| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Paleobotany |
2State Key Laboratory of Systematic and Evolutionary Botany, Institute of Botany, Chinese Academy of Sciences, Beijing, 100093, P. R. China; 4Department of Earth Sciences, The Open University, Walton Hall, Milton Keynes, MK7 6AA, UK; 5Senckenberg Research Institute and Natural History Museum, Senckenberganlage 25, 60325, Frankfurt, Germany; 3State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an, 710075, P. R. China; 6Graduate University of the Chinese Academy of Sciences, Beijing, 100039, P. R. China
Received for publication August 3, 2006. Accepted for publication February 7, 2007.
ABSTRACT
The reconstruction of the climate in the Miocene Shanwang basin is an important link in understanding past climate and environmental changes in East Asia. A recent study showed that the mean annual temperature (MAT) estimates derived from leaf margin analysis (LMA) and the Climate Leaf Analysis Multivariate Program (CLAMP) conflicted with and were remarkably lower than those estimated by the coexistence approach (CA). Overlapping distribution analysis (ODA), a new method introduced here, is used to reconstruct the Shanwang Miocene climate based explicitly on local plant distribution data and associated meteorological stations. The Shanwang flora (17–15.2 Ma) suggests a MAT of 10.9–14.5°C and a mean annual precipitation (MAP) of 1107.3–1880.0 mm. This result is closer to the values derived from CLAMP and LMA than that obtained by CA. This report is the first comprehensive intercomparison of foliar physiognomic and nearest living relative climate proxies in a Chinese context and provides important cross validation of results.
Key Words: climate • Climate Leaf Analysis Multivariate Program (CLAMP) • coexistence approach • leaf margin analysis • Miocene • overlapping distribution analysis • Shanwang basin
During the Tertiary, global climate displayed an overall cooling trend within which a series of warming and cooling cycles occurred (e.g., Zachos et al., 2001
). With concerns over the rate and pattern of future climate change now firmly at the center of the international political agenda, developing reliable climate proxies to quantify the rates, patterns, and consequences of past climate change on the Earth system over land is an urgent priority. The strengths and weaknesses of each proxy, new or existing, need to be appreciated and evaluated alongside the data they provide about the nature of past change. Land plants are an ideal climate proxy for nonmarine environments because their morphology and distribution is strongly climate dependent. With well-preserved plant fossil beds, the Miocene Shanwang basin has the potential to provide useful insights into the Miocene climate of East Asia as well as providing the context for comparative evaluation of several plant-based proxies.
As early as 1940, Hu and Chaney (1940, pp. 88–89) reported on the Miocene Shanwang flora and inferred that the climate was warm-temperate to subtropical based on the tolerances of similar living taxa. Subsequent studies gave more evidence to support this qualitative inference (Song et al., 1964
; Writing Group of Cenozoic Plants of China, 1978
; Wang, 1981
).
Recently, two popular quantitative methods were used to reconstruct the palaeoclimate at Shanwang (i.e., the Climate Leaf Analysis Multivariate Program [CLAMP; available at http://tabitha.open.ac.uk/spicer/CLAMP/Clampset1.html;Wolfe, 1993
] and the coexistence approach [CA; Mosbrugger and Utescher, 1997
]) (Sun et al., 2002
; Liang et al., 2003
). The results from the two methods were significantly different. The mean annual temperature (MAT) of 9.511.2°C estimated by CLAMP (Sun et al., 2002
) was lower than the MAT of 15.317.2°C estimated by CA (Liang et al., 2003
), and the difference exceeded the statistical uncertainties of each method.
The aim of this paper is to reconstruct the Miocene climate at the Shanwang basin based on data from local plant distributions and meteorological stations and to evaluate and compare the applicability of the current versions of foliar physiognomic methods (in this case, leaf margin analysis [LMA] and CLAMP), and methods based on the climatic tolerance of nearest living relatives (NLR), (CA and overlapping distribution analysis [ODA]) that we introduce here.
