Climate-driven C4 plant distributions in China: divergence in C4 taxa

There have been debates on the driving factors of C4 plant expansion, such as PCO2 decline in the late Micocene and warmer climate and precipitation at large-scale modern ecosystems. These disputes are mainly due to the lack of direct evidence and extensive data analysis. Here we use mass flora data to explore the driving factors of C4 distribution and divergent patterns for different C4 taxa at continental scale in China. The results display that it is mean annual climate variables driving C4 distribution at present-day vegetation. Mean annual temperature is the critical restriction of total C4 plants and the precipitation gradients seem to have much less impact. Grass and sedge C4 plants are largely restricted to mean annual temperature and precipitation respectively, while Chenopod C4 plants are strongly restricted by aridity in China. Separate regression analysis can succeed to detect divergences of climate distribution patterns of C4 taxa at global scale.

There have been debates on the driving factors of C 4 plant expansion, such as PCO 2 decline in the late Micocene and warmer climate and precipitation at large-scale modern ecosystems. These disputes are mainly due to the lack of direct evidence and extensive data analysis. Here we use mass flora data to explore the driving factors of C 4 distribution and divergent patterns for different C 4 taxa at continental scale in China. The results display that it is mean annual climate variables driving C 4 distribution at present-day vegetation. Mean annual temperature is the critical restriction of total C 4 plants and the precipitation gradients seem to have much less impact. Grass and sedge C 4 plants are largely restricted to mean annual temperature and precipitation respectively, while Chenopod C 4 plants are strongly restricted by aridity in China. Separate regression analysis can succeed to detect divergences of climate distribution patterns of C 4 taxa at global scale.
Modern ecosystems, such as tropical savannas, temperate grasslands and semi-deserts, have a significant component of C 4 plants 1 . At global scale, only about 3% of total plant species is characterized by C 4 photosynthetic pathway, C 4 plants, however, account for roughly 25% of global terrestrial primary production, including important crops, weed plants and potential biofuels 2,3 . Understanding the occurrence and distribution of C 4 biota can yield important information regarding to global primary productivity and to the effects of climate changes on ecosystem structures and functions 2,4,5 , as well as C 4 plant's past, present and future.
The abundance of C 4 species in particular regions and their distribution in relation to climate have been well reported in North America 6 , Africa 7,8 , Europe 9,10 , Australia 11 , Middle East 12,13 , but has not been studied details in China and this knowledge is essential for formulating generalization regarding to global C 4 occurrence and their relation with climate. The vast area and varied terrain in China with complex ecosystem components (e.g. rain forests, wet lands, temperate grasslands, deserts and tundra) and great climate changes contain more different C 4 information. Moreover, the deserts in China and Asia differ markedly from the arid ecosystems of North America, Australia and Europe in the taxonomic groups of C 4 species 5,14 . In China deserts and arid regions, Chenopodiaceae is the leading C 4 family, but their distribution in relation to climate has not yet been addressed, this is very important for understanding the effects of climate changes on ecosystem structures and functions, particularly with the increasing of desertification in west China in recent decades 15 .
Although the occurrence and distribution of C 4 plants have been documented at different scales over the past couple of decades, there have been debates on C 4 plant expansion at large-scale 1,16-19 , for example, (i) what is the driving factor for C 4 plant expansion, decrease in atmospheric CO 2 concentration in the late Miocene or climate (both ancient and modern) variability 19−21 ? It had been hypothesized that PCO 2 decline caused C 4 plant expansion rapidly during the late Micocene (~8 to 4 Ma) 1,16,17 , but some evidences suggested that C 4 plant expansion was likely driven by addition factors, such as enhanced low-latitude aridity, seasonal precipitation 19 and fire 20 in the Miocene. The present-day global distribution of C 4 grasses is largely restricted to warmer climate and precipitation, for strong positive relationships between C 4 grass abundance and growing season temperature at continental scales 5,18 . Few found that the restriction of C 4 grasses to warmer areas was due largely to their evolutionary history 2 . (ii) whether the different C 4 taxa have similar climate distribution pattern in present-day at large-scale 21,22 ? Indeed, there are few large data sets with which to examine occurrence and climate distribution pattern of different C 4 taxa in modern vegetation at continental areas, resulting severely limits the accuracy understanding C 4 plant expansion and the ecological implications.
