1Research paper

2 Phylogenetic Age Differences in Tree Assemblages across the Northern Hemisphere Increase with
3 Long-term Climate Stability in Unstable Regions

4Gang Feng1,2, Ziyu Ma2, Blas M. Benito2, Signe Normand2, Alejandro Ordonez2, Yi Jin3, Lingfeng
5Mao4,*, and Jens-Christian Svenning2

61School of Life SciencesEcology and Environment, Inner Mongolia University, 010021, Hohhot, China

72Section for Ecoinformatics and Biodiversity, Department of Bioscience, Aarhus University, Ny

8Munkegade 114, DK-8000 Aarhus C, Denmark

93College of Life Sciences, Zhejiang University, Hangzhou, 310058, China

104Department of Plant Biology, Michigan State University, East Lansing, Michigan, 48824, USA

11G Feng (qaufenggang@163.com)

12ZY Ma (paul.ziyu.ma@gmail.com)

13BM Benito (blasbenito@gmail.com)

14S Normand (signe.normand@bios.au.dkgmail.com )

15A Ordonez (alejandro.ordonez@biologybios.au.dk)

16Y Jin (jinyi1984@zju.edu.cn)

17LF Mao (maolingfeng2008@163.com)

18JC Svenning (svenning@bios.au.dk)

19GF and ZYM contributed equally to this work.

20Keywords Quaternary-scale climate variability, North Hemisphere, phylogenetic age differences,

21speciation, extinction, tree assemblage.

22Running title Long-term climate stability and phylogenetic age differences.

23Corresponding author Lingfeng Mao.

24The number of words in the Abstract 243.

25The number of words in main body of the paper 2764.

26The number of references 60.

27Abstract

28Aims Long-term climate stability is hypothesized to drive the emergence of species assemblages with
29large species age differences due to the accumulation of relict species and/oras well as relatively newly
30arisen species via reduced extinction and increased speciation. Few studies have addressed these
31predictions and so far no study has done so for plants at near-global scalesinacross the Northern
32Hemisphere. Here, we linked Quaternary-scale climate variability to phylogenetic age differences
33between the oldest and youngest group of species in tree assemblages in 100 km × 100 km grid cells
34across the Northern Hemisphere to test these predictions.

35Location Northern Hemisphere

36Methods Last Glacial Maximum (LGM)-to-present shifts in temperature and precipitation were used as
37proxies for Quaternary-scale glacial-interglacial variability. Simultaneous Autoregressive (SAR)
38models were used to assess the relations between phylogenetic age differences and Quaternary-scale
39climate variability.

40Results We find that phylogenetic age differences overall were largest in China and smallest in Europe,
41and decline with increasing temperature instability as predicted, but only in Europe and North America.
42In China, the relatively mild Quaternary climate changes do not appear to have strongly affected age
43differences in tree assemblages.

44Main conclusion

45Our results show that phylogenetically diverse assemblages with large phylogenetic age differences
46among species are associated with relatively high long-term climate stability, Our results indicate that
47the impact of with intra-regional links between long-term climate variability on theand phylogenetic
48composition of tree assemblages is especially strong in the more most unstable regions, with weak or
49no effect in the most stable areas. These findingsrefore, point to future climate change as a key risk to
50the preservation of the phylogenetically diverse assemblages in regions characterized by relatively high
51paleoclimate stability, with China a key examplethe rich flora in China, preserved through the region’s
52relatively high long-term climate stability, but now facing unprecedented threat by the rapid increase of
53human activities, needs special attention for future biodiversity conservation.
54Introduction

55The causes of the gradual decrease in species richness from the equator towards the poles, is a central
56question in biogeography and macroecology (Willig et al., 2003; Hillebrand, 2004; Jansson & Davies,
572008; Rolland et al., 2014). Speciation, extinction and dispersal are the fundamental processes
58influencing species richness and isare expected to vary along latitudes and among continents. Regions
59with long-term climate stability, i.e., glacial refugias, should have higher species diversity due to recent
60relatively high speciation rates and as well as the preservation of ancient species from climate-driven
61extinctions (Tzedakis et al., 2002; Mckenna & Farrell, 2006; Moreau & Bell, 2013).

62THence, these regions with higher speciation and lower extinction rates are expected to have include
63members of both basal and crown lineages of a phylogeny. Therefore, phylogenetic age differences
64between the basal and crown lineages are predicted to be relatively larger in climatic stable areas than
65anywhere else (Pellissier et al., 2014). Previous studies of this hypothesis have focused on animals such
66as leaf beetles, ants, fishes and marine bivalves (Jablonski et al., 2006; Mckenna & Farrell, 2006;
67Moreau & Bell, 2013; Pellissier et al., 2014). In contrast, no few studies has have assessed this
68hypothesis for plants at global scale or across different continents.

