1. Introduction

Global patterns of biodiversity in animals and plants emerge from the interplay between ecological constraints—such as habitat availability and resource limitation—and evolutionary processes that generate and maintain species richness over time (Ricklefs, 2004; Wiens, 2011; Rabosky and Hurlbert, 2015). Evolutionary dynamics span from the deep-time origin of major clades and their ecological expansions to more recent events of speciation and extinction that shape present-day assemblages (Dimitrov et al., 2023). While these processes have been extensively studied in animals (e.g., Stadler, 2011; Jetz et al., 2012), mosses (Qian, 2025), and vascular plants (especially angiosperms; e.g., Niklas et al., 1983; Magallón and Castillo, 2009), analogous insights remain limited for liverworts, one of the three major lineages of bryophytes and one of the earliest lineages of the extant land plants.

Although early evolutionary radiations—such as those associated with the initial diversification of angiosperms—have long been a focus of evolutionary research, current or "tip" diversification rates, which reflect recent evolutionary dynamics, have only recently gained attention. In avian taxa, for example, diversification rates increase with elevation and in regions of low species richness, implying the presence of unfilled ecological niches (Quintero and Jetz, 2018). Similarly, recent global studies in angiosperms have revealed that diversification rates are often highest in temperate regions with lower species richness (Igea and Tanentzap, 2020; Tietje et al., 2022; Dimitrov et al., 2023). Liverworts present a contrasting pattern: recent evidence suggests that current diversification rates peak in relatively species-poor liverwort assemblages at low elevations, where certain canopy-dwelling lineages in tropical rainforests are undergoing active adaptive radiations (Maul et al., 2025). In contrast, mid-elevation, where liverwort species richness peaks, only have intermediate levels of diversification, whereas at the highest elevation, assemblages are species poor and have low diversification rates. Collectively, these findings challenge the assumption that areas of high species richness necessarily harbor lineages with high contemporary diversification rates, suggesting that current biodiversity patterns may reflect the legacy of deep-time diversification (Stephens et al., 2025), and may continue to change as species accumulate under variable ecological and evolutionary pressures.

Liverworts comprise approximately 7500 described extant species worldwide (Laenen et al., 2018) and are globally distributed, from tropical rainforests to polar tundras. Their species richness peaks in moist tropical environments, particularly at mid-elevations, where high humidity and canopy microclimates provide ideal conditions for epiphytic growth (Lombo-Sanchez et al., 2024; Wang et al., 2025). In contrast, phylogenetic diversity and mean genus age—both representing the breadth of evolutionary history—tend to peak at high elevations and in temperate forests, reflecting an origin of liverworts under cool climatic conditions (Laenen et al., 2014; Qian and Kessler, 2024; Qian et al., 2024d, 2025b; Maul et al., 2025). These studies have also revealed that temperature-related variables often exert stronger effects on phylogenetic diversity than precipitation-related variables, and climatic extremes play a larger role than seasonality. In contrast to these studies on the deep evolutionary history of liverworts, our understanding of tip diversification rates in liverworts remains limited, being limited to a single study analyzing diversification rates along a set of elevational gradients (Maul et al., 2025). The global patterns of tip diversification rates in liverworts thus remains unknown, as is whether areas of high liverwort richness correspond to those of high diversification rates, or how these rates relate to environmental variables across geographic scales.

Several methodological approaches are available to estimate tip diversification rates (Morlon, 2014). Among them, the method-of-moments estimator (Magallón and Sanderson, 2001), also known as the MS approach (Meyer and Wiens, 2018), provides a tractable way to estimate net diversification by dividing lineage species richness by clade age (Stanley, 1979; Magallón and Sanderson, 2001). This method does not require full phylogenetic trees and has been widely applied to infer diversification patterns in various taxa, including angiosperms (Eriksson and Bremer, 1992; Magallón and Sanderson, 2001; Boucher et al., 2020), birds (Cooney et al., 2016), amphibians (Adams et al., 2009), and fishes (Tedesco et al., 2017), as well as across broader taxonomic scales (Wiens, 2015a, 2015b; Scholl and Wiens, 2016). This is also the method applied by Maul et al. (2025) in their study of elevational patterns of liverwort diversification rates. This approach is particularly useful for studies on diversification of those groups of organisms, such as liverworts, for which a species-level phylogeny is not available. While alternative frameworks such as Bayesian analysis of macroevolutionary mixtures (BAMM; Rabosky, 2014) have also been employed, concerns over methodological robustness and reliability have been raised (Moore et al., 2016), with recent assessments favoring the MS approach for its consistency and performance (Meyer and Wiens, 2018; Stephens et al., 2025).