MATERIALS AND METHODS
The Shanwang basin (36°33' N, 118°44' E), approximately 700 m long and 500 m wide, runs NW-SE in Shandong Province, eastern China (Fig. 1). The basin (today about 250 m above mean sea level [a.m.s.l.]) lies within moderate topography ranging from 200 m to 400 m a.m.s.l. It was formed by a maar volcano on a basalt platform (C. S. Li et al., 2000
). Ponding of streams formed a small lake in Miocene times (F. L. Li et al., 2000
). Mammalian faunas co-occurring with the leaves have been assigned to MN5 (Qiu, 1990
; Steininger et al., 1996
) that equates to 17–15.2 Ma on the timescale of Berggren et al. (1995)
. K-Ar dates for the underlying basalt (Yang and Yang, 1994
) range from 18–16 Ma, while the upper basalt has been determined as 10–9 Ma (Wang and Jin, 1986
; Chen and Pen, 1985
; Zhu et al., 1985
). The Shanwang Formation consists of six units (Li, 1991
). The second unit is diatomaceous, contains abundant fossils, and is now further divided into 19 subunits (C. S. Li et al., 2000
).
|
; Writing Group of Cenozoic Plants of China, 1978
; Wang et al., 2003
). The distributions of the nearest living relatives of 107 species are known and were selected for the reconstruction of the Miocene climate and elevation (Appendix S1, see Supplemental Data accompanying the online version of this article).
Sun (2000)
and Sun et al. (2002)
who used the CLAMP methodology and Liang et al. (2003)
who used CA studied six subunits (15, 14, 13, 7, 5, and 4) within the 19 subunits of the Shanwang diatomite. For intercomparison purposes, the same records of fossil plants in these six subunits are employed here to reconstruct the palaeoclimate using LMA, CLAMP, CA, and ODA.
Palaeoclimate proxies based on fossil leaves and pollen may be divided into two types: those that are based on aspects of plant architecture constrained by environmental conditions (physiognomic approaches) and those based on the environmental tolerances of assumed living relatives (nearest living relative approaches). Each has specific advantages and disadvantages.
Foliar physiognomic techniques
Leaf margin analysis
There are two physiognomic techniques in common usage. The first of these is simple LMA first introduced by Bailey and Sinnott (1915
, 1916
) and more recently revisited by Wolfe (1979)
, Wing and Greenwood (1993)
, and Wilf (1997). LMA relies on the correlation that exists between the proportion of toothed vs. nontoothed (entire) woody dicot leaves in a given patch of stable (nonpioneer) vegetation and the mean annual temperature. In humid to mesic vegetation, the relationship is essentially a straight line, the slope and intercept of which differ between the northern and southern hemispheres.
The percentage of entire margined species (PEMS) was employed to predict MAT from the relationship shown by the following equation for the northern hemisphere based on southeastern Asia data: MAT = 1.141 + (0.306 x PEMS) as shown by Wolfe (1979)
and Wing and Greenwood (1993)
.
The MAT errors are calculated by Wilf's (1997) sample error equation:
|
| (1) |
).
The underlying cause of this relationship between margin form and temperature is poorly known, but it is likely to encompass the generation of turbulence and thinning of the boundary layer, gas exchange, water relations, transpiration, and photosynthesis, particularly early in the growing season (e.g., Mauseth, 1988
; Schuepp, 1993
; Royer and Wilf, 2006
). In reality, more than one factor is likely to influence the adaptive morphology of any leaf architectural characteristic (Spicer et al., 2005
). However, the advantage of LMA is that it is simple to compute. The disadvantage is that it only returns one climate variable: the mean annual temperature (MAT). More advanced multivariate physiognomic analyses return far more variables and confirm not only the strong correlation between leaf margin type and MAT (Spicer et al., 2005
), but also parameters associated with water availability.
CLAMP
A further development of the LMA introduced by Wolfe (1993)
is the so-called Climate Leaf Analysis Multivariate Program (CLAMP). As its name suggests, CLAMP extends LMA beyond its underlying simplistic assumption that a single leaf character (in this case the geometry of the leaf margin) is correlated with a single climate variable (the MAT). The CLAMP methodology is the most comprehensive foliar physiognomic technique currently available, and with the existing calibration data sets (PHYSG3AR and PHYSG3BR) (see the CLAMP website for details: http://tabitha.open.ac.uk/spicer/CLAMP/Clampset1.html, it is capable of yielding values for up to 13 palaeoclimate variables, although 11 cited here in the methods section are most commonly returned (see the CLAMP website for details). Inevitably, the multivariate nature of the technique means that the computational simplicity of univariate methods such as LMA is lost, but this cost is more than outweighed by the precision obtained across an array of temperature related parameters that include the length of the growing season and enthalpy (a property of the atmosphere useful in determining palaeoaltitudes; Forest et al., 1995
; Wolfe et al., 1998
; Spicer et al., 2003
). At the heart of the CLAMP technique is the assumption that foliar physiognomy is determined by the physical laws related to fluid flow, diffusion, and irradiance/heat balance and that consequent constraints over plant architecture are largely invariant, within certain limits, over the last 100 Ma. For any given situation there will be an optimum leaf architecture that satisfies a variety of constructional and environmental constraints whilst returning maximum efficiency, particularly with respect to photosynthetic productivity.