69Although tree assemblages in China, North America, and Europe have similarities in composition
70(notably shared genera and families) and structure, China has the highest species richness among the
71three regions (Latham & Ricklefs, 1993; Fang et al., 2012). The relative stable climate during the
72Neogene and Quaternary glacial-interglacial cycles in East Asia has been suggested to be one important
73factor behind this diversity discrepancy, notably in contrast to Europe (Qian & Ricklefs, 1999; López-
74pujol et al., 2006). However, few studies have directly connected paleoclimatic stability with tree
75assemblage composition to examine its importance for the diversity discrepancy among these regions.
76Northern Hemisphere regional tree floras were strongly affected by the late Neogene and Quaternary
77cooling and drying (Svenning, 2003; Eiserhardt et al., 2015), with especially the former leading to
78phylogenetically clustered survivorship via cooling filtering on phylogenetically conserved cold
79tolerance (Eiserhardt et al., 2015). It is less clear to what extent speciation in Northern Hemisphere
80trees has been affected by Quaternary climate, as most tree species in the region appear to be older
81(Willis & Andel, 2004; López-pujol et al., 2006; Kerkhoff et al., 2014); nevertheless, as similar
82dynamics have been going on since the Neogene (e.g., Willis et al., 1999), there may be deeper time
83correlations to Quaternary patterns in climate stability.

84In this paper, we study tree assemblages in China, North America and Europe to test the hypothesis that
85long-term climate stability drives the emergence of species assemblages with large species age
86differences due to the accumulation of both relict species and/or newly arisen species via reduced
 87extinction and increased speciation. In terms of climate stability we focus on Quaternary climate, given
88the massive impact of the Pleistocene glaciations on Northern Hemisphere tree floras (Davis, 1983;
89Svenning, 2003), using the LGM-present shift to represent the general pattern of glacial-interglacial
90climate shifts across the period (Jansson, 2003). We address the following predictions: (1) Comparing
91the three regions, overall assemblage phylogenetic age differences should be greatest in the most stable
92region (China) and smallest in the least stable region (Europe) (Qian & Ricklefs, 1999; López-pujol et
93al., 2006). (2) Within regions, phylogenetic age differences of tree assemblages should be negatively
94correlated with LGM-present climate change, with the weakest relation in the region most mildly
95affected by Quaternary climate change, i.e., China.

96Materials and Methods

97Tree distribution data. In this study we focused on tree species with a maximum canopy height ≥10 m
98because of data availability. Gymnosperm species were not included in our study because of its effect
99on phylogenetic age differences. The study covered three regions: China, North America and Europe.
100Tree distribution data for China came from the Chinese Vascular Plant Distribution Database, which is
101mainly compiled from Flora Republicae Popularis Sinicae (Delecti Florae Reipublicae Popularis
102Sinicae Agendae Academiae Sinicae, 1959-2004), provincial and regional floras. Tree distribution data
103for North America was compiled in the Atlas of United States Trees (Little, 1971-1978), digitized and
104available at the USGS Geosciences and Environmental Change Science Center. Tree distribution for
105Europe was compiled from the Atlas Florae Europaeae (AFE; Jalas & Suominen, 1994–1999),
106combined with the range maps used in Montoya et al. (2007), and supplemented with expert
107information on the distribution of Acer sempervirens and Malus trilobata (personal communication
108Arne Strid). All the distribution maps where converted to Lambert azimuthal equal-area projection and
109rasterized to a 100 × 100 km2 resolution. For our analyses we only used grid cells with more than 50%
110of land area. Finally, tThe study covered 2,154 tree species in China, 159 in Europe and 367 in North
111America.

112Phylogeny. To estimate tree species age, we used a time-calibrated species-level phylogeny covering
11331,749 plant species (Zanne et al., 2014; Qian & Jin, 2015). This phylogeny was generated based on
114seven gene regions and includes 98.6% families as well as 51.6% genera of all extant seed plants in the
115world (Zanne et al., 2014). Because species age derived from terminal branch length could be sensitive
116to the algorithum of adding species to the base tree, we randomly (100 times) added the species not
117included in the phylogeny within their relevant genera and families, producing 100 phylogenies as a
118null comparison. The mean terminal branch length of each species from the 100 random phylogenies
119was used to represent the species age.
120Environmental variables. Contemporary climate was characterized by mean annual temperature
121(MAT) and mean annual precipitation (MAP), which were downloaded from WorldClim (Hijmans et
122al., 2005). Altitude range per grid cell was computed based on an digital elevation model available in
123the same source. Temperature and precipitation during the Last Glacial Maximum (LGM) were
124extracted from the Community Climate System Model version 3 (CCSM3; Hijmans et al., 2005; Otto-
125Bliesner et al., 2006) and the Model for Interdisciplinary Research on Climate version 3.2 (MIROC3.2;
126Hasumi & Emori 2004). LGM temperature and precipitation were then summarized as the mean of the
127two models. LGM-present anomalies (contemporary climate minus LGM climate) in MAT and MAP
128and LGM-present velocities in MAT and MAP were used to directly represent paleoclimate change.
129Velocity is the ratiodescribes the speed of movement of climate across the time period evaluated,
130estimated by between relating the rate of temporal climate change and to the rate of the local spatial
131gradient in climate change (Loarie et al., 2009). All environmental variables were processed in ArcMap
13210.1.