In this study, we apply the MS approach to examine global patterns of tip diversification rates (i.e., species-level diversification rates within genera) in liverworts and evaluate geographic patterns and climatic correlates of these rates. Specifically, we address the following questions: (1) How do diversification rates vary across climatic gradients and among biomes? (2) Are current climatic conditions more influential in shaping diversification patterns than historical climate change, and do temperature-related or precipitation-related variables, as well as climatic extremes versus seasonality, better predict diversification? (3) Finally, do regions with higher liverwort species richness contain lineages with elevated diversification rates?

2. Materials and methods 2.1. Diversification rate estimation

Liverworts have an age of about 500 million years (Laenen et al., 2014; Harris et al., 2022). Our preliminary observation based on the information of genus stem ages available in Laenen et al. (2014) (Fig. S1) and species richness per genus available in Brinda and Atwood (2023) showed that 47% of liverwort genera had a stem age of < 30 million years with an average of 17.8 million years per genus and these genera included about 41% of the global species richness of liverworts. Furthermore, many species in those genera older than 30 million years are younger than 20 million years (e.g., Villarreal et al., 2016). Thus, the majority of extant liverwort species are relatively young with respect to the age of the whole liverwort lineage. As a result, the method used to quantify diversification rates of liverworts should be able to account for diversification at tips of a phylogenetic tree. Because most genera of liverworts lack well resolved phylogenies for their species, phylogeny-based estimators of diversification rates, such as BAMM (Rabosky, 2014), are not appropriate to estimate tip diversification rates for liverworts.

Among the diversification rate estimators that do not depend on phylogenies, the method-of-moments estimator (Magallón and Sanderson, 2001) has been a commonly used approach (e.g., Eriksson and Bremer, 1992; Hughes and Eastwood, 2006; Rabosky and Matute, 2013, Alfaro et al., 2007; Wiens, 2015a, b; Scholl and Wiens, 2016). This approach follows from the idea that a lineage's net diversification rate can be calculated by dividing the number of extant species in a lineage by its age (Stanley, 1979; Magallón and Sanderson, 2001; Meyer and Wiens, 2018). We estimated net diversification rate (r) of each genus, using the following formula: r = ln[n(1 − ε) + ε]/t, where n is the number of extant species in a lineage (i.e., a genus in our case), t is the age of the lineage, and ε is the relative extinction rate suggested to vary from 0 to 0.9 (Magallón and Sanderson, 2001). The stem age of each genus was obtained from Laenen et al.'s (2014) genus-level phylogeny for liverworts with the lower and upper bounds of the lognormal distribution associated to each fossil being set so as to encompass the timespan of the geological era attributed to the fossil (Laenen et al., 2014), and the number of species in each genus was determined based on Brinda and Atwood (2023). Laenen et al. (2014) provided stem ages for 84 % of the global liverwort genera, which included about 91% of the liverwort species in the world. Genera that were absent from Laenen et al. (2014) were not considered in the present study. Following previous studies (e.g., Stephens et al., 2025; Wu and Wiens, 2022; Qian, 2025), we used three values of ε (i.e., 0.0 for no extinction, 0.5 for intermediate extinction, 0.9 for high extinction; Gómez-Rodríguez et al., 2015) to estimate diversification rates for each genus; we found that the three sets of diversification rates estimated using the three values of ε were nearly perfectly correlated with one another (Spearman's ρ ranging from 0.981 to 0.998), suggesting that using different values of ε would have little impact on the results of subsequent analyses on geographic patterns and climatic correlates of diversification rates. Following previous studies (e.g., Davies et al., 2004; Boucher et al., 2020; Maul et al., 2025; Qian, 2025; Qian et al., 2025a), we used the set of estimated diversification rates based on ε being set to zero in subsequent analyses. Using the same approach of estimating diversification rates among different studies on different lineages of bryophytes (e.g., Maul et al., 2025; Qian, 2025; this study) allows direct comparison of the results of the studies.

