1Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

www.nature.com/scientificreports

The mossy north: an inverse latitudinal diversity gradient in European bryophytesRubén G. Mateo1,2,3, Olivier Broennimann1, Signe Normand4, Blaise Petitpierre1, Miguel B. Araújo5,6,7, Jens-C. Svenning4, Andrés Baselga8, Federico Fernández-González3, Virgilio Gómez-Rubio9, Jesús Muñoz10, Guillermo M. Suarez11, Miska Luoto12, Antoine Guisan1,* & Alain Vanderpoorten2,*

It remains hotly debated whether latitudinal diversity gradients are common across taxonomic groups and whether a single mechanism can explain such gradients. Investigating species richness (SR) patterns of European land plants, we determine whether SR increases with decreasing latitude, as predicted by theory, and whether the assembly mechanisms differ among taxonomic groups. SR increases towards the south in spermatophytes, but towards the north in ferns and bryophytes. SR patterns in spermatophytes are consistent with their patterns of beta diversity, with high levels of nestedness and turnover in the north and in the south, respectively, indicating species exclusion towards the north and increased opportunities for speciation in the south. Liverworts exhibit the highest levels of nestedness, suggesting that they represent the most sensitive group to the impact of past climate change. Nevertheless, although the extent of liverwort species turnover in the south is substantially and significantly lower than in spermatophytes, liverworts share with the latter a higher nestedness in the north and a higher turn-over in the south, in contrast to mosses and ferns. The extent to which the similarity in the patterns displayed by spermatophytes and liverworts reflects a similar assembly mechanism remains, however, to be demonstrated.

The existence of a latitudinal diversity gradient (LDG) peaking near the equator and decreasing towards the poles has been a persistent feature during the history of life on earth1 (but see ref. 2). This gradient has been quoted as one of the few laws in ecology3, and it demonstrates remarkable consistency across geographic areas, scales, hab-itats, and taxonomic groups4–7. Mounting evidence suggests that this convergence of distribution patterns across taxonomic groups is due to environmental forcing8.

On the one hand, macroclimate (primarily energy and water) is postulated to control species richness through the ecological sorting of regional and global species pools according to species climatic tolerances or by affecting rates of speciation9. In particular, dry or cold environments are specifically challenging for plants because adap-tations must evolve to enable the tolerance or avoidance of extremely low water potentials10. In lineages that have successfully adapted to high-latitude environments, increased seasonal variability at higher latitudes is assumed to have resulted in broader thermal tolerances and consequently larger ranges: i.e., Rapoport’s rule11.

1Department of Ecology and Evolution, University of Lausanne, Biophore, CH-1015, Lausanne, Switzerland. 2Institute of Botany, University of Liège, B-4000, Liège, Belgium. 3Institute of Environmental Sciences, University of Castilla-La Mancha, ES-45071, Toledo, Spain. 4Section for Ecoinformatics & Biodiversity, Department of Bioscience, Aarhus University, Ny Munkegade 114, DK-8000, Aarhus C, Denmark. 5Department of Biogeography and Global Change, National Museum of Natural Sciences (CSIC), ES-28006, Madrid, Spain. 6InBIO/CIBIO, University of Évora, Largo dos Colegiais, 7000, Évora, Portugal. 7Center for Macroecology, Evolution and Climate, Natural History Museum of Denmark, University of Copenhagen, Universitetsparken 15, DK-2100, Copenhagen, Denmark. 8Department of Zoology and Physical Anthropology, University of Santiago de Compostela, ES-15782, Santiago de Compostela, Spain. 9Department of Mathematics, University of Castilla-La Mancha, ES-02071, Albacete, Spain. 10Real Jardi n Bota nico (CSIC), ES-28014, Madrid, Spain. 11CONICET, Facultad de Ciencias Naturales e I.M.L, UNT, Universidad Nacional de Tucamán, 4000, Tucumán, Argentina. 12Department of Geosciences and Geography, University of Helsinki, 00014, Helsinki, Finland. *These authors jointly supervised this work. Correspondence and requests for materials should be addressed to R.G.M. (email: rubeng.mateo@gmail.com)

Received: 26 November 2015

accepted: 19 April 2016

Published: 06 May 2016

OPEN

www.nature.com/scientificreports/

2Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

In addition, habitat heterogeneity promotes species richness because diverse habitats allow greater niche sep-aration and therefore promote species coexistence12–14. Habitat heterogeneity, when represented by topographic heterogeneity, is also related to historical factors because mountains facilitate the long-term survival of species by allowing climate tracking via short-distance altitudinal migration15.

Investigating the factors determining the distribution of 1016 European plant species, Normand et al.16 con-firmed the key role of extant climate and also demonstrated the crucial role of historical factors. Decreased extinc-tion and increased speciation in climatically stable areas is expected to contribute to shaping extant distribution patterns17,18. For example, the increase in dung beetle species richness towards lower latitudes in Europe results from both the orderly exclusion of species towards the north, resulting in a higher nestedness of northern assem-blages compared with southern assemblages and from the high species turnover caused by steep ecological gradi-ents in southern areas19. In addition to climate stability, the time required for colonisation can also contribute to the observed latitudinal species richness gradient13,15. In particular, latitudinal species richness gradients might result from the incomplete post-glacial recolonisation of high-latitude regions. In this context, less mobile organ-isms are expected to exhibit steeper latitudinal species richness gradients than vagile organisms20.

There are, however, some notable exceptions to these patterns, as groups originating during warmer periods of earth’s history display a steeper latitudinal gradient, whereas groups originating during colder periods display a shallower diversity gradient due to a weak affinity or no affinity for lower latitudes21. For example, grasses are among the relatively few higher-order lineages that exhibit a shallow, atypical latitudinal gradient due to the climatic specialisation of particular lineages to cold and arid environments22. Similarly, although approximately 50% of extant gymnosperm species occur primarily between the tropics, the diversity of gymnosperms decreases at equatorial latitudes23.

A latitudinal species richness gradient was observed in spermatophytes24 and in ferns25 wherein, however, regional species richness patterns do not always correlate with latitude26,27, and even challenged in bryophytes, a group with approximately 20,000 species of mosses, liverworts, and hornworts, which represents the second most speciose lineage of land plants after the angiosperms. Similar levels of bryophyte species richness have been repeatedly reported from tropical and extra-tropical areas28–31, and Rozzi et al.32 even documented an inverted bryophyte species richness gradient that increased towards the pole in southern south America. Such weak, if not inverted, gradients of species richness towards high latitudes appear consistent with three major features of bryophyte biology. First, bryophytes typically fail to radiate in contrasted environments33, reducing their opportu-nities to diversify along the steep ecological gradients found at low latitudes. Second, bryophyte species, including tropical species, are inherently better adapted to cold conditions than are angiosperms34. They are universally able to grow at low temperatures, showing a growth reduction of less than 50% at 5 °C compared with growth at their optimal temperatures35. Simultaneously, because they are poikilohydric, they are much less well equipped to face drought and warm conditions. Bryophytes therefore exhibit lower temperature optima than higher plants35; all temperate and boreal species investigated by Furness & Grime35 died when kept continuously at 35 °C, and most shoots died at > 30 °C. Third, bryophyte species display larger geographic ranges than angiosperm species and display a high dispersal capacity, resulting in a much lower global rate of species turn-over than angiosperms36, suggesting that rapid post-glacial recolonisation prevents the formation of the richness gradients that result from limited post-glacial dispersal processes, as has been observed in vascular plants15.

Here, we examined the spatial variation in species richness and beta-diversity (disentangling its two compo-nents: species replacement or turnover, and species loss or nestedness) in European bryophytes and compared them with those observed in ferns and spermatophytes. Specifically, we tested three hypotheses: 1) as a result of differences in temperature optima and drought tolerance between bryophytes and spermatophytes, bryophyte species richness should decrease towards lower latitudes, whereas spermatophytes should display the opposite pattern; 2) given the failure of bryophytes to radiate in ecologically contrasted areas with a long history of climate stability such as the Mediterranean, and given their large range sizes, the spatial turnover of bryophyte richness across latitudinal gradients should be less pronounced than in spermatophytes; 3) in spermatophytes, nestedness should increase in the north as a result of limitations in the ability to evolve adaptations to cold conditions and the consequent ordered loss of species towards the north, whereas species turn-over should increase in the south. In bryophytes, we expect to observe a reverse pattern as a result of the inherent cold tolerance in the north and the filtering-out of species towards the south.

ResultsConsistent with our first hypothesis, European ferns, mosses, liverworts and spermatophytes exhibited contrast-ing patterns of species richness around a central axis running approximately through the chains of the Pyrenees and the Alps, close to the 46th parallel (Figs 1 and 2). Ferns, mosses and liverworts exhibited a similar impov-erishment of species richness towards the Mediterranean and a peak of richness at mid-latitudes that reaches Scandinavia in ferns and mosses, but not in liverworts (Figs 1 and 2). Thus, regions of significant spatial associ-ation in the distribution of the four taxonomic groups are distributed along this central axis, whereas negative associations are found southwards and northwards of it (Fig. 3).