133Statistical Analyses. Tree species in each grid cell were sorted according their ages and then were
134equally divided into three equally-sized age groups. The difference in median age between the oldest
135and youngest group was then used to represent the age differences, to use a metric that should be
136relatively robust to the coarse resolution of the available phylogeny. For each random estimated
137phylogeny, we got a phylogenetic age difference in each grid cell. We , and then used the mean of the
138phylogenetic difference ofestimated based on each of the 100 restimatedandom phylogenies in each
139grid cell was the final phylogenetic age difference for thisa given grid cell. Environmental variables
140that were not normally distributed were log-transformed. Finally, all variables were standardized to
141allow the estimation of standardized regression coefficients.

142Multiple comparisons of phylogenetic age differences among the three regions was were computed
143using Tukey’s HSD test. Relations between phylogenetic age differences and the environmental
144predictors were estimated by Simultaneous Autoregressive (SAR) models to account for spatial
145autocorrelation. To find the best combination of variables associated with phylogenetic age differences,
146we did multiple linear regressions. Because the climate change velocity in climate is highly correlated
147with anomaly in climate and altitude range, we divided the models into two groups: group 1 includes
148MAT, MAP, velocity in MAT and velocity in MAP; group 2 includes MAT, MAP, anomaly in MAT,
149anomaly in MAP and altitude range. In each group, SAR models for all possible combinations of
150variables were run with SAR models., and Akaike’s Information Criterion (AIC) was used to choose
151the best combination of predictors for each group. All analysis were done in R (version 3.1.1, R Core
152Team).
153Results

154Tukey’s HSD test showed that phylogenetic age differences in tree assemblages in Europe were smaller
155than in North America, which again were further smaller than in China (Fig. 1, p < 0.01). Single
156Single-variable Simultaneous Autoregressive models showed that the LGM-present anomaly in
157temperature was one of the two variables most associated with phylogenetic age differences in both
158Europe and North America (Fig. 1, 2; Table S1, p < 0.05 in Europe and < 0.01 in North America). In
159contrast, the two variables most associated with phylogenetic age differences in China are were the two
160contemporary climate variables (Fig. 1, 2; Table S1, p < 0.01 for mean annual temperature). Ignoring
161significance levels, all associations with paleoclimate instability tended to be negative and all
162associations with contemporary temperature and precipitation tended to be positive. This indicates that
163phylogenetic age differences tend to be larger in regions areas with a more stable paleoclimate in
164Europe and North America, and in regions with higher contemporary temperature and more
165precipitation in China (Fig. 2; Table S1).

166The multiple regressions provided similar results, indicating the important role of paleoclimate in North
167America and Europe (although the standardized coefficients in Europe were not significant), and the
168important role of contemporary climate in China in shaping the phylogenetic age differences (Table 1).
169Notably, the best combination of variables includes both contemporary and paleoclimate in North
170America, emphasizing the complementary effect of these two sets of variables (Table 1).

171Discussion

172Long-term climate stability is hypothesized to drive the emergence of species assemblages with large
173species age differences due to the accumulation of both relict species and newly arisen species via
174reduced extinction and increased speciation (Fjeldså & Lovett 1997; Dynesius & Jansson 2000). Our
175results provide overall support for this idea. Notably, we found that phylogenetic age differences are (1)
176smaller in Europe than in the other regions consistent with stronger climate-driven extinction in this
177more climatically unstable region; (2) negatively correlated with LGM-present climate change in
178Europe and North America; (3) not correlated with LGM-present climate change, but linked with
179contemporary climate in China. These links to climate instability in Europe and North America are
180consistent with the hypothesis that phylogenetic age differences are larger in areas with stable climate
181due to the co-occurrence of both ancient and new species, preserved from climate-driven extinction
182and/or promoted by recent speciation, respectively. The weak link to climate instability in China is
183consistent with the relatively minor effect of glacial-interglacial climate change on the Chinese flora.
184We acknowledge that there is limited evidence for tree speciation during the Quaternary, so the
185geographic variability in phylogenetic age differences must primarily reflect differential extinction
186rates (with the extinctions causing phylogenetic filtering, cf. Eiserhardt et al. 2015). Still, relative
187recent (but likely mostly pre-Quaternary) speciation may also play a role, if areas that were stable
188during Quaternary were also relatively stable on longer time scales. Additionally, newly formed species
189with small ranges would have been more susceptible to climate-driven extinction during the
190Quaternary, which may also contribute to the pattern.