2.2. Regional assemblages of liverworts

Previous studies on global liverworts have used species lists in 390 geographic units (primarily countries, provinces or states) worldwide (Qian et al., 2024d; Wang et al., 2025), which were derived from the data set compiled by Collart et al. (2021). The present study also used liverwort species lists in the 390 geographic units. We used the Bryophyte Nomenclator database compiled by Brinda and Atwood (2023) to standardize the liverwort names. After the nomenclature standardization, there were 5793 liverwort species in the 390 geographic units, and 94.4% of these species had data for ages of their respective genera.

For each geographic unit, each species was assigned the diversification rate of its genus, as in previous studies (e.g., Maul et al., 2025; Stephens et al., 2025; Qian, 2025). The mean diversification rate for each geographic unit was the arithmetic mean diversification rate (MDR) across all species in the geographic unit, as in Wu and Wiens (2022), Maul et al. (2025), and Stephens et al. (2025). It should be noted that a low or high MDR in a region or environmental condition reflects that the species assemblage of the region or environmental condition is composed of species from clades with low or high diversification rates, respectively; they should not be taken to indicate that the region or environmental condition had a low or high diversification rate (Stephens et al., 2025).

In addition to determining the number of species (i.e., species richness) in each geographic unit, we also determined area-corrected species richness (hereafter, species density), by dividing the number of species in each geographic unit by the log₁₀-transformed area (in square kilometer) of the geographic unit (Guo et al., 2021; Qian et al., 2024b).

To explore patterns of variation in MDR among different biomes, we assigned each of the 390 geographic units to one of Whittaker's biomes, which were defined based on mean annual temperature and annual precipitation (e.g., Whittaker, 1975). For the 11 geographic units that were located outside Whittaker's biome framework, we assigned them to their respective closest biomes. As a result, the number of geographic units assigned to each of Whittaker's biomes is as follows: tropical rain forest (21 geographic units), tropical seasonal forest/savanna (85), subtropical desert (38), temperate rain forest (3), temperate deciduous forest (107), woodland/shrubland (69), temperate grassland/desert (11), boreal forest (taiga) (38), and tundra (18). We then calculated the mean value of MDR for each biome.

2.3. Climate data

Current and historical climates have been considered as major factors affecting plant speciation and distributions (Cai et al., 2023; Qian et al., 2024b). We related the mean diversification rates of liverworts in the 390 geographic units to climatic variables reflecting current climatic conditions and historical climate change during the Quaternary. Previous studies, including those on liverworts (e.g., Qian and Kessler, 2024; Qian et al., 2024d; Qian et al., 2025b; Wang et al., 2025), showed that the following current climatic variables are among main climatic variables shaping plant distributions: mean annual temperature (T_mean), minimum temperature of the coldest month (T_min), temperature seasonality (T_seas), annual precipitation (P_mean), precipitation during the driest month (P_min), and precipitation seasonality (P_seas). Accordingly, we used them to characterize current climatic conditions. The historical climate change variables used in this study included the differences in mean annual temperature and annual precipitation between the Last Glacial Maximum and the present, which are temperature anomaly (T_anom) and precipitation anomaly (P_anom), respectively. T_mean, T_min and T_seas were considered temperature-related variables of current climate; P_mean, P_min and P_seas were considered precipitation-related variables; T_min and P_min were considered climate extreme variables; T_seas and P_seas were considered climate seasonality variables in current climate conditions. Data for the climatic variables for each geographic unit were obtained from the CHELSA climate database (v.2.1) at the 30-arcsecond resolution (https://chelsa-climate.org/bioclim/). For each climatic variable, we used the average value of all data points at the 30-arcsecond resolution within each geographic unit and used the average values in data analyses.