These differences in species richness patterns among taxonomic groups were paralleled by substantial differ-ences in their patterns of species turnover (β SIM) and nestedness-resultant dissimilarity (β SNE) in northern (> 46° N) and southern areas (< 46° N) (Fig. 4, Tables 1 and 2). In spermatophytes, β SIM was significantly higher in the south than in the north (p < 0.01). Spermatophytes thus exhibited significantly and substantially higher β SIM than ferns, mosses and liverworts in the south (p = 0.001), but not in the north. Liverworts exhibited a similar trend as spermatophytes, with β SIM in the south being significantly higher than in the north (p = 0.026). Mosses and ferns exhibited the reverse pattern, with significantly higher (p = 0.036 in ferns) or similar (p = 0.37 in mosses) β SIM in the north than in the south. Thus, there was no difference of β SIM between mosses and ferns in the north

www.nature.com/scientificreports/

3Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

Figure 1. Potential species richness of spermatophytes (a), ferns (b), mosses (c) and liverworts (d) across Europe. Maps based on ensemble stacked species distribution models (S-SDMs) of 1359, 79, 810 and 224 species of spermatophytes, ferns, mosses and liverworts, respectively. Maps generated by R.G. Mateo using the ArcMap extension in ArcGIS 10.2 (ESRI Inc., Redlands, CA, USA, http://www.esri.com).

Figure 2. Predicted numbers of species of spermatophytes (green), ferns (black), mosses (red) and liverworts (blue) in 100 km latitudinal bands across Europe. Dashed lines indicate crude SR values predicted by S-SDMs, solid lines correspond to SR values normalized according to species-area relationships (see Supplementary Methods 5).

www.nature.com/scientificreports/

4Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

(p = 0.429) or in the south (p = 0.111), whereas both groups exhibited significantly higher β SIM than liverworts in the north (p < 0.001 in both mosses and ferns), but not in the south (p = 0.067 in mosses and 0.266 in ferns).

In turn, β SNE was significantly higher in the north than in the south (p < 0.001) in spermatophytes. Again, although β SNE was significantly higher for ferns, mosses and liverworts than for spermatophytes in the south (p < 0.001), whereas the opposite trend was observed in the north (p < 0.001 in mosses and liverworts and p = 0.011 in ferns), liverworts exhibited a similar trend to spermatophytes as β SNE in the north was significantly higher than in the south (0.019). The opposite trend was observed in mosses and ferns, wherein β SNE was sig-nificantly lower in the north than in the south (p < 0.001) in the latter, and similar in the north and the south (p = 0.220) in the former. Thus, β SNE was significantly higher for liverworts than for mosses and ferns in the north (p < 0.001).

DiscussionAlthough some previous studies report shallow increases in species richness towards low latitudes21 and even of inverted patterns in specific taxonomic groups37, we report here a marked increase in species richness towards high latitudes for an entire phylum of land plants at the continental scale. The investigated area in the present study did not encompass the tropics, so that the hypothesis of a global LDG in bryophytes cannot be rejected (but see ref. 29). Nonetheless, our results contrast with the suggestion of a consistent LDG pattern across geographic areas, scales, habitats and taxonomic groups4–7. This pattern was revealed by the analysis of both raw data and stacked species distribution models (see Methods), reducing the potential effect of dispersal limitations on extant species richness patterns. Thus, in contrast with less-mobile organisms, for which historical factors can account for an inverted LDG37, major ecological factors also contribute to the observed patterns of increased species

Figure 3. Correlation between the species richness of taxonomical groups across Europe corrected for spatial autocorrelation, as measured by re-scaled Lee’s L bivariate spatial association. Regions of significant spatial association using a Monte Carlo test on Lee’s statistic at the 95% level. ‘Positive’ indicates values of the Lee’s statistic ranked in the top 97'5% of Monte Carlo values, whilst ‘Negative’ indicates a statistic ranked among the bottom 2'5% Monte Carlo values. Maps generated by V. Gómez-Rubio using R 3.2.2 (R Core Team, https://www.r-project.org). (a) Correlation between mosses and spermatophytes. (b) Correlation between liverworts and spermatophytes. (c) Correlation between ferns and spermatophytes.

Figure 4. Density plots representing (a) the distribution of the turnover (β SIM) and (b) nestedness-resultant multiple-site dissimilarity (β SNE) across 1000 samples of 50 pixels. Components of multiple-site dissimilarity were computed for potential species composition in mosses (red), liverworts (blue), ferns (black) and spermatophytes (green) in northern (latitude above 46th parallel, solid line) and southern Europe (latitude below 46th parallel, dashed line).

www.nature.com/scientificreports/

5Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

richness towards high latitudes in European bryophytes, consistent with the suggestion that habitat suitability and diversity prevail over historical factors (time, speciation, dispersal) in explaining patterns of biodiversity for bryophytes38.

In spermatophytes, the marked increase in SR towards the south is paralleled by an increase in species turn-over and a reduced nestedness compared with northern areas. The higher spatial species turnover in southern vs. northern areas of Europe is consistent with the strong topographical variation in the Mediterranean basin and with the long-term isolation of specialised populations along ecological gradients, which have accumulated mutations within a relatively stable environment resulting in high rates of local endemism39. The markedly lower levels of species turnover in southern mosses, liverworts and ferns compared with southern vascular plants is consistent with our second hypothesis that, unlike spermatophytes, ferns and bryophytes have failed to radi-ate in situ along the strong ecological gradients of the Mediterranean. The significantly higher species turnover observed in southern liverworts as compared to northern ones points, however, to the large difference between assemblages dominated by leafy species in the most humid areas as compared to the assemblages dominated by highly specialized thalloid species such as Riccia, characterized by an annual life-cycle and very large spores able to persist underground during the drought season in the most xeric areas40.

The higher levels of nestedness-resultant dissimilarity observed in northern spermatophytes compared with southern ones point, in turn, to the exclusion of species from northern areas, in support of our third hypothesis that the failure to evolve adaptations to cold climates is a key mechanism of the LDG in this group22. Ferns and mosses exhibited the reverse trend. Fern assemblages, whose distribution and richness patterns are indeed mainly controlled by precipitation levels41–43, were significantly more dissimilar due to nested patterns in the south than in the north, pointing to the exclusion of drought-intolerant species from the south. Moss assemblages, however, did not exhibit significantly higher levels of nestedness-resultant dissimilarity in the south than in the north. Since beta diversity, which represents the slope in species-area relationships44, does not significantly vary across latitudes in mosses, low levels of moss species richness in the south must be interpreted in terms of the lower carrying capacity (i.e., the intercept of the species-area relationship) of southern areas due to the severe constraint of the poikilohydric condition.

The significant difference in beta diversity along a latitudinal gradient among land plants sheds light on the question of whether the turn-over in community composition is progressively slower from spermatophytes, ferns,

βSIM βSNE

SpermatophytesN < S N > S

p < 0.001 p = 0.001

FernsN > S N < S

p = 0.036 p = 0.001

MossesS= N S = N

p = 0.370 p = 0.220

LiverwortsN < S N > S

p = 0.026 p = 0.019

Table 1. Species turnover (βSIM) and nestedness (βSNE) in European spermatophytes (Sp), ferns (Fe), mosses (Mo) and liverworts (Li): significance level (p-value) of the difference between northern (N, latitude above 46th parallel) and southern Europe (S, latitude below 46th parallel) within taxonomic groups (Fig. 4).

Taxonomic groups

north south

βSIM βSNE βSIM βSNE

Mo vs. LiMo > Li Mo < Li Mo = Li Mo = Li

p < 0.001 p < 0.001 p = 0.067 p = 0.052

Mo vs. FeMo = Fe Mo > Fe Mo = Fe Mo = Fe

p = 0.429 p = 0.009 p = 0.111 p = 0.253

Mo vs. SpMo = Sp Mo = Sp Sp > Mo Mo > Sp

p = 0.292 p = 0.278 p = 0.002 p < 0.001

Li vs. FeFe > Li Li > Fe Fe = Li Li > Fe

p < 0.001 p < 0.001 p = 0.266 p = 0.009

Li vs. SpSp > Li Li > Sp Sp > Li Li > Sp

p < 0.001 p < 0.001 p < 0.001 p < 0.001

Fe vs. SpFe = Sp Sp > Fe Sp > Fe Fe > Sp

p = 0.178 p = 0.011 p < 0.001 p < 0.001

Table 2. Species turnover (βSIM) and nestedness (βSNE) in European spermatophytes (Sp), ferns (Fe), mosses (Mo) and liverworts (Li): significance level (p-value) of the difference between groups within northern (N, latitude above 46th parallel) and southern Europe (S, latitude below 46th parallel). (Fig. 4).

www.nature.com/scientificreports/

6Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

and bryophytes, in relationship with the difference in dispersal capacities between these groups36, and suggests different mechanisms of assembly in these groups. Liverworts, in particular, strikingly differed from all groups by exhibiting the highest levels of nestedness both in the north and in the south. Such a pattern suggests that liverworts are the most sensitive group to the impact of past climate change, and in particular, to the higher levels of drought that characterized the glacial periods of the Quaternary and particularly affected frost- and drought-sensitive taxa45. Nevertheless, although the extent of species turnover in the south was substantially and significantly lower than in spermatophytes, liverworts exhibit a similar pattern towards higher species turnover in the south than in the north, strikingly differing from mosses and ferns in this respect. The patterns of beta diversity displayed by ferns and mosses as compared to liverworts observed here is reminiscent of the differences in the slope of the species-area relationships in liverworts as compared to mosses and ferns36. In turn, the simi-larity in the patterns of beta diversity between liverworts and spermatophytes is puzzling. The extent to which the similarity in the patterns displayed by spermatophytes and liverworts along the latitudinal gradient reflects a sim-ilar mechanism of assembly remains, however, to be demonstrated. In the mid-western islands of the Canaries, Madeira and Azores for example, congeneric endemic species generally result from the diversification of a single common ancestor in angiosperms (cladogenetic speciation), but from several independent colonization events in bryophytes (anagenetic speciation)46. This suggests that, in the Mediterranean, the diversity of genera such as Riccia may not necessarily result, like in angiosperms, from a local radiation, but from the recurrent recruitment of pre-adapted species from south-west Asia, where these genera are highly diversified47. While the evolutionary processes underlying the unique diversity of Mediterranean angiosperms have been thoroughly studied39, the evolutionary history of the highly specialized Mediterranean liverwort flora remains a large avenue of research to better understand how poikilohydric organisms may thrive and diversify in dry environments.