191Climate change and trees in Europe

192Using 10 meters in maximum canopy height as a cutoff, Europe only has 159 tree species, much less
193than North America (367 species) and China (2154 species). A key factor in this diversity anomaly is a
194much greater extinction rate among trees in Europe during the late Neogene and Quaternary due to a
195stronger cooling and drying trend in this region (Davis, 1983; Qian & Ricklefs 1999; Eiserhardt et al.,
1962015). Phylogenetic structure of the temperate tree flora in Europe is more clustered than North
197America and China, because of the stronger phylogenetically selective extinction in Europe (Eiserhardt
198et al., 2015).

199Previous studies have shown strong evidence of glacial cooling leading to strong range contractions in
200surviving tree species in Europe, albeit a subset of the more hardy species remained relatively
201widespread (Willis & Andel, 2004; Svenning et al., 2008). Although less severe, many tree species
202have also exhibited range contractions during glacial maxima in North America (Ordonez & Williams
2032013; Bell et al., 2014) and East Asia (Ni et al., 2014; Zhao et al., 2016). These dynamics still shape
204diversity patterns within the regions, affecting endemism (Svenning et al., 2007; Feng et al., 2016),
205phylogenetic structure (Ma et al., 2016), and even functional diversity (Ordonez & Svenning, 2015,
2062016).

207Extending these findings, our results here showed that phylogenetic age differences in Europe are
208relatively smaller than North America and China (Fig. 1.), in line with the stronger impact of
209Quaternary climate change on tree diversity in Europe due to the selective loss of phylogenetic clades
210(Eiserhardt et al. 2015). Furthermore, the two variables most associated with phylogenetic age
211differences in Europe are were LGM-present anomaly in temperature and precipitation (Fig. 1 and 2),
212again emphasizing the role of Quaternary climate change in this region in shaping current diversity
213patterns.

214Climate change and trees in North America

215Although the similarities in flora between North America and China have been widely reported in the
216past decades, plant species richness in China is much higher than in North America (Wen, 1999; Milne
2172006; Fang et al., 2012). One of the drivers for this pattern is the more extensive continental ice sheet
218during Quaternary in North America, leaving albeitsouth-eastern and southern parts of western North
219America functioned as glacial refugia for many plant species (Milne 2006; Eiserhardt et al., 2015).
220Several studies have found that current tree diversity patterns in North America still partially reflects
221this dynamic. For example, accessibility to recolonization from glacial refugia and current climate
222codetermines the functional diversity pattern of broad-leaved trees across North America (Ordonez &
223Svenning, 2016). Further, there is a more clustered phylogenetic structure of temperate trees in western
224North America than eastern North America, linked to the more severe climate-driven extinctions in the
225former region (Eiserhardt et al., 2015). Still, glacial refugia in North-west North America along the
226coast have high local plant diversity as well as several endemic species, compared with inland glacial
227regions (Gavin 2009). In addition, phylogenetic diversity of North America angiosperms is also
228correlated with current climate, i.e., minimum temperature, with more clustered phylogenetic structure
229in colder regions (Qian et al., 2013).

230In line with these studies, the best combination of variables for phylogenetic age differences in North
231America includes both paleo- and contemporary climate (Table 1). Notably, the two variables most
232associated with phylogenetic age differences here are mean annual temperature and LGM-present
233anomaly in temperature (Fig. 1 and 2). In addition, the significantly negative associations between
234phylogenetic age differences and LGM-present climate change strongly support the hypothesis of the
235effect of long-term climate stability on phylogenetic structure of tree assemblages in North America.

236Climate change and trees in China

237Compared with Europe and North America, China was relatively mildly affected by the Last Glacial
238Maximum (Qian & Ricklefs 1999; López-pujol et al., 2006), which is consistent with our finding that
239the phylogenetic age differences in China is most associated with two contemporary climate variables,
240i.e., mean annual temperature and precipitation (Fig. 1 and 2). Species richness of trees in China is well
241linked to contemporary climate, especially temperature related variables (Wang et al., 2009, 2011).

242Like in other regions, there is evidence for the impact of Quaternary glacial-interglacial climate change
243on tree assemblage composition and structure in China, with effects on e.g.. endemism in woody plant,
244phylogenetic structure of forest tree communities, and spatiotemporal biome shifts (Ni et al., 2014;
245Feng et al., 2014, 2016). However, it is also clear that there were widely distributed glacial refugia for
246forests and trees not only in southern China, but also in northern and north-western China (Tian et al.,
2472009; Yang et al., 2009; Ma et al., 2012), likely providing a key explanation for the survival of many
248tree genera that went extinct in North America and Europe during the late Cenozoic and Quaternary
249(e.g., Eiserhardt et al. 2015). These dynamics are consistent with the weak, negative relation between
250phylogenetic age differences in China and LGM-present climate change variables found in the present
251study.