(19.67%) and Chenopodiaceae (17.52%; Chenopod, hereafter). Relative lower C 4 plant occurrence is due largely to there is no tropical savannas (with more C 4 grasses) in China. In general, total C 4 species abundance decreases from south to north and from east to west in China (Fig. 1a,b). The total C 4 species abundance in Heilongjiang (most northern territory) is only 1/3 of that in Yunnan (most southern territory), while that in western province of Qinghai is less than 1/3 of that in Taiwan (Fig. 1b). The total C 4 species abundance is strongly and positively related with mean annual temperature (T m ) (R 2 = 0.56, P < 0.0001) and mean annual precipitation (P m ) (R 2 = 0.47, P < 0.0001; Fig. 2a-c). Multiple regression of the total C 4 species abundance (Y totalC4 ) against climate variables shows that there is a strongly and positively correlation between Y totalC4 and T m , P m and aridity (A I ) as model: This indicates that these climate factors affect the distribution of total C 4 species abundance in China.
Stepwise multiple regression analysis exhibits that T m has highest contributions (61.5%) to total C 4 species distribution, while the impacts of P m (2.0%) and A I (2.6%) are relative less ( Table 2). C 4 /C 3 proportion in China flora is about 1.2%, ranging from 0.85% in Tibet to 4.77% in Shandong province (Fig. 3). Most humid southern provinces (e.g. Yunnan, Guangxi and Sichuan) have lower C 4 /C 3 , even though the occurrence of C 4 species is high in these regions. There are no significant relations between C 4 /C 3 proportions and climate variables in present-day vegetation in China (P > 0.05; Fig. 2d-f), indicating that C 4 /C 3 proportion dose not exhibit certain ecological pattern, even the C 4 occurrence dose significantly related with plant abundance at large-scale region. C 4 distribution patterns predicted by total C 4 species abundance appear to be insensitive to climate factors known to influence C 4 occurrence and expansion because of the different adaptive strategies for C 4 taxa to climate variables. Both grass and sedge C 4 species abundances are strongly and positively related with P m (R 2 = 0.51, P < 0.001; R 2 = 0.58, P < 0.001) and T m (R 2 = 0.50, P < 0.001; R 2 = 0.65, P < 0.001), but significantly and negatively with aridity (R 2 = 0.23, P < 0.01; R 2 = 0.55, P < 0.001) ( Fig. 4a-c). Multiple regressions of grass C 4 species abundance (Y grassC4 ) and sedge C 4 species abundance (Y sedgeC4 ) against climate variables manifest that Y grassC4 and Y sedgeC4 are strong correlated with T m , P m and A I as models: sedgeC4 m m I But stepwise multiple regression analysis demonstrates that grass C 4 plant distribution is largely restricted to P m (50.6%), and T m and A I functions are not significant (P > 0.05). Sedge C 4 species is mainly limited to T m (64.7%) and the impacts of P m (0.2%) and A I (2.3%) are very less and no significant (P > 0.05; Table 2). On the contrary, Chenopod C 4 plant abundance is strongly and positively related with A I (R 2 = 0.88, P < 0.001), and significantly and negatively with both P m (R 2 = 0.92, P < 0.001) and T m (R 2 = 0.25, P < 0.001; Fig. 4a-c). Stepwise regression of Chenopod C 4 plant abundance against climate variables shows that there is a strong correlation between Y Chenopod C4 and T m , P m and A I as model: − .

Discussion
There were many studies on the C 4 plant expansion and distribution in relation to climate over the past couple of decades 1, [16][17][18][19] , but only few manifested the detail floristic data of C 4 occurrence in large regional scale (e.g.       of Chenopod C 4 species is only 1/3 of that in Middle East and 1.5 times of that in Mongolia 14 , probably because China arid regions is smaller than Middle East, but larger than Mongolia. This knowledge is essential for building global C 4 plant database and formulating generalization regarding their relation to global climate. What is the driving factor for C 4 plant expansion and distribution remains controversial. The evidences of palaeovegetation and fossil tooth enamel indicated the global expansion of C 4 plants may be related to lower PCO 2 in the Miocene 1,16,17 , but some evidences suggested that C 4 plant expansion was likely driven by climate variables and fire in both old world and present-day vegetation 19,20 . Our data clearly demonstrate that C 4 plant distributions are restricted to mean annual climate variables (e.g. T m , P m and A I ) in the present-day vegetation in China (Fig. 2). From the south to the north, T m governs the vegetation changes, while from the east to the west moisture gradient (P m ) drives plant distributions 23,24 . Within China, the total C 4 species abundance is strongly and positively related with T m and P m , this suggests that there is remarkably strong tendency for C 4 species to grow in hot and wet conditions, even though the stepwise multiple regression analysis exhibits that the impact of P m is relative less ( Table 2). This is much different with previous observations 6,11,25 , their evidences manifest that July average daily temperature is a critical factor for C 4 distribution and the precipitation gradients seem to have much less impacts. Such significant difference is probably because almost 2/3 identified C 4 species are perennial grasses, sedges and some Polygonaceae species, relative higher P m and T m are not only favor for their growth in growing seasons, but also for their survival in winters 15,23,24 . This is also partly supported by the observation of soil organic carbon and present vegetation which indicates that C 4 fraction of Inner Mongolia grassland has increased by approximately 10% in the past decades because of increasing of temperature 26 .