MethodsImplementation of species distribution models to circumvent sampling bias. Although Europe is arguably the continent for which the information on species distributions is most detailed, the correspond-ing databases and richness patterns are, even in the best-known groups such as vascular plants, globally biased because some areas have been more intensively investigated than others48. This problem is exacerbated in less-studied organisms such as bryophytes49. To circumvent this issue, we employed species distribution models (SDMs)50, which have become a powerful tool for generating maps of potential distribution, or ecological suita-bility, in areas where distribution information is scarce or lacking.

Available bryophyte distributions include 113,321 records for 1726 species (see Supplementary Methods 1). After removal of the species with less than 15 presences, the data contained 1040 species (including 224 of the 453 liverwort (49%), 810 of the 1,292 moss (63%) and 6 of the 8 hornwort species (73%) of Europe) at a 100 km pixel resolution. Bryophyte species richness was split into mosses and liverworts, two lineages of about 12,000 and 5,000 species. Hornworts should, for consistency, have also been analysed separately. Hornworts are, however, a small group of only about 250 species worldwide whose diversity pales in comparison to the much more diverse liverworts and mosses. The number of hornwort species in our data set did not warrant separate analyses and, because hornworts exhibit a suite of functional vegetative traits and ecological features that are similar to those of thalloid liverworts, the data from the two groups were merged (hereafter referred to as liverworts). Species that were present in fewer than 15 pixels were removed, leaving a total of 1040 species of bryophytes, representing 58% of the total number of species in Europe.

We ran an ensemble model using three different techniques: Generalised Linear Models, Maxent, and Random Forests, as implemented in the R (R Core Team60) package BIOMOD 2.051 (see Supplementary Methods 1). We used the 35 macroclimatic variables of CliMond52 as environmental predictors, as well as monthly and annual potential evapotranspiration53. To avoid multicollinearity, we ran a Pearson correlation analysis eliminating one of the variables in each pair with a correlation value greater than 0.8, as advised by Dormann et al.54. A final set of six variables was used to run the models (see Supplementary Methods 1). For proper evaluation, the models were trained on 70% of the data and evaluated on the remaining 30%. This split-sampling was replicated 10 times. For each species, the potential distribution was considered as a consensus across statistical techniques, evaluation indices and thresholds used to binarise continuous predictions. These individual potential species distributions were stacked (S-SDMs, stacked species distributions models55) to depict the potential SR across Europe (see Supplementary Methods 1).

To validate the resulting potential SR (see Supplementary Methods 2), we compared the maps generated by S-SDMs with 1) the observed bryophyte richness values from a literature review (Supplementary Methods 3), 2) macroecological models (MEMs) of SR for the same study area56 and 3) a sampling effort map for bryophytes in Europe (see Supplementary Methods 3).

To obtain comparable results for bryophytes and vascular plants, we further generated a potential richness S-SDM for ferns and spermatophytes (see Supplementary Methods 1). Data for 2,728 native spermatophyte and fern species from the Atlas Florae Europaeae at 50 km pixel resolution16 were upscaled to a 100 km pixel resolu-tion for consistency with the bryophyte data. Gymnosperms should have been analysed separately but, as in the case of hornworts (see above), the low number of species (20) did not warrant a specific analysis, so that sperma-tophytes were analysed globally. After removal of the species with less than 15 presences, the data included 1,359 and 79 spermatophyte and fern species, representing 12% and 49% of their total diversity in Europe, respectively. The potential richness generated by the S-SDM was then compared with the observed richness values for all the species available in this study (see Supplementary Methods 2)57.

The predictions of the S-SDMs were highly correlated with the observed richness values and the potential richness values of the macroecological models (MEMs, Supplementary Methods 2), supporting the notion that S-SDMs appear as a very promising tool for modelling species assemblages and providing reliable predictions of the geographical variation in species richness55,58. Moreover, these predictions showed only a low correlation with

www.nature.com/scientificreports/

7Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

the map of sampling effort, indicating that the effects of sampling bias were adequately removed (Supplementary Methods 2).

Comparison of SR patterns between bryophytes and vascular plants. The potential richness pre-dicted by S-SDMs was employed to compare the spatial patterns of SR between bryophytes and vascular plants, using three approaches: 1) comparison of potential richness maps, 2) spatial turnover and nestedness, and 3) a latitudinal band analysis of SR.

Comparison of potential richness maps of bryophytes and vascular plants. We began with a com-parison of potential richness maps using two different techniques. First, we calculated and mapped the local Lee’s L bivariate spatial association59 using our own implementation of this statistic with the R language60, which is now included in the ‘spdep’ package (Supplementary Methods 4). In contrast to bivariate association measures such as Pearson’s correlation, Lee’s L captures spatial associations among observations in terms of their point-to-point relationships across two spatial patterns.

Spatial turnover and nestedness in species composition. In the second approach, assemblage multiple-site dissimilarity was measured using the Sørensen index (β SOR) and was partitioned into its turnover (β SIM) and nestedness-resultant (β SNE) components to distinguish between the contribution of spatial species replacement and species loss, respectively61, along the environmental gradients. Potential values for β SIM and β SNE were computed independently for mosses, liverworts, ferns, and spermatophytes in northern and south-ern Europe (defined by the limit of the 46th parallel). Multiple-site dissimilarity was computed 1000 times for randomly sampled subsets of 50 pixels (command beta.sample in R package betapart62), and the resulting dis-tributions of β SIM and β SNE values across the 1000 samples were used to empirically assess whether there were significant differences between northern and southern Europe and between mosses, liverworts, ferns and spermatophytes.

Latitudinal band analysis of species richness. Lastly, we plotted a set of the environmental variables indicated in Table 3 and the potential species richness values for mosses, liverworts, ferns, and spermatophytes for each 100 km latitudinal band across Europe (Supplementary Methods 5).

References1. Crane, P. R. & Lidgard, S. Angiosperm diversification and paleolatitudinal gradients in cretaceous floristic diversity. Science 246,

675–678 (1989).2. Mannion, P. D., Upchurch, P., Benson, R. B. J. & Goswami, A. The latitudinal biodiversity gradient through deep time. Trends Ecol.

Evol. 29, 42–50 (2013).3. Lomolino, M. V., Riddle, B. R., Whittaker, R. J. & Brown, J. H. Biogeography (Sinauer Associates Inc., Sunderland, MA, 2010).4. Gaston, K. J. Global patterns in biodiversity. Nature 405, 220–227 (2000).5. Hillebrand, H. On the generality of the latitudinal diversity gradient. Am. Nat. 163, 192–211 (2004).6. Beaugrand, G., Rombouts, I. & Kirby, R. R. Towards an understanding of the pattern of biodiversity in the oceans. Glob. Ecol.

Biogeogr. 22, 440–449 (2013).7. Jablonski, D. et al. Out of the tropics, but how? Fossils, bridge species, and thermal ranges in the dynamics of the marine latitudinal

diversity gradient. Proc. Natl. Acad. Sci. USA 110, 10487–94 (2013).8. Heikkinen, R. K., Marmion, M. & Luoto, M. Does the interpolation accuracy of species distribution models come at the expense of

transferability? Ecography 35, 276–288 (2012).9. Currie, D. J. et al. Predictions and tests of climate-based hypotheses of broad-scale variation in taxonomic richness. Ecol. Lett. 7,

1121–1134 (2004).10. Chaves, M. M., Maroco, J. P. & Pereira, J. S. Understanding plant responses to drought – from genes to the whole plant. Funct. Plant

Biol. 30, 239–264 (2003).11. Stevens, G. C. The latitudinal gradient in geographical range: how so many species coexist in the tropics. Am. Nat. 133, 240–256

(1989).12. Simpson, G. G. Species density of north American recent mammals. Syst. Zool. 13, 57 (1964).13. Currie, D. J. Energy and large-scale patterns of animal- and plant-species richness. Am. Nat. 137, 27–49 (1991).14. Tognelli, M. F. & Kelt, D. A. Analysis of determinants of mammalian species richness in south America using spatial autoregressive

models. Ecography 27, 427–436 (2004).15. Svenning, J. C. & Skov, F. Could the tree diversity pattern in Europe be generated by postglacial dispersal limitation? Ecol. Lett. 10,

453–460 (2007).

Environmental variable Driver

Mean annual temperature Kinetic energy

Standard deviation of mean annual temperature Environmental heterogeneity

Mean annual precipitation in each pixel Water balance

Standard deviation of mean annual precipitation Environmental heterogeneity

Standard deviation of altitude Environmental heterogeneity

Mean potential evapotranspiration in each pixel Potential energy

Standard deviation of potential evapotranspiration Environmental heterogeneity

Mean distance to the coast (continentality) Spatial and climatic

Mean distance to refugia16 Spatial and historical

Table 3. Variables used in the canonical analyses and macroecological models. Each variable define a possible driver for species richness patterns.

www.nature.com/scientificreports/

8Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

16. Normand, S. et al. Postglacial migration supplements climate in determining plant species ranges in Europe. Proc. Biol. Sci. 278, 3644–53 (2011).

17. Dynesius, M. & Jansson, R. Evolutionary consequences of changes in species’ geographical distributions driven by Milankovitch climate oscillations. Proc. Natl. Acad. Sci. USA 97, 9115–20 (2000).