252Conclusions

253Our study extend the results from studies on the effects of long-term climate stability on phylogenetic
254structure for other organism groups and regions (Jablonski et al., 2006; McKenna & Farrell, 2006;
255Moreau & Bell, 2013; Pellissier et al., 2014) to show its applicability to the North Hemisphere tree
256floras. Notably, our results show that phylogenetically diverse assemblages with large phylogenetic age
257differences among species are associated with relatively high long-term climate stability, with intra-
258regional links between long-term climate variability and phylogenetic composition especially strong in
259the more unstable regions. These findings point to future climate change as a key risk to the
260preservation of the phylogenetically diverse assemblages in regions characterized by relatively high
261paleoclimate stability, with China a key example.The different patterns among Europe, North America
262and China further show that although long-term climate variability has left strong legacy on the
263hemisphere-scale assemblage phylogenetic composition pattern, the strength varies among different
264regions.

265Acknowledgments

266We thank the Chinese Scholarship Council(201204910187) for the support to GF. GF was also funded
267by the Starting Funding for Scientific Research from Inner Mongolia University (21400-5165111). J.-
268C.S. and A.O. were supported by the European Research Council (ERC-2012-StG-310886-
269HISTFUNC). B.B. was supported by Aarhus University and Aarhus University Research Foundation
270under the AU IDEAS program (via Centre for Biocultural History). Z.M. and J.-C.S. acknowledge
271support from the Danish Council for Independent Research | Natural Sciences (12-125079). J.-C. S.
272also considers this work a contribution to his VILLUM Investigator project “Biodiversity Dynamics in a
273Changing World”, funded by VILLUM FONDEN.

274Biosketch Gang Feng is interested in combing community ecology with macroecology and
275biogeography to assess the relative roles of local, regional and historical factors in shaping diversity
276patterns.

277References

278Bell, D.M., Bradford, J.B. & Lauenroth, W.K. (2014) Early indicators of change: divergent climate
279 envelopes between tree life stages imply range shifts in the western United States. Global
280 Ecology and Biogeography, 23, 168-180.
281Corlett, R.T. (2015) The Anthropocene concept in ecology and conservation. Trends in Ecology and
282 Evolution, 30, 36-41.

283Davis, M.B. (1983). Quaternary history of deciduous forests of eastern North America and Europe.
284 Annals of the Missouri Botanical Garden, 70, 550-563.

285Delecti Florae Reipublicae Popularis Sinicae Agendae Academiae Sinicae ed. (1959–2004) Flora
286 Reipublicae Popularis Sinicae. Beijing: Science Press.

287Dynesius, M. & Jansson, R. (2000) Evolutionary consequences of changes in species' geographical

288 distributions driven by Milankovitch climate oscillations. Proceedings of the National Academy
289 of Sciences, 97, 9115-9120.

290Eiserhardt WL et al. (2015). Climate-driven extinctions shape the phylogenetic structure of temperate
291 tree floras. Ecology Letters, 18, 263-272.

292Fang, J., Wang, X., Liu, Y., Tang, Z., White, P.S. & Sanders, N.J. (2012) Multi-scale patterns of forest
293 structure and species composition in relation to climate in northeast China. Ecography, 35,
294 1072-1082.

295Feng, G., Mi, X., Eiserhardt, W.L., Jin, G., Sang, W., Lu, Z., Wang, X., Li, X., Li, B., Sun, I., Ma, K. &
296 Svenning, J.C. (2015) Assembly of forest communities across East Asia-insights from
297 phylogenetic community structure and species pool scaling. Scientific Reports, 5, 9337.

298Feng, G., Mao, L., Sandel, B., Swenson., N.G. & Svenning, J.C. (2016) High plant endemism in China
299 is partially linked to reduced glacial-interglacial climate change. Journal of Biogeography, 43,
300 145-154.

301Fjeldså, J. & Lovett, JC. (1997) Geographical patterns of old and young species in African forest biota:
302 the significance of specific montane areas as evolutionary centres. Biodiversity and
303 Conservation, 6, 325-346.

304Gavin, D.G. (2009) The coastal-disjunct mesic flora in the inland Pacific Northwest of USA and
305 Canada: refugia, dispersal and disequilibrium. Diversity and Distributions, 15, 972-982.

306Hasumi, H. & Emori, S. (2004) K-1 coupled gcm (miroc) description. Center for Climate System
307 Research, University of Tokyo, Tokyo.

308Hawkins, B.A. & DeVries, P.J. (2009) Tropical niche conservatism and the species richness gradient of
309 North American butterflies. Journal of Biogeography, 36, 1698-1711.
310Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.J. & Jarvis, A. (2005) Very high resolution
311 interpolated climate surfaces for global land areas. International Journal of Climatology, 25,
312 1965-1978.

