https://doi.org/10.5061/dryad.k0p2ngfff

Description of the data and file structure

This file consists of supplementary tables 1, 2, 3, 4, and 5. The contents of each file, the software used and the analysis results are different. The details are described as follows：

Supplementary Table 1：Species composition and important values of lithogenic bryophytes in different elevation gradients

Supplementary Table 1 shows that 58 bryophyte species belonging to 13 families and 27 genera were recorded in the karst urban survey sample, and the species composition was different in different altitude gradients, and the important values were significantly different.

We surveyed lithophyte moss in the downtown area of Guiyang with a sample method (Nanming District, Yunyan District, Huaxi District, Wudang District, Baiyun District, and Guanshan Lake District). Geographical elevation data (DEM) for the study area was obtained from the geospatial data cloud (http://www.gscloud.cn/). The 6 central urban areas were divided into 4 altitude segments based on equal height difference intervals: N1 (989-1091m), N2 (1092-1193m), N3 (1194-1295m), N4 (1296-1398m). Using the fishing net grid tool in ArcGIS 10.2 software with a 1km interval between sample points, a total of 117 sample sites were investigated after excluding some inaccessible or non-conforming sample sites. For each sampling plot, a 10 m × 10 m large sample square was designated, which was further subdivided into five 2 m × 2 m middle sample squares. Within each middle sample square, five 10 cm × 10 cm small sample squares were established using a metal grid, following the five-point sampling technique. Bryophyte coverage and the number of specimens collected were documented for each quadrat, with the process replicated three times on each substrate type during the survey. In total, 427 bryophyte specimens were gathered and subsequently identified under a microscope in the laboratory. In the karst urban survey plots, a total of 58 bryophyte species were identified, encompassing 13 families and 27 genera.

Supplementary Table 2：Chi-square test and Jaccard index comparison of dominant species of lithophyte moss at different elevations in urban areas (N1-N4:989-1398m)

The Yates'scorrectionformula Chi-square test (χ²) was used to calculate the interspecific association of the dominant species in the community. This study employed the Jaccard index (JI) to quantify the probability and extent of association between species pairs.

Supplementary Table 3：Semi-matrix of Pearson and Spearman rank correlation test for dominant species at N1-N4 (989-1398m) elevation

Supplementary Table 3 shows that the correlation coefficients of lithophytic moss in different altitude gradients are quite different. To assess the degree of correlation between species, the Pearson correlation coefficient and the Spearman rank correlation coefficient were utilized, which are quantitative indicators of interspecific correlation. Spearman rank correlation coefficient needs to convert the abundance vector into a rank vector, and then calculate the rank vector by substituting it into the formula.

Supplementary table 4：Select environment variables for interpretation before CCA

Supplementary table 5：CCA Analysis Results

To study the relationship between species composition, environmental factors, and community stability of plant communities at different elevations, Detrended Correspondence Analysis (DCA) in Canoco 5.0 was used to sort the distribution of plant species in the sample plots. It overcame the arcuation effect based on Correspondence Analysis (CA) data of all species at various points. The results showed that the length of the ranking axis was greater than 4. Therefore, Canonical Correspondence Analysis (CCA) based on an unimodal model was used for direct ranking. For each scale, two matrices are created: one for "vegetation attributes" × "sample site" (response variable), and the other for "habitat variable" × "sample site" (explanatory variable). The ranking results objectively reflect the ecological relationship between plant communities and the environment. Through the selection and screening method mentioned above, only the factors that have a significant impact on species distribution are added to the final ranking model, and the model is tested by Monte Carlo.

Supplementary Tables 4 and 5 show that the Canonical Correspondence Analysis ranking results of 117 plots and 9 environmental factors and stability can better explain the changes in species composition. Nine environmental factors were considered in this study, comprising elevation, slope aspect, slope gradient, vegetation coverage, canopy density, light intensity, air temperature, air humidity, and rock exposure.

Code/software

Supplementary Table 1：The relative coverage and relative frequency of bryophytes are used to calculate the ecological importance value of bryophytes. The calculation formula is as follows:

  L=(MI+NI)/2                              (1)

Where L is the ecological importance value; MI is relative coverage; and NI is the relative frequency. Relative coverage = average coverage of one bryophyte/sum of average coverage of all bryophyte species, relative frequency = frequency of one bryophyte/sum of frequency of all bryophyte species.

Supplementary Table 2：The Yates'scorrectionformula Chi-square test (χ²) was used to calculate the interspecific association of the dominant species in the community, which is the qualitative index of interspecific association:

                             (2)

N is the quadrat number. When ad>bc, the interspecies pairs are positive, and when ad<bc, the interspecies pairs are negative. If 3.841<χ²<6.635, the association was significant, χ²≥6.635 was extremely significant, otherwise, it was not significant. sp. assoc() in R language spaa package was used to calculate the population and interspecies association.

This study employed the Jaccard index (JI) to quantify the probability and extent of association between species pairs:

                (3)

The JI range is [0, 1]. The larger the value is, the stronger the positive correlation is; otherwise, the closer the negative correlation is. The JI value is calculated using sp.assoc() in the R language spaa package.

Supplementary Table 3：To assess the degree of correlation between species, the Pearson correlation coefficient and the Spearman rank correlation coefficient were utilized, which are quantitative indicators of interspecific correlation. This approach allows for a comprehensive enhancement of the χ² test.

          (4)

The  is the Pearson correlation coefficient between species i and species k in the quadrat, N is the total number of quadrats,  and  are the important value of species i and species k, respectively. They form vector sums  and . The  and  are the average values of the importance of species i and i in the j-quadrat, respectively. The value of (0,1]  is positively correlated, the value of [-1, 0)  is negatively correlated, and the value of (0)  is not correlated.

Spearman rank correlation coefficient needs to convert the abundance vector into a rank vector, and then calculate the rank vector by substituting it into the formula. Spearman rank correlation coefficient is calculated as follows:

              (5)

Where,  is the Spearman rank correlation coefficient between species i and species k in the j quadrat, N is the total quadrat number,  and  is the rank of the abundance value of species i and species k in the j quadrat respectively.

Within the R language spaa package, the sp.pair() function was used to calculate the correlation coefficients, and the significance of these coefficients was tested using the corr.test() function from the psych package. Data visualization was conducted using OriginPro 9.0.