18. Araújo, M. B. et al. Quaternary climate changes explain diversity among reptiles and amphibians. Ecography 31, 8–15 (2008).19. Hortal, J. et al. Ice age climate, evolutionary constraints and diversity patterns of European dung beetles. Ecol. Lett. 14, 741–8 (2011).20. Baselga, A., Lobo, J. M., Svenning, J.-C., Aragón, P. & Araújo, M. B. Dispersal ability modulates the strength of the latitudinal

richness gradient in European beetles. Glob. Ecol. Biogeogr. 21, 1106–1113 (2012).21. Romdal, T. S., Araújo, M. B. & Rahbek, C. Life on a tropical planet: niche conservatism and the global diversity gradient. Glob. Ecol.

Biogeogr. 22, 344–350 (2013).22. Visser, V., Clayton, W. D., Simpson, D. A., Freckleton, R. P. & Osborne, C. P. Mechanisms driving an unusual latitudinal diversity

gradient for grasses. Glob. Ecol. Biogeogr. 23, 61–75 (2014).23. Fragnière, Y., Bétrisey, S., Cardinaux, L., Stoffel, M. & Kozlowski, G. Fighting their last stand? A global analysis of the distribution

and conservation status of gymnosperms. J. Biogeogr. 42, 809–820 (2015).24. Kreft, H. & Jetz, W. Global patterns and determinants of vascular plant diversity. Proc. Natl. Acad. Sci. USA 104, 5925–5930 (2007).25. Kessler, M. Biogeography of ferns. In Fern Ecology (eds Mehltreter, K., Walker, L. R., Sharpe, J. M.). Cambridge University Press

(2010).26. Salazar, L. et al. Diversity patterns of ferns along elevational gradients in Andean tropical forests. Plant Ecol Divers 8, 13–24 (2015).27. Parra, M. J., Rodríguez, R., Cavieres, L., Muñoz-Tapia, L. & Atala, C. Latitudinal patterns in pteridophyte distribution of continental

Chile. Gayana - Botanica 72, 58–69 (2015).28. Shaw, A. J., Cox, C. J. & Goffinet, B. Global patterns of moss diversity: taxonomic and molecular inference. Taxon 54, 337–352

(2005).29. Geffert, J. L., Frahm, J. P., Barthlott, W. & Mutke, J. Global moss diversity: spatial and taxonomic patterns of species richness. J. Bryol.

35, 1–11 (2013).30. Tan, B. C. & Pocs, T. Bryogeography and conservation of bryophytes. In Bryophyte Biology (eds Shaw, A. J. and Goffinet, B.) 403–448,

Cambridge (Cambridge University Press, 2000).31. Chen, S.-B. et al. Latitudinal diversity gradients in bryophytes and woody plants: Roles of temperature and water availability. J. Syst.

Evol. 53, 535–545 (2015).32. Rozzi, R. et al. Changing lenses to assess biodiversity: patterns of species richness in sub-Antarctic plants and implications for global

conservation. Front. Ecol. Environ. 6, 131–137 (2008).33. Patiño, J., Carine, M., Fernández-Palacios, J. M., Otto, R., Schaefer, H. & Vanderpoorten, A. The anagenetic world of spore-

producing land plants. New Phytol. 201, 305–311 (2014).34. Glime, J. M. Bryophyte ecology (Michigan Technological University, Botanical Society of America & International Association of

Bryologists, 2007).35. Furness, S. B. & Grime, J. P. Growth rate and temperature responses in bryophytes: II. A comparative study of species of contrasted

ecology. J. Ecol. 70, 525–536 (1982).36. Patiño, J. et al. Differences in species–area relationships among the major lineages of land plants: a macroecological perspective.

Glob. Ecol. Biogeogr. 23, 1275–1283 (2014).37. Pyron, R. A. & Burbrink, F. T. Can the tropical conservatism hypothesis explain temperate species richness patterns? An inverse

latitudinal biodiversity gradient in the New World snake tribe Lampropeltini. Glob. Ecol. Biogeogr. 18, 406–415 (2009).38. Patiño, J. et al. Accounting for data heterogeneity in patterns of biodiversity: an application of linear mixed effect models to the

oceanic island biogeography of spore-producing plants. Ecography 36, 904–913 (2013).39. Thompson, J. D. Plant Evolution in the Mediterranean (Oxford University Press, Oxford, 2005).40. Bishler, H. Liverworts of the Mediterranean: Ecology, diversity and distribution (J. Cramer, Berlin, 2004).41. Ferrer-Castán, D. & Vetaas, O. R. Pteridophyte richness, climate and topography in the Iberian Peninsula: Comparing spatial and

nonspatial models of richness patterns. Glob. Ecol. Biogeogr. 14, 155–165 (2005).42. Kessler, M. Patterns of diversity and range size of selected plant groups along an elevational transect in the Bolivian Andes. Biodiver.

Conser. 10, 1897–1921 (2001).43. Bickford, S. A. & Laffan, S. W. Multi-extent analysis of the relationship between pteridophyte species richness and climate. Glob.

Ecol. Biogeogr. 15, 588–601 (2006).44. Triantis, K. A., Guilhaumon, F. & Whittaker, R. J. The island species–area relationship: biology and statistics. J. of Biogeogr. 39,

215–231 (2012).45. Svenning, J.-C. Deterministic Plio-Pleistocene extinctions in the European cool-temperate tree flora. Ecol. Lett. 6, 646–653 (2003).46. Patiño, J. & Vanderpoorten, A. Macaronesia is a departure gate of anagenetic speciation in the moss genus Rhynchostegiella.

J. Biogeogr. 42, 2122–2130 (2015).47. Kürschner, H. & Frey, W. Liverworts, Mosses and Hornworts of southwest Asia (Marchantiophyta, Bryophyta, Anthocerotophyta).

Nova Hedwigia, Beihefte 139, 1–240 (2011).48. Petřík, P., Pergl, J. & Wild, J. Recording effort biases the species richness cited in plant distribution atlases. Perspect. Plant Ecol. Evol.

Syst. 12, 57–65 (2010).49. Mateo, R. G., Vanderpoorten, A., Munoz, J., Laenen, B. & Desamore, A. Modeling species distributions from heterogeneous data for

the biogeographic regionalization of the European bryophyte flora. Plos One 8, e55648 (2013).50. Guisan, A. & Zimmermann, N. E. Predictive habitat distribution models in ecology. Ecol. Model. 135, 147–186 (2000).51. Thuiller, W., Lafourcade, B., Engler, R. & Araújo, M. B. BIOMOD - a platform for ensemble forecasting of species distributions.

Ecography 32, 369–373 (2009).52. Kriticos, D. J. et al. CliMond: global high resolution historical and future scenario climate surfaces for bioclimatic modelling.

Methods Ecol. Evol. 3, 53–64 (2012).53. Zomer, R. J., Trabucco, A., Bossio, D. A. & Verchot, L. V. Climate change mitigation: a spatial analysis of global land suitability for

clean development mechanism afforestation and reforestation. Agric. Ecosyst. Environ. 126, 67–80 (2008).54. Dormann, C. F. et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance.

Ecography 36, 027–046 (2013).55. Mateo, R. G., Felicísimo, Á. M., Pottier, J., Guisan, A. & Muñoz, J. Do stacked species distribution models reflect altitudinal diversity

patterns? Plos ONE 7, e32586 (2012).56. Gotelli, N. J. et al. Patterns and causes of species richness: a general simulation model for macroecology. Ecol. Lett. 12, 873–886

(2009).57. Humphries, C., Araújo, M. B., Williams, P., Lampinen, R., Lahti, T. & Uotila, P. Plant diversity in Europe: Atlas Florae Europaeae and

WORLDMAP. Acta Botanica Fennica 162, 11–21 (1999).58. Aranda, S. C. & Lobo, J. M. How well does presence-only-based species distribution modelling predict assemblage diversity? A case

study of the Tenerife flora. Ecography 34, 31–38 (2010).59. Lee, S. I. Developing a bivariate spatial association measure: An integration of Pearson’s r and Moran’s I. J. Geogr. Syst. 3, 369–385

(2001).

www.nature.com/scientificreports/

9Scientific RepoRts | 6:25546 | DOI: 10.1038/srep25546

60. R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.

61. Baselga, A. Partitioning the turnover and nestedness components of beta diversity. Glob. Ecol. Biogeogr. 19, 134–143 (2010).62. Baselga, A. & Orme, C. D. L. betapart: an R package for the study of beta diversity. Methods Ecol. Evol. 3, 808–812 (2012).

AcknowledgementsRGM was funded by a Marie Curie Intra-European Fellowship within the 7th European Community Framework Programme (ACONITE, PIEF-GA-2013-622620). Many thanks are due to two referees for their constructive comments on the manuscript.

Author ContributionsR.G.M., S.N., J.M. and A.V. collected the data. R.G.M., O.B., B.P., V.G.R. and A.B. analysed the data. All the authors contributed to the writing of the manuscript.