313Hillebrand, H. (2004) On the generality of the latitudinal diversity gradient. American Naturalist, 163,
314 192-211.

315Huang, J., Chen, B., Liu, C., Lai, J., Zhang, J. & Ma, K. (2012) Identifying hotspots of endemic woody
316 seed plant diversity in China. Diversity and Distributions, 18, 673-688.

317Jablonski, D., Roy, K. & Valentine, J.W. (2006) Out of the tropics: evolutionary dynamics of the
318 latitudinal diversity gradient. Science, 314, 102-106.

319Jalas, J. & Suominen, J. (eds) (1972-1994). Atlas Florae Europaeae: Distribution of Vascular Plants in
320 Europe. The Committee for Mapping the Flora of Europe and Societas Biologica Fennica
321 Vanamo, Helsinki, Finland.

322Jansson, R. (2003). Global patterns in endemism explained by past climatic change. Proceedings of the
323 Royal Society of London. Series B: Biological Sciences, 270, 583-590.

324Jansson, R. & Davies, T.J. (2008) Global variation in diversification rates of flowering plants: energy
325 vs. climate change. Ecology Letters, 11, 173-183.

326Kerkhoff, A.J., Moriarty, P.E. & Weiser, M.D. (2014) The latitudinal species richness gradient in New
327 World woody angiosperms is consistent with the tropical conservatism hypothesis. Proceedings
328 of the National Academy of Sciences, 111, 8125-8130.

329Lamanna, C., Blonderb, B., Violled, C. et al. (2014) Functional trait space and the latitudinal diversity
330 gradient. Proceedings of the National Academy of Sciences, 111, 13745-13750.

331Latham, R.E. & Ricklefs, R.E. (1993) Global patterns of tree species richness in moist forests: energy
332 diversity theory does not account for variation in species richness. Oikos, 67, 325–333.

333Little, E.L. 1971-1978. Atlas of United States trees. United States Dept of Agriculture Miscellaneous
334 publication.

335Loarie, S.R., Duffy, P.B., Hamilton, H., Asner, J.P., Field, C.B. & Ackerly, D.D. (2009) The velocity of
336 climate change. Nature, 462, 1052-1055.

337López-Pujol, J., Zhang, F. & Ge, S. (2006) Plant biodiversity in China: richly varied, endangered, and
338 in need of conservation. Biodiversity and Conservation, 15, 3983-4026.
339Ma, S., Zhang, M. & Sanderson, S.C. (2012). Phylogeography of the rare Gymnocarpos przewalskii
340 (Caryophyllaceae): indications of multiple glacial refugia in north-western China. Australian
341 Journal of Botany, 60, 20-31.

342Ma, Z., Sandel, B. & Svenning. J.C. (2016) Phylogenetic assemblage structure of North American trees
343 is more strongly shaped by glacial-interglacial climate variability in gymnosperms than in
344 angiosperms. Ecology and Evolution, doi:10.1002/ece3.2100.

345McKenna, D.D. & Farrell, B.D. (2006) Tropical forests are both evolutionary cradles and museums of
346 leaf beetle diversity. Proceedings of the National Academy of Sciences, 103, 10947-10951.

347Montoya, D., Rodríguez, M.A., Zavala, M.A. & Hawkins, B.A. (2007). Contemporary richness of
348 holarctic trees and the historical pattern of glacial retreat. Ecography, 30, 173-182.

349Moreau, C.S. & Bell, C.D. (2013) Testing the museum versus cradle tropical biological diversity
350 hypothesis: phylogeny, diversification, and ancestral biogeographic range evolution of the ants.
351 Evolution, 67, 2240-2257.

352Ni, J., Cao, X., Jeltsch, F. & Herzschuh, U. (2014) Biome distribution over the last 22,000 yr in China.
353 Palaeogeogr. Palaeogeography Palaeoclimatology Palaeoecology, 409, 33–47.

354Ordonez, A. & Svenning, J.C. (2015) Geographic patterns in functional diversity deficits are linked to
355 glacial-interglacial climate stability and accessibility. Global Ecology and Biogeography, 24,
356 826–837.

357Ordonez, A. & Svenning, J.C. (2016) Functional diversity of North American broad-leaved trees is
358 codetermined by past and current environmental factors. Ecosphere, 7, e01237.

359Ordonez, A. & Williams, J.W. (2013) Climatic and biotic velocities for woody taxa distributions over
360 the last 16 000 years in eastern North America. Ecology Letters, 16, 773-781.

361Otto-Bliesner, B.L., Brady, E.C., Clauzet, G., Tomas, R., Levis, S. & Kothavala, Z. (2006) Last glacial
362 maximum and Holocene climate in CCSM3. Journal of Climate, 19, 2526-2544.