This environmental adaptation is achieved within the context of the capabilities imparted by the genome, itself the product of long-term natural selection. Non-adapted physiognomies fail to survive, and over time there is a degree of convergence largely independent of taxonomy. Of course, in reality, environmental conditions are always in a state of flux as is interplant competition, so adaptation is also a dynamic process. Because of this, congruence between foliar physiognomy and an inherently dynamic environment can never be perfect. Inevitably, this degrades precision.
In CLAMP, 31 physiognomic character states that encompass lobing, margin geometry, apex and base shape, and lamina size and shape are used. At each modern vegetation plot used to calibrate CLAMP, the full morphological range of leaves of at least 20 taxa of woody dicots, including shrubs and lianas, are scored for the 31 character states. This data array is then compared to a similar one composed of climate variables observed at each plot site. Whenever possible, 30-yr (or greater) climate averages are used, measured within 1 km of and at the same altitude as the plot site. Typically, 11 climate variables (MAT; warm month mean temperature, WMMT; cold month mean temperature, CMMT; length of the growing season, LGS; mean growing season precipitation, MGSP; mean monthly growing season precipitation, MMGSP; precipitation during the three wettest months, 3-WET; precipitation during the three driest months, 3-DRY; specific humidity, SH; relative humidity, RH; and Enthalpy) are correlated with the foliar physiognomic data using canonical correspondence analysis (ter Braak, 1986
) in the form of the program CANOCO v. 4 (see the CLAMP website for details). This particular multivariate statistical engine is used because it is robust to incomplete data (important when dealing with fossil material that may be missing some character states), makes no assumptions about the Gaussian distribution of variables and does not assume that the variables, either foliar or climatic, are independent of one another (which they obviously are not). No single architectural feature of a leaf or whole plant determines adaptive success (Lande and Arnold, 1983
), and so no single feature can be expected to correlate with a single climatic variable. Moreover, many interacting traits may influence fitness (Ackerly et al., 2000
).
The fossil leaves are scored for the 31 foliar character states in an identical manner to those living leaf forms making up the calibration data sets. The fossil data, lacking any accompanying climate data, are introduced into the analysis as passive samples, i.e., their inclusion does not disturb the structure of the multidimensional physiognomic space defined by the modern calibration samples. The relationship between modern leaf physiognomy and modern climate data define the positions of vectors, one for each climate variable, running through physiognomic space, and these vectors can be calibrated from the observed climate. The position of the fossil assemblage in physiognomic space, when projected normally on to the calibrated vector, yields the palaeoclimate prediction. In practice, second order polynomial regressions are plotted between the climate vector score as defined by coordinates in the first (greatest) four axes of variation and the matching observed climate variable. The standard deviation of the residuals about this line is used as a measure of the uncertainty of the climate estimates. The calibration data sets currently in use comprise either 173 sites from predominantly North America and Japan and include sites where some significant cold is experienced (data set PHYSG3AR) or 144 sites derived from within this data set where cold sites are excluded (PHYSG3BR). As such, these data sets are not relevant to tropical vegetation or sites in the southern hemisphere. Exceptions to this are cited in Kennedy et al. (2002)
.
In our analysis, we used the PHYSG3BR data set, the default values in CANOCO 4.0, and placed the results in a spreadsheet appropriate to PHYSG3BR as available from the CLAMP website.