Additional InformationSupplementary information accompanies this paper at http://www.nature.com/srepCompeting financial interests: The authors declare no competing financial interests.How to cite this article: Mateo, R. G. et al. The mossy north: an inverse latitudinal diversity gradient in European bryophytes. Sci. Rep. 6, 25546; doi: 10.1038/srep25546 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license,

unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

The mossy North: an inverse latitudinal diversity gradient in European bryophytes R.G. Mateo, O. Broennimann, S. Normand, B. Petitpierre, M.B. Araújo, J.-C. Svenning, A. Baselga, F. Fernández-González, V. Gómez-Rubio, J. Muñoz, G.M. Suarez, M. Luoto, A. Guisan and A. Vanderpoorten

E-mail: rubeng.mateo@gmail.com

Table of contents

• Supplementary Methods 1: Implementation of species distribution models to

circumvent sampling bias

• Supplementary Methods 2: Validation of stacked species distribution models

• Supplementary Methods 3: List of references employed to build a network of

observed richness patterns for European bryophytes

• Supplementary Methods 4: Comparison of potential richness maps

• Supplementary Methods 5: Comparison of the latitudinal patterns of species

richness for mosses, liverworts, ferns and spermatophytes

Supplementary Methods 1: Implementation of species distribution models to

circumvent sampling bias

Species distribution models (SDMs) have become a powerful tool to predict

distributions in areas where presence points are scarce or lacking1. Two features of

the present study justify, however, the use of SDMs to predict species richness

patterns across Europe. First, error rates diminish drastically when SDMs are applied

to large areas that are representative of the distribution ranges of individual species.

Second, stacked SDMs were shown to provide reliable predictions of spatial variation

in species richness2.

Species occurrences

Bryophyte distributions include 113,321 records for 1,726 species at 100 km pixel

resolution (see ref. 3) following the military grid reference system (MGRS), while

data for 2,728 native species of vascular plants from Atlas Florae Europaeae at 50 km

pixel size (see ref. 4) were upscaled to 100 km pixel size for consistency with

bryophyte data. Species present in less than 15 pixels were removed and were

distributed homogeneously across Europe, leaving a total of 1,438 vascular plants

species (1,359 spermatophytes and 79 ferns) and 1,040 bryophytes species (810

mosses and 224 liverworts). Bryophyte sampling bias was removed by random sub-

sampling of intensively surveyed areas (see ref. 2).

Atlas of Flora Europeae provides a good representation of vascular plant diversity

patterns in western Europe vascular plants, but not in eastern Europe5-6. Therefore,

eastern MGRS pixels were discarded in following analyses.

Environmental predictors

As environmental predictors we used the 35 macroclimatic variables of CliMond

(https://www.climond.org/)7, as well as monthly and annual potential

evapotranspiration (http://www.cgiar-csi.org, see ref. 8). To avoid multicollinearity,

we run a Pearson correlation analysis eliminating one of the variables in each pair

with a correlation value higher than 0.8. The final set of six variables used to run the

models were ‘mean diurnal temperature range’, ‘temperature seasonality’,

‘precipitation seasonality’, ‘mean moisture index of warmest quarter’, ‘mean moisture

index of wettest quarter’ and ‘annual potential evapotranspiration’.

Background selection and statistical modelling

For each species, we generated 10 sets of pseudo-absences equalling the number of

presences and sampled from pixels not adjacent to reported occurrences, that were

later used to produce an ensemble model7 using three different techniques:

generalized linear models (GLM)10, Maxent11, and Random Forests (RF)12, as

implemented in the R13 package BIOMOD 2.014. The performance of the models was

assessed by randomly splitting 10 times the data into a 70% dataset to generate the

models and a 30% dataset to estimate their predictive accuracy. After elimination of

all models with an AUC<0.8, we generated for each species an ensemble model,

consisting in a weighted mean of the models predictions, where the contribution of

each individual technique was proportional to its predictive accuracy.

Evaluation statistics and binarization of species’ potential distribution

Because different ensemble models could generate different models15, we generated

two ensemble models per species, including either (1) only models with an AUC>0.8

(ROC consensus model), or (2) models with a true skill statistic (TSS) > 0.7 (TSS

consensus model). The contribution of each model to the final ensemble model was

proportional to their goodness-of-fit statistics.

If stacking of the binary models reduces the over-prediction2, the selection of an

appropriate threshold still can reduce error rates in both individual and ensemble

SDMs16. To take into account the effects of the selected threshold used to generate

binary models on species’ over-prediction, we produce three different binary models

per ensemble model, in total of six binary models per species. For the ROC consensus

model, we generated models: (1) optimizing the ROC statistics; (2) applying a

maximum of 5 % of omission error (i.e. percentage of the presence predicted as

absences, omission error17; and (3) applying a maximum of 10 % of omission error.

For the TSS consensus model, we also generated three richness models: (1)

optimizing the TSS statistics; (2) applying a maximum of 5 % of omission error; and

(3) applying a maximum of 10 % of omission error.

We stacked (i.e., summed) the different binarised SDMs predictions so that we

obtained 6 stacked species distribution models (S-SDMs) per species group

(bryophytes and vascular plants), which represent the potential species richness for

both groups (Figs. S1 and S2). To compare differences between models, we

calculated the Pearson correlation coefficients by pairs of S-SDMs (Tables S1 & S2).

As correlations were very high, we discarded possible differences between S-SDMs,

so in the following analyses we employed only the S-SDMs obtained by optimizing

the ROC statistics for all the subsequent analysis.

Over-fit testing for bryophytes

By definition, rare species distributions usually have few observations, and are prone

to over-fitting when modelled. To ensure that models were not over-fitted for large set

of predictors, we employed the ‘ensembling of bivariate models' approach presented

by Lomba et al.18. It involves generating bivariate models (using only two

independent variables every time) and averaging them with a weighted ensemble

approach. We have run this approach for bryophytes and climatic variables to

compare the results obtained and discard over-fitting problems. As in the original S-

SDMs we generated 300 models for every species; in order to be comparable we have

run 480 bivariate models per species (4 pseudo-absence sets, 4 replicates, 10 groups

of pairs of independent variables, and 3 modelling techniques). We generated a

consensus (only models with an AUC>0.8) and a binary model per species

(optimizing the ROC statistic). The final richness model generated was compared

with the original richness model to discard over-fitting. The Pearson correlation

coefficient between both approaches was 0.99, hence we consider that over-fitting

problems can be disregarded in the modelling approach followed.

Figure S1. Predicted richness models (S-SDMs) generated for bryophytes. For the

ROC consensus model, we generated three models: (1) optimizing the ROC statistics

(ROC), (2) applying a maximum 5 % of omission error (ROC.COM5), and (3)

applying a maximum of 10 % of omission error (ROC.COM10). For the TSS

consensus model, we generated also three models: (1) optimizing the TSS statistics

(TSS), (2) applying a maximum of 5 % of omission error (TSS.COM5), and (3)

applying a maximum of 10 % of omission error (TSS.COM10). Maps generated by

R.G. Mateo using the ArcMap extension in ArcGIS 10.2 (ESRI Inc., Redlands, CA,

USA, http://www.esri.com).

Figure S2. Predicted richness models (S-SDMs) generated for vascular plants. For

abbreviations see Figure S1. Maps generated by R.G. Mateo using the ArcMap

extension in ArcGIS 10.2 (ESRI Inc., Redlands, CA, USA, http://www.esri.com).

Table S1. Pearson correlation coefficient between all the S-SDMs generated for

bryophytes. For abbreviations see Figure S1.

ROC ROC.COM5 ROC.COM10 TSS TSS.COM5 TSS.COM10

ROC 1.00 0.98 1.00 1.00 0.99 1.00

ROC.COM5 0.98 1.00 0.99 0.98 1.00 0.99

ROC.COM10 1.00 0.99 1.00 1.00 1.00 1.00

TSS 1.00 0.98 1.00 1.00 0.99 1.00

TSS.COM5 0.99 1.00 1.00 0.99 1.00 0.99

TSS.COM10 1.00 0.99 1.00 1.00 0.99 1.00

Table S2. Pearson correlation coefficient between all the S-SDMs generated for

vascular plants. For abbreviations see Figure S1.

ROC ROC.COM5 ROC.COM10 TSS TSS.COM5 TSS.COM10

ROC 1.00 1.00 1.00 1.00 1.00 0.99

ROC.COM5 1.00 1.00 0.99 0.99 1.00 0.99

ROC.COM10 1.00 0.99 1.00 0.99 0.99 1.00

TSS 1.00 0.99 0.99 1.00 1.00 0.99

TSS.COM5 1.00 1.00 0.99 1.00 1.00 0.99

TSS.COM10 0.99 0.99 1.00 0.99 0.99 1.00

References

1. Guisan, A. & Zimmermann, N.E. Predictive habitat distribution models in ecology.

Ecol. Model. 135, 147-186 (2000).

2. Mateo, R. G., Felicísimo, Á. M., Pottier, J., Guisan, A. & Muñoz, J. Do stacked

species distribution models reflect altitudinal diversity patterns? PLoS ONE 7, e32586

(2012).

3. Mateo, R. G., Vanderpoorten, A., Muñoz, J., Laenen, B. & Désamoré, A. Modeling

species distributions from heterogeneous data for the biogeographic regionalization of

the European bryophyte flora. PLOS ONE 8, e55648 (2013).

4. Normand, S. et al. Postglaciation migration supplements climate in determining

plant species ranges in Europe. Proc. R. Soc. Lond. B Biol. Sci. 278, 3644-3653

(2011).

5. Ronk A, Szava-Kovats R, Pärtel M. Applying the dark diversity concept to plants at

the European scale. Ecography 38, 1015–1025 (2015).

6. Kalwij JM, Robertson MP, Ronk A, Zobel M, Pärtel M. Spatially-Explicit

Estimation of Geographical Representation in Large-Scale Species Distribution

Datasets. PLoS ONE 9, e85306 (2014).

7. Kriticos, D. J. et al. CliMond: global high-resolution historical and future scenario

climate surfaces for bioclimatic modelling. Methods Ecol. Evol. 3, 53-64 (2011).

8. Zomer, R. J., Trabucco, A., Bossio, D. A. & Verchot, L. V. Climate change

mitigation: A spatial analysis of global land suitability for clean development

mechanism afforestation and reforestation. Agric. Ecosyst. Environ. 126, 67-80

(2008).