2.4. Data analysis

We used Pagel's λ (Pagel, 1999) and Blomberg's K (Blomberg et al., 2003) to assess whether there is a significant phylogenetic signal in diversification rates of liverworts, using the genus-level phylogeny derived from Laenen et al. (2014) and the function phylosig in the package phytools (Kembel et al., 2010). A value of 0 indicates no phylogenetic signal in diversification rate across the phylogeny, whereas a value of 1 indicates a strong phylogenetic signal, which matches expectations under the Brownian motion model. A strong phylogenetic signal would indicate that closely related genera would have similar diversification rates.

We assessed the relationship of MDR in geographic units with latitude and each of the climatic variables using Spearman's rank correlation (ρ). Following previous studies (e.g., Qian et al., 2025a), we used ordinary least squares model to conduct a variation partitioning analysis (Legendre and Legendre, 2012) to determine relative effects of current climatic variables and historical climate change variables on MDR, which partitioned the amount of the explained variation in MDR into three portions: variation explained uniquely by current climatic variables, variation explained uniquely by historical climate change variables, and variation explained jointly by the two sets of variables. Similarly, we conducted a variation partitioning analysis to determine whether temperature-related variables or precipitation-related variables in current climates have a stronger influence on MDR, and another variation partitioning analysis to determine whether climate extreme variables or climate seasonality variables have a stronger influence on MDR.

Previous studies on plant diversity (e.g., Weigand et al., 2020; Qian et al., 2023, 2024a) have found marked differences between continental regions. We divided the globe into three longitudinal segments, namely the New World, the western Old World (west of 75°E), and the eastern Old World (east of 75°E), as in Qian et al. (2024a) and Wang et al. (2025). In addition to conducting analyses including all the 390 regional liverwort assemblages across the world, we also conducted Spearman's rank correlation and variation partitioning analyses for each of the three longitudinal segments. We used the package SYSTAT (Wilkinson et al., 1992) for statistical analyses.

A frequently discussed hypothesis is that temperature and precipitation drive the variations of both diversification rate and species richness, and diversification rate drives the variation of species richness (e.g., Rohde, 1992; Allen and Gillooly, 2006). We used structural equation modeling (SEM) approach to assess the direct and indirect effects of temperature, precipitation, and mean diversification rate on species density (i.e., area-corrected species richness). In the framework of the SEM, mean annual temperature and annual precipitation were exogenous variables, and mean diversification rate was endogenous variable. With the SEM, we tested whether diversification rate directly influences species richness when the influences of major climatic factors on both diversification rate and species richness are accounted for. We used the R package 'lavaan' (https://cran.r-project.org/web/packages/lavaan) for our SEM analyses.

3. Results

There was weak, but statistically significant, phylogenetic signal in diversification rates of liverworts (Pagel's λ = 0.161, Blomberg's K = 0.138, P < 0.05). Even though phylogenetic signal in diversification rates was weak, the tendency of closely related genera to have similar rates of diversification can be clearly seen in Fig. 1. For example, the vast majority of the genera located in the upper-right quarter of the phylogenetic tree show dots in the two cold colors, indicating low diversification rates, whereas the vast majority of the genera located in the upper-left quarter of the phylogenetic tree have dots in the two warm colors, indicating high diversification rates.

Geographic units with high MDR were generally located in tropical areas in Africa, Asia, and Americas (Fig. 2). As a result, MDR was negatively correlated with latitude (Spearman's rank correlation ρ = −0.400; Table 1). This result qualitatively held true when the three longitudinal segments were considered separately (ρ = −0.623, −0.469, and −0.268 for the New World, eastern Old World, and western Old World, respectively).