363Pellissier, L., Leprieur, F., Parravicini, V., Cowman, P.F., Kulbicki, M., Litsios, G., Olsen, S.M., Wisz,
364 M.S., Bellwood, D.R. & Mouillot, D. (2014) Quaternary coral reef refugia preserved fish
365 diversity. Science, 344, 1016-1019.

366Pyron, R.A. & Wiens, J.J. (2013) Large-scale phylogenetic analyses reveal the causes of high tropical
367 amphibian diversity. Proceedings of the Royal Society B: Biological Sciences, 280, 20131622.
368Qian, H. & Jin, Y. (2015). An updated megaphylogeny of plants, a tool for generating plant
369 phylogenies, and an analysis of phylogenetic community structure. Journal of Plant Ecology,
370 doi: 10.1093/jpe/rtv047.

371Qian, H. & Ricklefs, R.E. (1999) A comparison of the taxonomic richness of vascular plants in China
372 and the United States. American Naturalist, 154, 160-181.

373R Core Team. 2014. R: a language and environment for statistical computing. Vienna (Austria): R
374 Foundation for Statistical Computing.

375Rolland, J., Condamine, F.L., Jiguet, F. & Morlon, H. (2014) Faster speciation and reduced extinction
376 in the tropics contribute to the mammalian latitudinal diversity gradient. PLoS Biology, 12,
377 e1001775.

378Shen, Z., Fei, S., Feng, J., Liu, Y., Liu, Z., Tang, Z., Wang, X., Wu, X., Zheng, C., Zhu, B. & Fang J.
379 (2012) Geographical patterns of community‐based tree species richness in Chinese mountain
380 forests: the effects of contemporary climate and regional history. Ecography, 35, 1134-1146.

381Stevens, R.D. (2011) Relative effects of time for speciation and tropical niche conservatism on the
382 latitudinal diversity gradient of phyllostomid bats. Proceedings of the Royal Society B:
383 Biological Sciences, 278, doi:10.1098/rspb.2010.2341.

384Svenning, J.C. (2003). Deterministic Plio-Pleistocene extinctions in the European cool-temperate tree
385 flora. Ecology Letters, 6, 646-653.

386Svenning, J.C. & Skov, F. (2005) The relative roles of environment and history as controls of tree
387 species composition and richness in Europe. Journal of Biogeography, 32, 1019-1033.

388Svenning, J.C. & Skov, F. (2007) Ice age legacies in the geographical distribution of tree species
389 richness in Europe. Global Ecology and Biogeography, 16, 234-245.

390Svenning, J.C., Normand, S. & Skov, F. (2009) Plio‐Pleistocene climate change and geographic
391 heterogeneity in plant diversity–environment relationships. Ecography, 32, 13-21.

392Tian, B., Liu, R., Wang, L., Qiu, Q., Chen, K. & Liu, J. (2009) Phylogeographic analyses suggest that a
393 deciduous species (Ostryopsis davidiana Decne., Betulaceae) survived in northern China during
394 the Last Glacial Maximum. Journal of Biogeography, 36, 2148-2155.

395Tzedakis, P.C., Lawson, I.T., Frogley, M.R., Hewitt, G.M. & Preece, R.C. (2002) Buffered tree
396 population changes in a Quaternary refugium: evolutionary implications. Science, 297, 2044-
397 2047.
398Wang, Z., Brown, J.H., Tang, Z. & Fang, J. (2009) Temperature dependence, spatial scale, and tree
399 species diversity in eastern Asia and North America. Proceedings of the National Academy of
400 Sciences, 106, 13388-13392.

401Webb, C.O. & Donoghue, M.J. (2005) Phylomatic: tree assembly for applied phylogenetics. Molecular
402 Ecology Notes, 5, 181-183.

403Webb, C.O., Ackerly, D.D. & Kembel, S.W. (2008) Phylocom: software for the analysis of
404 phylogenetic community structure and trait evolution. Bioinformatics, 24, 2098–2100.

405Wiens, J.J. & Donoghue, M.J. (2004) Historical biogeography, ecology and species richness. Trends in
406 Ecology and Evolution, 19, 639-644.

407Willig, M.R., Kaufman, D.M. & Stevens, R.D. (2003) Latitudinal gradients of biodiversity: pattern,
408 process, scale, and synthesis. Annual Review of Ecology, Evolution, and Systematics, 34, 273-
409 309.

410Willis, K.J., Kleczkowski, A. & Crowhurst, S.J. (1999). 124,000-year peridicity in terrestrial vegetation
411 change during the late Pliocene epoch. Nature, 397, 685-688.

412Willis, K.J. & Andel, T.H.V. (2004). Trees or no trees? The environments of central and eastern Europe
413 during the Last Glaciation. Quaternary Science Reviews, 23, 2369-2387.

