig. 3a); for (A+C)/P and MAP, the relationship is slightly weaker (R2=0.778, p<0.001) (Fig. 3b), but also reaches high values. No significant correlations with MAT were found: A/C and MAT (R2=0.127, p>0.001), (A+C)/P and MAT (R2=0.045, p>0.001).

The positive correlations of the calculated ratios with the mean annual precipitation once again emphasize that the features of sedimentation are important in these calculations. However, within the same landscape type, the relationships between the A/C and ( + ) / ratios and MAP may not be comparable due to various factors, such as soil salinity, plant community composition, human activity and others, which can influence the values of the ratios. Therefore, when interpreting the results, it is necessary to consider various factors (in particular, the sample type) assuming also random and inaccessible ones for study and accounting.

Conclusion

Differences in the (A+C)/P ratio in recent samples collected using Tauber traps in anthropogenically modified plain areas of Altai Krai allow us to potentially divide the territory into steppe and forest steppe zones. This ratio has a positive correlation with the mean annual precipitation, which is due to the sedimentation processes of this type of sample. However, in addition to precipitation, a variety of factors can influence these ratios and must be considered when interpreting the results: the proportion of indicator taxa in the community composition and their pollen productivity, pollen volatility and its ability to persist for a long time, and the influence of vegetation including an anthropogenically modified one on pollen redistribution and deposition.

For topsoil samples, the calculated ratios require a more detailed study with a larger number of replications from different locations.

Our data and other studies confirm that the A/C and (A+C)/P ratios can be used to detect vegetation changes and serve as qualitative or semi-quantitative indicators of moisture in the steppe and forest-steppe landscapes of Altai Krai but require uniform sample groups and a larger number of collected samples to minimize error. More research is needed on the relationship between the calculated ratios and climatic parameters, as well as the feasibility of using this ratio for different sample types and various natural zones (for example, topsoil samples in forest steppe landscapes). We believe that the pollen ratio is a very promising tool for biome paleoreconstruction based on the principle of actualism, as it unifies and formalizes pollen content data in samples, allowing us to better understand the mechanisms by which vegetation is reflected in recent and fossil pollen spectra.

Acknowledgments

The selection, palynological analysis of soil samples, and calculation of the pollen ratios of taxa were carried out using the Russian Science Foundation Grant No. 25-77-00070. Within the framework of the State Assignment of the Institute for Water and Environmental Problems of Siberian Branch of the Russian Academy of Sciences, the analysis of climatic parameters (project FUFZ-2026-0010) and structure of vegetation cover (project FUFZ-2026-0003) were completed.