MDR tended to decrease along the gradient from tropical rain forest (biome 1) through temperate rain forest (biome 4) to boreal forest (biome 8) but then increased toward tundra (biome 9) (Fig. 3). Within warm climate conditions, MDR decreased monotonically from 0.170 to 0.151 along the gradient from tropical rain forest (biome 1) through tropical seasonal forest/savanna (biome 2) to subtropical desert (biome 3) (Fig. 3).

When all the 390 geographic units in the world were considered as a whole, MDR was positively associated with mean annual temperature, annual precipitation, and precipitation seasonality, and negatively associated with temperature seasonality and temperature and precipitation anomalies (Table 1). When geographic units in the three longitudinal segments were analyzed separately, these relationships held true except for that of precipitation anomaly in the western Old World (Table 1).

Of the variation in MDR explained by a combination of current climatic variables and historical (Quaternary) climate change variables, the vast majority of the variation was explained independently by the current climatic variables; it was over 30 times greater than that explained independently by historical climate change variables (Fig. 4a). Of the variation in MDR explained by the six current climatic variables, the three temperature-related variables explained more variation in MDR than did the three precipitation-related variables (5.0% versus 4.0%; Fig. 4b). Of the variation in MDR explained by the combination of the two climate extreme variables and the two climate seasonality variables, the vast majority of the variation was jointly explained by the two sets of the variables; for the variation independently explained by either set of the variables, the climate seasonality variables independently explained more variation, compared with the climate extreme variables (2.4% versus 0.9%; Fig. 4c).

When liverwort assemblages in the three longitudinal segments were analyzed separately, the current climatic variables explained more variation in MDR than did the historical climate change variables in all the three longitudinal segments (Fig. 5). Temperature-related variables explained more variation in MDR in the New World but less variation in the eastern Old World, compared with the precipitation-related variables, with the two sets of the variables explaining approximately equal amount of the variation in the western Old World (Fig. 5). Nearly all the variation in MDR explained by the four variables representing climate extreme and seasonality was explained jointly by the two sets of the variables in the New World and the eastern Old World, but climate seasonality variables explained more variation than did climate extreme variables in the western Old World (Fig. 5).

MDR was positively correlated with species density, regardless of whether all liverwort assemblages in the world were considered as a whole (Spearman's correlation coefficient ρ = 0.298) or liverwort assemblages in each of the three longitudinal segments were considered separately (Fig. 6). When the relationships between MDR and species density were assessed after accounting for the effects of temperature and precipitation on both of them in structural equation models (SEMs), the positive relationships between MDR and species density held true both for the world as a whole and for each of the three longitudinal segments (Fig. 7).

4. Discussion

In this first study exploring the global variation of tip diversification rates in liverworts, measured as mean diversification rate (MDR), we obtained the following main results. First, MDR peaked at tropical latitudes and in humid and hot environments, with MDR being influenced more strongly by current climate, temperature-related variables, and climatic seasonality than by historical climate change, precipitation-related variables, and climatic extremes, respectively. Second, we found a positive relationship between MDR and liverwort species density, with the latter being more strongly directly influenced by climate than by MDR. Third, the above-described patterns differed little among longitudinal segments.

These results differ in several important aspects from the only previous study of tip diversification rates in liverworts along elevational gradients reported by Maul et al. (2025). Thus, although they also found the highest MDR in tropical conditions, they found that the richest liverwort assemblages at mid-elevations only had intermediate levels of MDR, contrasting with the positive MDR−species density relationship found here. These differences likely reflect the different spatial scales of the studies, with our study looking at large geographic units that often encompass extensive elevational gradients, whereas Maul et al. (2025) specifically assessed MDR at different elevations. The two studies thus complement each other. The negative relationships of MDR with latitude and the positive relationships of MDR with mean annual temperature and annual precipitation found for liverworts in the present study are consistent with those for mosses at a global scale (Qian, 2025). This suggests that both groups of bryophytes have similar current diversification dynamics.