9. Araújo, M. B. & New, M. Ensemble forecasting of species distributions. Trends

Ecol. Evol. 22, 42-47 (2007).

10. McCullagh, P. & Nelder, J. A. Generalized Linear Models, 2nd ed (Chapman &

Hall, London, 1989).

11. Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maximum entropy modeling of

species geographic distributions. Ecol. Modell. 190, 231-259 (2006).

12. Breiman, L. Random forests. Mach. Learn. 45, 5-32 (2001).

13. R Core Team (2013). R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-

project.org/".

14. Thuiller, W., Lafourcade, B., Engler, R. & Araújo, M. B. BIOMOD - a platform

for ensemble forecasting of species distributions. Ecography 32, 369-373 (2009).

15. Marmion, M., Parviainen, M., Luoto, M., Heikkinen, R. K. & Thuiller, W.

Evaluation of consensus methods in predictive species distribution modelling. Divers.

Distrib. 15, 59-69 (2009).

16. Trotta-Moreu, N. & Lobo, J. M. Deriving the species richness distribution of

Geotrupinae (Coleoptera: Scarabaeoidea) in Mexico from the overlap of individual

model predictions. Environ. Entomol. 39, 42-49 (2010).

17. Fielding, A. H. & Bell, J. F. A review of methods for the assessment of prediction

errors in conservation presence/absence models. Environ. Conserv. 24, 38-49 (1997).

18. Lomba, A. et al. Overcoming the rare species modelling paradox: a novel

hierarchical framework applied to an Iberian endemic plant. Biol. Conserv. 143, 2647-

2657 (2010).

Supplementary Methods 2: Validation of stacked species distribution models

Potential richness maps obtained by stacked species distribution models (S-SDMs)

have been rarely evaluated until recently in a few studies1-3, and never for

inconspicuous organism like bryophytes. The first step was to evaluate the S-SDMs

obtained to check if the sampling bias was solved. We compared the S-SDMs

obtained for bryophytes with i) observed species richness values of bryophytes from a

literature review, ii) a macroecological model (MEM)4, and iii) a sampling effort map

for bryophytes in Europe.

For the first comparison we produced a dataset of observed species richness for 45

UTM squares (100x100 km) scattered across Europe and intensively sampled

according to a previous review of the literature (Figure S3, see Appendix S5 for the

complete list of references).

Then we generated a set of variables describing extant macroclimatic conditions,

environmental heterogeneity, spatial patterns and historical factors across Europe (see

Table 1 of the main manuscript). We selected the variables that best describe patterns

of bryophyte species richness across the 45 UTM squares through a multimodel

inference approach5 approach. We employed the MuMin R package6 to select the

variables with higher relative importance along a set of all the possible generalized

linear models in groups of four variables to avoid overparameterization, following the

sample size rule-of thumb of 10:1 subjects to predictors in multiple regression7. The

four variables with higher importance along all the possible GLM models were:

continentality, standard deviation of annual temperature, standard deviation of

altitude, and distance to refugia. The GLM model run with the four variables obtained

the lower AIC value (1851).

The importance of distance to refugia suggests at first sight that species are still

concentrated close to the areas least affected by Pleistocene ice age cooling, and

hence, exhibit post-glacial dispersal limitations. Such a result was unexpected for two

reasons. Fist, the relevance of distance to southern refugia in extant bryophyte species

richness patterns implicitly suggests that species survived the ice age in

Mediterranean refugia (southern refugium hypothesis) rather than in extra-

Mediterranean micro-refugia (northern refugium hypothesis), in contrast to some

phylogeographical evidence8. Second, such an important contribution of historical

factors in extant patterns of bryophyte species richness, also observed in other groups

such as angiosperms9 and reptiles10, is at odds with the high long-distance dispersal

capacities, which were thought to erase any historical signal in extant patterns of

species richness11. In fact, bryophytes quickly and massively responded to past

climate change based on phylogeographic12, 13 and macrofossil 14, 15 evidence,

supporting the view that bryophyte distributions and species richness patterns are at

equilibrium with contemporary ecological factors11. An important caveat to such

interpretations, however, is the strong correlation between variables. For instance,

distance to refugia, which appears at first sight as a historical factor, is strongly

correlated with evapotranspiration16, confounding the interpretation that can be made

from the selection of the best-fit variables in the model.

The four variables selected were employed as predictors of a macroecological

model (MEM)4 built on the 45 observed species richness values. Generalized linear

models (GLM)17; R library ‘glm’) were implemented using a quadratic function,

calibrated with a Poisson distribution and a logarithmic link function. This MEM was

consequently projected onto Europe to predict bryophyte species richness across the

continent (Figure S3).

On the other hand, we downloaded all the collections available for bryophytes in

the Global Biodiversity Information Facility (GBIF) database across Europe. We

counted the number of collections per UTM squares (100x100 km) to generate a map

of sampling effort for bryophytes in Europe (Figure S3).

Finally we calculated the Pearson correlation coefficients between the three maps

generated on the 45 pixels with observed richness values from bibliography. Those

values were highly correlated with the potential richness obtained by S-SDMs and

MEM, and very low correlated with sampling effort. In conclusion, the potential

bryophyte richness (S-SDMs) map is not biased by sampling effort, and it is a good

tool to study biodiversity patterns for bryophytes in Europe.

We also compare the S-SDMs obtained for vascular plants and observed richness

patterns. We generated a map of observed richness patterns (Figure S4) for vascular

plants with all the data available in the study9.

Figure S3: Potential species richness for bryophytes calculated with stacked species

distribution models (S-SDMs) and with a macroecological model (MEM), and a map

of sampling effort for bryophytes in the GBIF database (number of collections

available in the GBIF database for bryophytes in each pixel). Maps generated by R.G.

Mateo using the ArcMap extension in ArcGIS 10.2 (ESRI Inc., Redlands, CA, USA,

http://www.esri.com).

Table S3: Pearson correlation coefficients between observed bryophyte species

richness (45 pixels, information extracted from bibliography), potential bryophyte

species richness calculated with S-SDMs, potential bryophyte species richness

estimated with a MEM, and sampling effort for bryophytes in the GBIF database.

Observed S-SDMs GBIF MEM

Observed 1.00 0.91 0.23 0.90

S-SDMs 0.91 1.00 0.27 0.88

GBIF 0.23 0.27 1.00 0.22

MEM 0.90 0.88 0.22 1.00

Figure S4: Potential plant richness obtained by stacked species distribution models

(S-SDMs) and observed vascular plant richness in the complete database of Atlas

Florae Europaeae (AFE). Maps generated by R.G. Mateo using the ArcMap extension

in ArcGIS 10.2 (ESRI Inc., Redlands, CA, USA, http://www.esri.com).

References

1. Dubuis, A. et al. Predicting spatial patterns of plant species richness: a comparison

of direct macroecological and species stacking approaches. Divers. Distrib. 17, 1122-

1131 (2011).

2. Distler, T., Schuetz, J. G., Velásquez-Tibatá, J. & Langham, G. M. Stacked species

distribution models and macroecological models provide congruent projections of

avian species richness under climate change. J. Biogeogr. 42, 976-988 (2015).

3. Mateo, R. G., Felicísimo, Á. M., Pottier, J., Guisan, A. & Muñoz, J. Do stacked

species distribution models reflect altitudinal diversity patterns? PLoS ONE 7, e32586

(2012).

4. Gotelli, N. J. et al. Patterns and causes of species richness: a general simulation

model for macroecology. Ecol. Lett. 12, 873-886 (2009).

5. Grueber, C. E., Nakagawa, S., Laws, R. J. & Jamieson, I. G. Multimodel inference

in ecology and evolution: challenges and solutions. J. Evol. Biol. 24, 699-711 (2011).

6. Barton, K. MuMIn: multi-model inference. R package, version 1.14.0. 2015. URL:

http://r-forge.r-project.org/pro-jects/mumin/

7. Harrell, F. E. Regression Modeling Strategies with Applications to Linear Models,

Logistic Regression, and Survival Analysis (Springer, New York, 2001).

8. Désamoré, A., Laenen, B., Stech, M., Papp, B., Hedenäs, L., Mateo, R.G. &

Vanderpoorten, A. How do temperate bryophytes face the challenge of a changing

environment? Lessons from the past and predictions for the future. Glob. Chang. Biol.

18. 2915-2924 (2012).

9. Normand, S., Ricklefs, R. E., Skov, F., Bladt, J., Tackenberg, O. & Svenning, J. C.

Postglacial migration supplements climate in determining plant species ranges in

Europe. Proc. R. Soc. Lond. B 278, 3644-3653 (2011).

10. Araújo, M.B., Nogués-Bravo, D., Diniz-Filho, J.A.F., Haywood, A.M., Valdes,

P.J. & Rahbek, C. Quaternary climate changes explain diversity among reptiles and

amphibians. Ecography 31. 8-15 (2008).

11. Patiño, J. et al. Accounting for data heterogeneity in patterns of biodiversity: an

application of linear mixed effect models to the oceanic island biogeography of spore-

producing plants. Ecography, 36, 904-913 (2013).

12. Van der Velde, M. & Bijlsma, R. Phylogeography of five Polytrichum species

within Europe. Biol. J. Linn. Soc. 78. 203-213 (2003).

13. Szövényi, P., Sundberg, S. & Shaw, A.J. Long-distance dispersal and genetic

structure of natural populations: an assessment of the inverse isolation hypothesis in

peat mosses. Mol. Ecol. 21. 5461-5472 (2012).

14. Jonsgard, B. & Birks, H.H. Late-glacial mosses and environmental reconstructions

at Krakenes, western Norway. Lindbergia 20. 64-82 (1996).