The C 4 /C 3 proportion in China flora is about 1.2% and much lower than 3% estimated at global scale 4,18 . This is mainly due to complex relief in China, 2/3 of the total area is mountains and plateaus 23 . More mountains and high moisture in southern provinces lead to vast forest vegetation with relative more tree species and lower fractions of grasses and Chenopod species, even the species abundances for both C 3 and C 4 plants are much high in the southern regions. In addition, lower C 4 plant occurrence in China is for the absence of tropical savannas, which is estimated with more C 4 grasses 1 , but large area of temperate grasslands and deserts (40% of China land) devotes considerable C 4 plant resources 5,15 . There are no significant relations (P > 0.05) between C 4 /C 3 proportions and climate variables in present-day vegetation in China ( Fig. 2d-f), indicating that C 4 /C 3 proportions do not show certain ecological pattern (Fig. 3), even though C 4 occurrence dose significantly related with plant abundance at large-scale region.
Separate analysis for different C 4 taxa succeed in detecting C 4 distribution patterns accurately at continental scale, for the separate analysis can eliminate the noise signal from the C 4 taxon with different responses to climate variables and adaptive strategies. Previous studies had found that grass and sedge C 4 plants were largely restricted to July average daily temperature 6,11,18,25 . However, it is different in China, separate multiple regression analyses display that the grass and sedge C 4 plant abundances are mainly restricted to P m and T m respectively ( Table 2). Most grass C 4 species are terrarium plants and their distributions are mainly restricted to P m , while almost all sedge C 4 species are aquatic plants and the distributions of sedge C 4 taxon is governed by T m in China. This explanation is also supported by the relative higher proportions of perennial C 4 grasses and sedges in the floristic data (Supplementary Table S1). Unlike the grass and sedge C 4 plants, the distribution of Chenopod C 4 taxon is drove by aridity (Table 2; Fig. 4). Chenopod C 4 plants are favored arid regions with hot summer and sufficient summer precipitation, because most of Chenopod C 4 plants are annual plant species, they can use seasonal precipitation efficiently in dry and hot conditions where the precipitation mainly falls in growing season, and these species can withstand severe droughts as seeds 15 . Even though there are a few studies on the occurrence of C 4 Chenopods in particular regions 13,14 , but C 4 Chenopod distribution in relation to large-scale climate change has remain largely unexplored and this is important for understanding the effects of climate changes on arid ecosystems with the increase of desertification in west China 15 . In the dry west China (e.g. Xinjiang, Qinghai and Inner Mongolia), 30-45% of the total C 4 species is Chenopod plants, while that in the east and south China is less than 1%. Predominant C 4 Chenopods in hot and arid ecosystems, as well as strong relations with aridity (Fig. 4), imply that the expansion of C 4 Chenopods may be enhanced in China with an increase in hot and aridity worldwide as some climate-change scenarios suggested. Moreover, previous studies proved that the advantage of C 4 plants in water-limited deserts are not considered critical for establishing C 4 grass distribution pattern, and are commonly invoked to explain the dominance of C 4 dicots 21,27 , even though they did not separate Chenopod C 4 species from dicots.