Liverwort species richness is well known to peak at tropical latitudes, especially in humid mountain forests (Wang et al., 2017, 2025). We here find that tropical biomes are also the present-day centers of the formation of new species, resulting in a positive relationship between MDR and species density. This is consistent with the finding of Qian (2025) for mosses but contrasts with the results from three previous studies on angiosperms showing that tip diversification rates are highest in temperate regions (Igea and Tanentzap, 2020; Tietje et al., 2022; Dimitrov et al., 2023). A direct comparison of the studies on angiosperms is unfortunately not feasible due to the differing methodologies employed. Tietje et al. (2022) and Dimitrov et al. (2023) applied mean root distance (MRD) and BAMM, respectively—both methods incorporate branching information from the root to the tips of a phylogeny, which may introduce biases (Moore et al., 2016; Meyer et al., 2018; Meyer and Wiens, 2018). In contrast, our use of MDR does not consider relationships among genera and has been specifically recommended for analyses of this nature (Moore et al., 2016; Meyer et al., 2018; Meyer and Wiens, 2018). In any case, liverworts differ from many other major plant lineages such as angiosperms and ferns (Qian et al., 2023, 2024c), in that their initial evolutionary diversification took place under temperate, rather than tropical, climatic conditions (Maul et al., 2025; Qian et al., 2025b). This has led to fundamentally different evolutionary dynamics in this group, and it may thus not be surprising that we find a different relationship between MDR and climatic conditions in liverworts as compared to angiosperms. From a more abstract perspective, however, these studies coincide: in both cases, MDR is low under the ancestral climatic conditions of the lineages (tropical in angiosperms, temperate in liverworts) and peaks under climatic conditions to which the groups have only relatively recently become adapted to (temperate in angiosperms, tropical in liverworts). This suggests that the relatively newly colonized biomes have abundant empty niche space, favoring speciation (Igea and Tanentzap, 2020; Tietje et al., 2022; Dimitrov et al., 2023; Maul et al., 2025).

Our structural equation models show that the relationship between MDR and species richness in liverworts is not simply a result of spurious covariation due to a direct influence of climate on both MDR and species richness. We found a direct influence of MDR on liverwort richness even though the direct effect of annual precipitation was stronger in all cases. This suggests that patterns of species richness are most strongly determined by climatic conditions, in accordance with previous studies (e.g., Qian and Kessler, 2024; Wang et al., 2025). This likely reflects the physiological adaptations and limitations of liverworts, which are poikilohydric and can therefore not control the hydration level of their tissue (Qian et al., 2025b; Wang et al., 2025), making them highly dependent on external climatic and microclimatic conditions (Dai et al., 2025; Qian et al., 2025b; Maul et al., 2025). These physiological effects are not only influenced by precipitation, which is a proxy for water input, but also by temperatures, which influence evapotranspiration and hence also the hydration state of the liverworts.

Looking in more detail, our variation partitioning approach shows that while the combined effect of temperature and precipitation has the strongest correlation with MDR, the influence of temperature alone is stronger than that of precipitation alone. This is in accordance with previous results on phylogenetic diversity of liverworts (Qian et al., 2025b), but contrasts with findings on species richness in liverworts (Wang et al., 2025). Again, this emphasizes that patterns of species richness and of evolutionary dynamics (either phylogenetic diversity or diversification rates) follow different dynamics and are differently influenced by environmental factors. More generally, this shows the value of separately studying different aspects of biodiversity, considering evolutionary patterns in addition to species richness (Cadotte and Davies, 2016).