Nearest living relative techniques
The NLR methodologies have had the longest use in palaeobotany. The principle was even applied in China as long ago as 1086 by Shen Kuo (Deng, 1976
). Unfortunately, this early application of the method fell foul of the technique's greatest source of potential error, i.e., misidentification, in that it equated the extinct sphenophyte Neocalamites to bamboo. Nevertheless NLR-based climate proxies are well established and seem to work well for Quaternary material and even for Neogene fossils for which only small amounts of evolutionary change are likely to have taken place. For Palaeogene and older material, the fundamental assumption of NLR techniques, that there has been little change in environmental tolerance within a lineage over time, is unlikely to be true. This effect may be minimized by using large numbers of taxa and eliminating those that prove to be inconsistent with the behavior of the majority. Where there is consistency and the environmental tolerances have the greatest overlap, there is likely to be most confidence in the environmental relationship. This is the basis of the coexistence analysis (CA) of Mosbrugger and Utescher, (1997) and of overlapping distribution analysis (ODA). All NLR techniques rely on correct identification of the nearest living relative, ideally to the species level. For calibration, it is assumed that the nearest living relative occupies the entire geographic area that has the climate to which it is adapted.
In Quaternary work, methods have been devised that seek palaeobiological analogues and, in some cases, weight taxa based on their modern specificity for particular conditions (e.g., Guiot, 1987
). This approach may be appropriate for quaternary studies, but such techniques become highly susceptible to evolutionary change in a few key taxa. In CA and ODA, all taxa have equal weight, but some may be excluded because they do not behave congruently due to competitive (evolutionary) change in an environmental range. This attribute makes them more appropriate for older floras, but as with all NLR techniques, not where significant evolutionary change is likely to have taken place. The exact historical time for which NLR methodology becomes inappropriate is difficult to define and will be dependent on biogeographic history. This work is part of an ongoing attempt to explore this issue.
An important aspect of CLAMP that sets it apart from NLR methods is that evolutionary dynamism is recognized and actually exploited: the basis for the technique is that there is evolutionary convergence of architectural adaptation to environmental variables. A major disadvantage of CLAMP is that is restricted to leaves; however, pollen has no known intrinsic link between preserved wall morphology and climate. This is particularly unfortunate because fossil pollen is far more ubiquitous than fossil leaves and any interpretations of palynological assemblages in environmental terms has to be conducted in an NLR context.
In contrast, the assumption underlying NLR approaches is that there is stasis in environmental tolerance despite a degree of observable taxonomic change. Rarely are the ancient taxa identical in all respects to the modern presumed descendant—a situation complicated by mosaic evolution (different organs evolving at different rates under different selection pressures) and the fact that fossil organs are isolated from one another and never preserved as a complete entity with all parts of the life cycle and genotype available for study. NLR methods require accurate identification, ideally to the species level, but often this is not possible. More often than not, and with increasing frequency as progressively older fossil assemblages are examined, only within-genus identification can be achieved. Given the wide environmental tolerance observed in many modern genera (the polar to equatorial occurrences of Salix provides a notable, if extreme, example), extinct species and genera pose particular problems for the NLR methodology. There is a strong argument that constructing a comparator data set of modern taxa based on climatic tolerances at the species level is unrealistic where fossils are concerned and imparts an overly optimistic suggestion of precision. A more appropriate calibration is at the genus level, but this inevitably provides for a much greater spread in climatic tolerances and hence lower precision.
Coexistence analysis
CA, introduced by Mosbrugger and Utescher (1997)
uses a calibration data set (CLIMBOT) containing over 800 Tertiary plant taxa that have modern relatives. The climatic tolerances of these modern relatives are included in their data set, as are their climatic tolerances with respect to 10 variables: MAT, WMMT, CMMT, MAP, RH, potential evaporation (PE), mean maximum monthly precipitation (MMaP), mean minimum monthly precipitation (MMip), mean precipitation in the warmest month (MWP), and aridity index (AI), which is defined as MAP/PE. Extreme values for these variables were obtained from six relevant climate stations and emphasis was placed on the inclusion of altitudinal variations.
A CA analysis is conducted by identifying the taxa of a given fossil flora to genus or species level. The corresponding NLRs are extracted from the CLIMBOT data base (or a modified version called PALAEOFLORA; Liang et al., 2003
) as well as the tolerance for any given climate parameter. A FORTRAN program CLIMST (referred to as CLIMSTAT in Liang et al. [2003
]) is then used to determine the interval of coexistence for the climate parameter and to calculate associated statistics. The procedure is run separately for each of the 10 climate parameters.
Overlapping distribution analysis
Here we introduce a method of determining palaeoclimate data from plant geographic and altitudinal distributions. As with all NLR techniques, the first step in ODA is to identify the fossil material in relation to the appropriate NLRs.