Driving factor for C 4 plant expansion at spatial and temporal large-scale is controversial. The palaeosol carbonate and fossil tooth enamel data implicated that the C 4 plant expansion may have been due to decreasing of PCO 2 in late Miocene 1 , but other evidences suggested that the development of low-latitude season aridity and changes in growing conditions led to the expansion of C 4 plants at ~7 Ma 19 . These different explanations are mainly due to the lack of direct evidences and extensive data analysis. Our mass flora data analysis partly supports Pagain's perspective 19 , but their evidences also can not explain the divergence distribution pattern of different C 4 taxa. The divergence in C 4 climate pattern implicates these C 4 taxa may be with different area of origins, evolutionary histories 2,21 , expansion mechanism and adaptive strategies, because the Chenopod C 4 taxon has a diametrically opposed distribution pattern with grass and sedge C 4 taxa (Fig. 4). In the previous studies 6,18,25 , it had been found that grass and sedge C 4 plants are governed by July average daily temperature, but the distribution of grass, sedge and Chenopod C 4 species in China are largely restricted to P m , T m and A I respectively, for the mean annual climate variables (especially P m and T m ) can accurately describe the climate restrictions of plant distributions in China 23,24 . Edwards and Still also proved that the restriction of C 4 grasses to warmer areas was due largely to their evolutionary history 2 .
Comparing with most previous researches 1, [16][17][18][19] this work provides detail floristic data of C 4 occurrence in large regional scale based on China flora sources, which is essential for building worldwide C 4 plant database, and also contributes direct evidence formulating generalization regarding the driving factors of C 4 plant expansion. We suggest that the restriction of C 4 distributions at continental scale is due to largely the annual climate variables (e.g. P m , T m and A I ) in present-day ecosystems in China. Different C 4 taxa may exhibit diametrically opposite pattern in relation to climate at large-scale likely due to their differences in adaptations, area of origins and evolutionary histories 2,21 . Our findings suggest that the expansion of C 4 Chenopods will increase with the increasing of aridity in western and central China as climate-change scenarios expected, on the contrary, that for grass and sedge C 4 species may decrease in the future. This may have huge impacts on vegetation dynamics and primary plant production for the C 4 plants accounts for roughly 1/4 of global terrestrial primary production 2,3 .

Methods
China topography and climate. China, occupied a large area, about 9.6 million km 2 (3°51′ -53°33.5′ N; 73°33′ -135°05′ E), stretches 5,026 km across the East Asian landmass. It is primarily mountains, plateaus and plains country, 2/3 of the total area is mountains and plateaus 23 . Land elevation in the east plains is about 100-200 m above sea level (as l), while that in the southwest mountains and plateaus are as high as 4000-8000 m as l. The relief is very complicated with both latitudinal and longitudinal climate zones, mixed with steep altitudinal gradients in the northwest and southwest parts, leading great changes in climate.
The climate in China is extremely diverse due to its wide coverage, assortment of terrains and distances to the sea for different locations. Most of China lies in the temperate belt, with its south in subtropical belt and north in subarctic belt. In general, the average temperature in China is 11.8 °C, varying from 31 °C in July to − 10 °C in January. Because of the Influences of both latitude and monsoon activities, temperatures vary a great deal, low temperature in winter is − 40°C in Mohe, the northernmost of China, while in hot summer temperature can be as high as 50 °C in Turpan basin, Xinjiang. The average annual precipitation is about 620 mm, ranging from 150-400 mm in the western deserts and semi-deserts to 500-800 mm in the central region and vast flat plains, and 800-1000 mm in the eastern and coastal areas. The main nature vegetation types include tropical rain forest, wet land, grassland, desert and tundra 23 . Obtaining C 4 taxon data and analysis. Local C 4 taxon data of 32 provinces and municipalities (Fig. 1a) were collected from the C 4 plant database of Plant Adaptation Strategy and Mechanism Group, Institute of Botany, CAS, Reipublicae Popularis Sinicae 23 , Catalogue of Life China and local flora sources 28-60 . Long-term (195028-60 . Long-term ( -2010 climate data were provided by the National Meteorological Information Center of China Meteorological Administration. C 4 /C 3 proportion refers to the ratio of C 4 species number to C 3 species number in local flora. Regressions of C 4 taxon (e.g. total C 4 species abundance, grass and sedge C 4 species abundances) against climate variables (e.g. temperature, precipitation and aridity) were performed using SPSS 17.0 in order to explain the distribution patterns of C 4 taxa accurately at global scale. Stepwise multiple regression analyses between C 4 taxon and climatic variables were used to quantify the critical restriction of C 4 taxon. All statistical analyses were performed using SPSS 17.0 (SPSS for Windows, Chicago, IL, USA).