In our further examination of climatic variables, we found that present-day climate exerts a consistently stronger influence on MDR than historical climatic variability. This is expected, as tip diversification rates reflect evolutionary dynamics during the recent tens of million years, and it aligns with observed patterns in phylogenetic diversity (Qian et al., 2024c). We also identified a strong combined influence of climatic seasonality and climatic extremes on MDR. Notably, the individual effect of climatic seasonality, though being modest, is stronger than that of climatic extremes. This contrasts with findings in phylogenetic diversity, where climatic extremes typically exert a much greater impact than seasonality—observed in both liverworts (Qian et al., 2025b) and other plant groups (ferns: Qian et al., 2023, 2024b; angiosperms: Qian et al., 2024c). Such results have been interpreted to suggest that the persistence of evolutionary lineages is more constrained by rare extreme events, which are difficult for species to adapt to, than by the more predictable patterns of seasonality, which allow for physiological adaptation. However, our findings indicate that this pattern may not hold for speciation. We propose that this could be because newly formed species of closely related lineages tend to be physiologically very similar, whereas more distantly related lineages at higher taxonomic levels (e.g., genera, families, orders), whose diversity is captured by phylogenetic diversity metrics, often differ markedly in physiological traits.

Finally, we found little difference in the geographic patterns and climatic correlates of MDR between the continental land masses, separated here as the New World, western Old World (Africa and Eurasia west of 75°E) and eastern Old World (Eurasia east of 75°E and Australasia). This contrasts with many previous studies finding strong differences in species richness and phylogenetic diversity in plant assemblages between continents (e.g., Weigand et al., 2020, Qian et al., 2023, 2024a, c). Since we calculated MDR on a global per-genus base, we applied the same genus-specific diversification rate to all continents on which the genus occurs, as in many previous studies (e.g., Maul et al., 2025; Stephens et al., 2025), potentially masking within-genus differences between continents.

As stated above, because a phylogeny that includes most species of liverworts worldwide is lacking, we are not able to use a phylogeny-based estimator to estimate diversification rates of liverworts. Although previous studies (Meyer et al., 2018; Meyer and Wiens, 2018) have shown that the method-of-moments estimator used in our study outperforms some phylogeny-based estimators of diversification rates (e.g., BAMM), the robustness of the results reported in our study should be tested when a well-resolved species-level phylogeny of liverworts is available. Furthermore, this study used a spatial framework comprising 390 geographic units, many of which encompass large, environmentally heterogeneous regions, and some of which each contain multiple biomes. Treating these diverse regions as single spatial units may introduce bias into the analysis. While we acknowledge the limitations imposed by this coarse spatial resolution, the global scope of our study ensures that between-unit variability remains substantially higher than within-unit variability, so that the resulting patterns are robust. Further, because these geographic units, or similar geographic units, have been used in many macroecological studies of plants, including studies on diversification rates (e.g., Qian et al., 2025a for ferns), using the same, or similar, set of geographic units in different studies would make the results of different studies directly comparable.

5. Conclusions

Our study shows that tip diversification rates tell a complementary story of the evolution of liverwort diversity to that recovered by studying phylogenetic diversity and species richness. Most importantly, we found that tropical regions of high liverwort diversity also have high current diversification rates. However, a study separating different elevational belts has shown that this only applies to the tropical lowlands, rather than to the montane cloud forests which harbor the highest liverwort species richness (Maul et al., 2025). This reflects that to fully understand patterns of diversity and the underlying evolutionary dynamics, we must consider different spatial and temporal scales (Cadotte and Davies, 2016). Also, we find that patterns in liverworts differ from those found in angiosperms and ferns, suggesting that different evolutionary pathways exist in different groups of organisms. A direct comparison of tip diversification rates between different plant groups differing fundamentally in physiological and life history traits, using a consistent method and spatial scale, would be desirable to better understand the differences and commonalities between these plant groups.

Acknowledgements

We thank anonymous reviewers for their helpful comments.

Data accessibility statement

Liverwort species distribution data that were used in the present study are available at the Figshare Repository at https://doi.org/10.6084/m9.figshare.22587199. Climate data that were used in the present study are available at the CHELSA (https://chelsa-climate.org/bioclim/).

CRediT authorship contribution statement

Hong Qian: Conceptualization, Data curation, Writing − original draft, Writing − review & editing, Investigation, Formal analysis. Michael Kessler: Writing − original draft, Writing − review & editing. Shenhua Qian: Data curation, Formal analysis.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.12.010.