414Wang, J., Gao, P., Kang, M., Lowe, A.J. & Huang, H. (2009) Refugia within refugia: the case study of a
415 canopy tree (Eurycorymbus cavaleriei) in subtropical China. Journal of Biogeography, 36,
416 2156-2164.

417Zanne, A.E., Tank, D.C., Cornwell, W.K., et al. (2014). Three keys to the radiation of angiosperms into
418 freezing environments. Nature, 506, 89-92.

419Zhao, Y., Yan, X., Muir, G., Dai, Q., Koch., M.A. & Fu., C. (2016). Incongruent range dynamics
420 between co-occurring Asian temperate tree species facilitated by life history traits. Ecology and
421 Evolution, 6, 2346-2358.
422Tables

423Table 1. Simultaneous Autoregressive models of phylogenetic age differences in median value between
424the oldest and youngest group in each 100 × 100 km 2 grid cell against combinations of environmental
425variables in China (820 grid cells), Europe (502 grid cells) and North America (1609 grid cells).
426Standardized coefficients (Coef) for the model with the lowest AIC for a given variable group (see
427below) are given, as well as the Akaike weight (w) for each variable based on the full model sets per
428group. Model sets involved all possible combinations of associated variables, for two groups of
429variables: Group 1: mean annual temperature/precipitation (MAT/MAP), LGM-present MAT/MAP
***
430velocity. Group 2: MAT, MAP, altitude range (ALT), LGM-present MAT/MAP anomaly. p < 0.01, **
431p < 0.05, * p < 0.1.

　 China Europe North America

Coef w Coef w Coef w
Group 1
MAT 0.31*** 1 0.413 0.60*** 1
MAP 0.19** 0.792 -0.052 0.547 -0.15*** 0.997
VelocityMAT 0.293 0.368 -0.31*** 1
VelocityMAP 0.451 0.358 0.298
R2 0.635 0.707 0.737
AIC 1509.3 815.8 2425.5
Group 2
MAT 0.31*** 1 0.321 0.52*** 1
MAP 0.19** 0.854 0.407 -0.08** 0.775
ALT 0.298 0.327 0.08** 0.853
AnomalyMAT 0.472 -0.24* 0.609 -0.35*** 1
AnomalyMAP 0.278 -0.14* 0.646 0.275
R2 0.635 0.709 0.736
AIC 1509.3 　 813.4 　 2434.1 　
432
433Figures

434

435Figure 1. Maps of phylogenetic age differences in median value between the oldest and youngest
436group in each 100 × 100 km2 grid cell, and its two most related variables in China (820 grid cells),
437Europe (502 grid cells) and North America (1609 grid cells), which are mean annual temperature
438(MAT) and mean annual precipitation (MAP) in China, LGM-present anomaly of mean annual
439precipitation and temperature (MAP/MAT Anomaly) in Europe, and MAT and LGM-present anomaly
440of mean annual temperature in North America.
441

442Figure 2. Scatter plots of phylogenetic age differences in median value between the oldest and
443youngest group in each 100 × 100 km2 grid cell against their two most associated variables in China
444(CN, 820 grid cells), Europe (EU, 502 grid cells) and North America (NA, 1609 grid cells). Mean
445annual precipitation (MAP) in China and AnomalyMAT in Europe are log transformed. Linear regression
446fits are given.
447Supporting information

448Table S1. Simultaneous Autoregressive models of phylogenetic age differences in median value
449between the oldest and youngest group in each 100 × 100 km2 grid cell against each environmental
450variable separately, forin China (820 grid cells), Europe (502 grid cells) and North America (1609 grid
451cells). Standardized coefficients (Coef), pseudo R2 and AIC are given. MAT/MAP = mean annual
452temperature/precipitation, VelocityMAT/P = glacial-contemporary MAT/MAP velocity, ALT = altitude
453range, AnomalyMAT/P = LGM-present MAT/MAP anomaly. The two variables have the lowest AIC in
454each region are in bold. *** p < 0.01, ** p < 0.05, * p < 0.1.

China Europe North America

　 Coef R2 AIC Coef R2 AIC Coef R2 AIC

MAT 0.29*** 0.633 1512.4 0.06 0.706 816.7 0.41*** 0.731 2462.4
MAP 0.12 0.626 1526.9 -0.05 0.707 815.8 -0.04 0.726 2488.2
VelocityMAT 0.05 0.626 1527.1 0 0.706 817.2 -0.11** 0.727 2484.3
VelocityMAP -0.04 0.626 1527.7 -0.01 0.706 817.1 -0.05 0.726 2487.6
ALT -0.03 0.626 1527.8 -0.02 0.706 816.9 -0.03 0.726 2488.1
AnomalyMAT -0.05 0.626 1528.2 -0.29** 0.708 814.1 -0.38*** 0.729 2475.0
AnomalyMAP -0.03 0.626 1528.3 -0.16* 0.708 813.9 -0.03 0.726 2488.7