15. Ellis, C.J. & Tallis, J.H. Climatic control of blanket mire development at Kentra

Moss, north-west Scotland. J. Ecol. 88. 869-889 (2000).

16. Fløjgaard, C., Normand, S., Skov, F. & Svenning, J.-C. Deconstructing the

mammal species richness pattern in Europe – towards an understanding of the relative

importance of climate, biogeographic history, habitat heterogeneity and humans.

Global Ecol. Biogeogr. 20 218-230 (2011).

17. McCullagh, P. & Nelder, J.A. Generalized linear models (Chapman & Hall,

London, 1989).

Supplementary Methods 3: List of references employed to build a network of

observed richness patterns for European bryophytes

1- National Inventory of Swiss Bryophytes of National Inventory of Swiss

Bryophytes (Institute of Systematic Botany, University of Zürich):

http://www.nism.uzh.ch/map/map_en.php. Available at: (accessed July 2014

2014).

2- Aleffi, M. & Schumacker, R. Check-list and red-list of the liverworts

(Marchantiophyta) and hornworts (Anthocerotophyta) of Italy. Flora

Mediterranea, 5, 73-161 (1995).

3- Atherton, I., Bosanquet, S. & Lawley, M. Mosses and Liverworts of Britain and

Ireland. A Field Guide. (British Bryological Society. London, 2010)

4- Cano, J.M., Ros, R.M. & Guerra, J. Flora briofítica de la provincia de Alicante.

Cryptogam. Bryol. 17, 251-277 (1996).

5- Cano, J.M., Guerra, J., Jiménez, J.A., Gallego, M.T. & Orgaz, J.D. An updated

Bryophytes Check-list of the Region of Murcia (Southeastern Spain). Anales de

Biología 32, 101-131 (2010).

6- Cano, M.J., Gallego, M.T., Garilleti, R., Juaristi, R., Lara, F., Abaigar, J.M.,

Mazimpaka, V., Rosselló, J.A., Sánchez-Moya, M.C. & Urdíroz, A.

Aportaciones al conocimiento de la flora briológica española. Notula XIII:

hepáticas y musgos de Mallorca (Islas Baleares). Boletín de la Sociedad

Española de Briología 18/19, 103-110 (2001).

7- Cogoni, A., Flore, F. & Aleffi, M. Survey of the bryoflora on Monte Limbara

(Northern Sardinia). Cryptogam. Bryol. 23, 73-86 (2002).

8- Cortini Pedrotti, C. New Check-list of the Mosses of ltaly. Flora Mediterranea

11, 23-107 (2001).

9- Cros, R.M., Sáez, L. & Brugués, M. The bryophytes of the Balearic Islands: an

annotated checklist. J. Bryol. 30, 74–95 (2008).

10- Dragicevic, S., Veljic, M. & Marín, P. New records to the moss flora of

Montenegro. Cryptogam. Bryol. 29, 397-400 (2008).

11- Draper, I., Albertos, B., Brugués, M., Cano, M.J., Cros, R.M., Gallego, M.T.,

Garilleti, R., Guerra, J., Lara, F. & Mazimpaka, V. Aportaciones al

conocimiento de la flora briológica española. Notula XIV: musgos, antocerotas

y hepáticas de la sierra de Aracena (Huelva). Boletín de la Sociedad Española

de Briología 24, 7-14 (2004).

12- Düll, R. Moosflora der nördlichen Eifel. (Bad Münstereifel, 1995).

13- Elías, M.J., Albertos, B., Brugués, M., Calabrese, G., Cano, M.J., Estébanez, B.,

Gallego, M.T., Garilleti, R., Guerra, J., Heras, P., Infantes, M., Lara, F., Martín,

M.A., Mazimpaka, V., Medina, R., Muñoz, J., Pokorny, L., Puche, F. &

Sánchez, J.A. Aportaciones al conocimiento de la flora briológica española.

Notula XV: musgos, antocerotas y hepáticas de la Sierra de Gredos (Ávila).

Boletín de la Sociedad Española de Briología 28, 25-31 (2006).

14- Frahm, J.P. La Bryoflore des Vosges et des zones limitrophes. Universität-

(Gesamthochschule-Duisburg, 1989).

15- Frahm, J.P. & Walsemann, E. Nachträge zur Moosflora von Schleswig-

Holstein. Mitt. Arbeitsgem. Geobotanik Heft 23, 1-205 (1973).

16- Ganeva, A. & Natcheva, R. Check-list of the bryophytes of Bulgaria with data

on their distribution. I. Hepaticae and Anthocerotae. Cryptogam. Bryol. 24, 229-

239 (2003).

17- Heras, P. & Infante, M. Actualización y contribuciones a la brioflora de Burgos.

Boletín de la Sociedad Española de Briología 27-31 (2004).

18- Ingerpuu, N. & Vellak, K. Bryologically important sites in Estonia. Lindbergia

25, 106 – 111 (2000).

19- Kalinauskaitė, P. & Pippo, S. The bryophyte flora of Bromarv, southwest

Finland, based on Hans Buch's Reliquiae and other collections. Bryobrothera 9,

1-49 (2010).

20- Loeske, L. Mossflora des Harzes. Bornträger, Leipzig. (1903).

21- Maksimov, A., Potemkin, A., Hokkanen, T. & Maksimova, T. Bryophytes of

fragmented old-growth spruce forest stands of the north Karelian Biosphere

Reserve and adjacent areas of Finland. Arctoa 12, 9-23 (2003).

22- Mårtensson, O. Mossflora och mossvegetation krin Sjön Keddek I

Taurejuätnodalen I lule Lappmark. (Stockholm, 1962).

23- Natcheva, R. & Ganeva, A. Check-list of the bryophytes of Bulgaria. II. Musci.

Cryptogam. Bryol. 26, 209-232 (2005).

24- Papp, B., Alegro, A., Šegota, V., Šapić, I. & Vukelić, J. (2013) Additions to the

bryophyte flora of Croatia. J. Bryol. 35, 140-143.

25- Pérez, P.H., Infante, M., Casas, C., Cros, R.M. & Brugués, M. Contribución a la

brioflora del Pirineo Aragonés. Boletín de la Sociedad Española de Briología

25, 25-32 (2004).

26- Puche, F. & Gimeno, C. Flora briofítica de la Muela de Cortes y del Macizo del

Caroche (Valencia, España). Boletín de la Sociedad Española de Briología 24,

15–26 (2004).

27- Sabovljević & Cvetić, T. Bryophyte Flora of Avala Mt. (C. Serbia, Yugoslavia).

Lindbergia 28, 90-96 (2003).

28- Sabovljevic, M. Bryophyte flora of South Banat. (Vojvodina, Yugoslavia).

Cryptogam. Bryol 24, 241-252 (2003).

29- Sabovljevic, M., Tsakiri, E. & Sabovljevic, A. Towards the bryophyte flora of

Greece, studies in Chalkidiki area (North Greece). Cryptogam. Bryol. 29, 143-

145 (2008).

30- Sérgio, C., Brugués, M., Cros, R.M., Garcia, C. & Louro, T. A new important

Mediterranean area for bryophytes in Portugal: Barrancos (Baixo Alentejo).

Boletín de la Sociedad Española de Briología 29, 25-33 (2006).

31- Sérgio, C., Brugués, M., Cros, R.M., Garcia, C. & Stow, S. First bryofloristic

study of the Tejo International Region (Portugal). Boletín de la Sociedad

Española de Briología 37, 1-10 (2011).

32- Vieira, C., Séneca, A. & Sérgio, C. The bryoflora of Valongo. The refuge of

common and rare species Boletín de la Sociedad Española de Briología, 1-15

(2004).

33- Werner, J. Check-list of the bryophytes of Luxembourg. J. Bryol. 17, 489-500

(1993).

34- Werner, J., Schneider, T., Schneider, C. & Mahévas, T. Les bryophytes de la

Lorraine extra-vosgienne. Liste critique annotée. Cryptogam. Bryol. 26, 347-

402 (2005).

35- Werner, J., Bardat, J., Vanot, M. & Prey, T. Bryophyte (Anthocerotae,

Hepaticae, Musci) check-list of upper Normandy (France). Cryptogam. Bryol.

30, 457-475 (2009).

Supplementary Methods 4: Comparison of potential richness maps

Potential richness (stacked species distribution models, S-SDMs) maps were

employed to compare the spatial patterns of species richness among mosses,

liverworts, ferns and spermatophytes. The first step was two couple pixel by pixel

richness patterns analyses.

First, we calculated and mapped local Lee’s L bivariate spatial association1 using

our own implementation of this statistics with the R language. This function is now

included in the ‘spdep’ package2. Lee’s L statistics is a measure of the pixel-to-pixel

correlation between two spatial variables, corrected for the spatial autocorrelation of

each variable. As Moran’s I index, Lee’s L is not necessarily centered at 0, and its

limits can be outside of the (-1,1) interval3. To facilitate the interpretation of the L

statistics, we centred it on 0 and fixed its boundaries between -1 and 1 by subtracting

the mean of the local Lee’s L values and dividing the result by the maximum value

that L could take. An associated quantile was computed for each value of local Lee’s

L statistics using a Monte Carlo test with 999 simulations so that significant positive

(quantile >0.975) or negative (quantile <0.025) spatial association can be detected.

To complement this analysis, we mapped the residuals of a Procrustes4 analysis

with the “vegan” R package. Procrustes rotation allows the comparison of the

ordinations of two data matrices through an algorithm that minimises the sum of

squared distances between corresponding points belonging to the two matrices.

Procrustean comparison was applied to two canonical correspondence analyses

(CCAs)5 performed with the ade4 R-package6 and conducted using the matrices of

spermatophytes, mosses, liverworts and ferns species probabilities in each 100 km

pixel against a set of environmental variables (Table 1) capable of explaining species

composition patterns. High residual values indicate areas where the major gradients

differ, i.e., where the influence of the environmental variables differed for

spermatophytes and the other three groups.

The CCA with spermatophytes and environmental variables produced two axes,

the first one expressing the temperature, and the other one environmental

heterogeneity. The highest residual values are given in the most heterogeneous areas

(Alps and Pyrenees), so in this area composition is mostly explained with these

environmental variables. The CCAs with ferns, mosses or liverworts and

environmental variables produced two axes, the first one expressing the temperature,

and the other one potential evapotranspiration and environmental heterogeneity.

Inertia of spermatophytes, ferns, mosses, and liverworts composition explained by

environmental variables was 65.05%, 68.67%, 67.44%, and 66.13%, respectively. The

CCA gets the most important gradients of the independent variables to explain the

richness patters for each group (Fig. S6), and the Procrustes analysis (Fig. S7)

compare both CCA to see if they are the same. High differences between gradients

(high values of residuals) indicate that environmental variables have different

influence for the taxonomical groups compared.

Figure S5. Correlation between the species richness of mosses, liverworts, ferns and

spermatophytes across Europe corrected for spatial autocorrelation, as measured by

re-scaled Lee's L bivariate spatial association. Regions of significant spatial

association using a Monte Carlo test on Lee's statistic at the 95% level. ‘Positive’

indicates values of the Lee's statistic ranked in the top 97'5% of Monte Carlo values,

whilst ‘Negative’ indicates a statistic ranked among the bottom 2'5% Monte Carlo

values. Maps generated by V. Gómez-Rubio using R 3.2.2 (R Core Team,

https://www.r-project.org). (a) Correlation between mosses and spermatophytes. (b)

Correlation between liverworts and spermatophytes. (c) Correlation between ferns and

spermatophytes.

Figure S6a. Results of canonical correspondence analysis of species composition of

spermatophytes and environmental variables (see Table 1).

Figure S6b. Results of canonical correspondence analysis of species composition of

ferns and environmental variables (see Table 1).

Figure S6c. Results of canonical correspondence analysis of species composition of

mosses and environmental variables (see Table 1).

Figure S6d. Results of canonical correspondence analysis of species composition of

liverworts and environmental variables (see Table 1).

Figure S7. Residuals of the Procrustes analysis based in the comparison of two CCA.

Map generated by R.G. Mateo using the ArcMap extension in ArcGIS 10.2 (ESRI

Inc., Redlands, CA, USA, http://www.esri.com). (a) CCA of species composition of

mosses and environmental variables vs. CCA of species composition of

spermatophytes and environmental variables. (b) CCA of species composition of

liverworts and environmental variables vs. CCA of species composition of

spermatophytes and environmental variables. (c) CCA of species composition of ferns

and environmental variables vs. CCA of species composition of spermatophytes and

environmental variables

References

1. Lee, S.I. Developing a bivariate spatial association measure: an integration of

Pearson's r and Moran's I. J. Geogr. Syst. 3, 369-385 (2001).

2. Bivand, R.S., Gómez-Rubio, V. & Rue, H. Spatial data analysis with R-INLA with

some extensions. J. Stat. Softw. 63, 31 (2015).

3. Goodchild, M.F. Spatial Autocorrelation. (Geo Books, Norwich, UK, 1986).

4. Oksanen, J. Multivariate analysis of ecological communities in R: vegan tutorial

[online]. Available from http://cc.oulu.fi/~jarioksa/opetus/metodi/vegantutor.pdf

[accessed May 2015].

5. Dray, S., Pélissier, R., Couteron, P., Fortin, M.J., Legendre, P., Peres-Neto, P.R.,

Bellier, E., Bivand, R., Blanchet, F.G., De Cáceres, M., Dufour, A.B., Heegaard, E.,

Jombart, T., Munoz, F., Oksanen, J., Thioulouse, J. & Wagner, H.H. Community

ecology in the age of multivariate multiscale spatial analysis. Ecol. Mon. 82, 257-275

(2012).

6. Dray, S., Dufour, A.B. & Chessel, D. The ade4 package-II: Two-table and K-table

methods. R News 7, 47-52 (2007).

Supplementary Methods 5: Comparison of the latitudinal patterns of species

richness for mosses, liverworts, ferns and spermatophytes

To illustrate the results obtained in the pixel analyses we provide the following graphs

of the latitudinal patterns of species richness (SR) of mosses, liverworts, ferns and

spermatophytes (Fig. S8a), as well as some environmental variables and other data

used in the study calculated for latitudinal bands of 100 km stretch across Europe

(Fig. S8b-i).

In order to represent adequately the latitudinal gradients of SR, the observed SR

values (OSi; in our case, modelled through stacked species distribution models, S-

SDMs) for every latitudinal band i of 100 km stretch were calculated for every

taxonomic group. Each latitudinal band has an effective area Ai (the sum of the areas

of the pixels included in the band, excluding sea surfaces). In the measure that SR is

dependent upon area, we need an estimation of the expected SR values (ESi) if SR

was an exact function of area. The most consistent mathematical relationship between

SR (S) and area (A) (species-area relationship, SAR) is the power function1: S=cAz.

For deriving a value of c, we assume that total species presences are the same in

expected and observed SR sets, i.e. ΣESi=ΣOSi. Hence, ESi=Ai z ΣOSi / Σ(Ai

z).

For the estimation of the value of z, SR was calculated for groups of contiguous

pixels ranging in area from ∼104 km2 (1 pixel) to ∼64*104 km2 (64 pixels), completely

nested within them and covering almost all Europe. Pixels covering sea surfaces were

excluded from this analysis. The species-area relationship was calculated via

regression of SR on area (Dengler 2009). The z values obtained were slightly higher

for spermatophytes (0.185) than for liverworts (0.165), mosses (0.163) and ferns

(0.133). These values are common for species-area curves from large mainland areas2.

According to these SAR equations for Europe, the species richness corrected for area

effects (CSRi) were calculated as: CSRi = OSRi – ESRi + MSR, being MSR the mean

species richness across the latitudinal gradient.

A simple test to prove the existence of a latitudinal gradient in SR across

latitudinal bands independent of their area was carried out. Linear regressions of the

SR values normalized by area on latitude were calculated and we checked if their

slopes (a) differ significantly from zero. A significant slope was found for mosses and

spermatophytes, positive for the former (a=6.87±1.692, t35=4.06, p=.0003) and

negative for the latter (a=-24.26±2.251, t35=-10.78, p<10-11). Slope was non-

significant for ferns (a=0.063±0.159, t35=.396, p=0.69) and only marginally

significant for liverworts (a=-1.77±0.872, t35=-2.03, p=0.0501).

Figure S8: Graphs of the latitudinal patterns of species richness (SR) of mosses,

liverworts, ferns and spermatophytes, as well as some environmental variables and

other data used in the study calculated for latitudinal bands of 100 km stretch across

Europe.

Figure S8a: Predicted numbers of species of spermatophytes (green), ferns (black),

mosses (red) and liverworts (blue) in 100 km latitudinal bands across Europe. Dashed

lines indicate crude SR values predicted by S-SDMs, solid lines correspond to SR

values normalized according to species-area relationships. In the lower graph SR

values of the four groups have been rescaled to allow comparisons.

200

400

600

800

1000

1200

36 41 46 51 56 61 66 71

Spe

cies

rich

ness

latitude (°)

0

50

100

150

200

250

36 41 46 51 56 61 66 71

Spe

cies

rich

ness

latitude (°)

36 41 46 51 56 61 66 71

Spe

cies

rich

ness

(res

cale

d)

latitude (°)

Figure S8b: Mean and SD elevation range by latitudinal band.

Figure S8c: Mean and SD annual temperature (ºC x 10) by latitudinal band.

0

100

200

300

400

500

600

700

80037.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

Al#tude Mean

Sd

-40

-20

0

20

40

60

80

100

120

140

160

180

37.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

TemperatureMean

Sd

Figure S8d: Mean and SD annual precipitation (mm/m2) by latitudinal band.

Figure S8e: Mean and SD annual PET (mm m-2 day-1) by latitudinal band.

0

100

200

300

400

500

600

700

800

900

1000

37.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

Precipita#on Mean

Sd

0

200

400

600

800

1000

1200

1400

37.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

PETMean

Sd

Figure S8f: Area covered by this study by latitudinal band.

0.0000

5.0000

10.0000

15.0000

20.0000

25.0000

37.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

Area

Figure S8g: Number of collection for bryophytes in the database employed in this

study, by latitudinal band.

Figure S8h: Number of collection for bryophytes in the database employed in this

study normalized by area, by latitudinal band.

0

1000

2000

3000

4000

5000

6000

7000

800037.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

Collec#ons(bryophytes)

0

100

200

300

400

500

600

700

37.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

70.2

Collec#ons(bryophytes)normalized

Figure S8i: Number of collection for bryophytes in the GBIF database by latitudinal

band.

References

1. Dengler J. Which function describes the species-area relationship best? A review

and empirical evaluation. J. Biogeogr. 36, 728-744 (2009).

2. Rosenzweig M.L. Species diversity in space and time. (Cambridge University

Press, Cambridge, 1995).

0

500

1000

1500

2000

250037.6

39.3

41.1

42.9

44.7

46.5

48.4

50.1

51.9

53.7

55.5

57.3

59.1

60.9

62.7

64.5

66.3

68.1

GBIF(bryophytes)