The complete process is as follows:
|
| (2) |
|
| (3) |
is the temperature lapse rate. These temperature lapse rates are obtained from the local meteorological data within the area of maximum overlap. Here they are taken as 0.5°C/100 m for MAT (Ni, 1997
; Ni and Song, 1998
), 0.45°C/100 m for CMMT, and 0.6°C/100 m for WMMT (Zhang and Lin, 1985
). From these, the maximum and minimum of the CMMT and WMMT and mean annual range of temperature (MART) within the zone of altitude overlap can be obtained.
The relative advantages and disadvantages of physiognomic and NLR techniques
NLR techniques are at their most robust when large numbers of taxa are involved because this minimizes the effects of incorrect identification and/or lack of appropriate modern relatives at the species level. Large comparator data sets are at the core of CA and ODA. Fortunately, both techniques are applicable not only to leaf assemblages, but also to pollen (and even wood) that have a higher preservation potential and can be sourced over a wider area. The restriction of LMA and CLAMP to locally derived leaf assemblages means that they may well yield data biased toward fluvial margin or lake shore microclimates, which are likely to be markedly cooler and more humid than regional conditions.
One way to completely eliminate the potential problems that bedevil NLR techniques (i.e., those of misidentification, lack of close living relatives, and changes in environmental tolerances within lineages over time) is to disconnect physiognomy from taxonomic identification. In CLAMP, there are no inherent assumptions of relationships between the fossil specimens and modern (or other fossil) taxa. Instead, all that is required is the partitioning of the fossil specimens into morphotypes that are considered to represent once living natural species. Nomenclature can be arbitrary, and there is no attempt to assign the morphotypes to living or extinct taxa. Moreover, uncertainty in morphotype assignment can be explored in successive analyses by experimentally assigning ambiguous specimens to several different morphotypes that display similarities. Variations in climate parameter predictions between these analyses can be added to the statistical uncertainties inherent in the calibration data sets. In the analyses that are the subject of this work, CLAMP morphotypes map on to taxa as recognized for LMA, CA, and ODA.
The use of CLAMP also largely overcomes the problem of a modern calibration species not fully occupying the potential climatic range it is capable of tolerating. CLAMP is not entirely immune to this phenomenon because the leaf morphologies used to calibrate CLAMP are based on those observed to occur in modern climate regimes, and this must include an element of biogeographic history at each of the calibration sites. However, because foliar physiognomy tends to exhibit convergence due to there being constrained architectural solutions to given environmental conditions, the absence of a taxon from a calibration site that it could otherwise have occupied is of little consequence. For this, and other reasons connected with statistical precision, a minimum of 20 morphotypes at a given site is required for any CLAMP analysis (Wolfe, 1993
).
There are no published minimums for LMA, CA, and ODA, but clearly small numbers of taxa are likely to result in large, unknown uncertainties. As a rule of thumb, we suggest that the 20 taxa required for CLAMP should apply to all the methods considered here, although in cases where taxa with highly specific climatic requirements occur this may be relaxed. In some cases, even the occurrence of a single taxon can provide useful insights into past climates (e.g., Wang et al., 2003
).
Both CLAMP and ODA provide the potential for estimating palaeoelevation. This is a powerful feature because understanding ancient topography is important in developing concepts of crustal dynamics and constraining the movement of air masses in palaeoclimate modeling. To achieve palaeoelevation estimates from CLAMP, values for enthalpy must be obtained for a site at a known height as well as at the unknown height. The difference between these two enthalpy estimates can then be used to yield a height difference. Ideally, a fossil assemblage identical in age to that at the unknown altitude, sandwiched between conformable marine sediments so it is likely to represent an elevation close to sea level, would be used to estimate the enthalpy at the known elevation (Forest et al., 1995
; Wolfe et al., 1998
). Unfortunately, this is not always possible, and in such cases model-derived enthalpy values can be used to provide the fixed datum (Spicer et al., 2003
). Sun et al. (2002)
did not attempt an elevation estimate using enthalpy and a quantitative estimate of the palaeoelevation of the Shanwang site is not considered further here.
Overall, no single method is infallible, nor can it be universally applied, and intercomparison studies such as this are needed in a wide variety of situations to explore temporal and biogeographical limits on each technique.
RESULTS
The Miocene Shanwang climate
A total of 127 fossil taxa from the Shanwang flora was recorded. Of them, 107 fossil species have NLRs with known distributions in China, covering 36 endemic species. These are listed in Appendix S1 (see Supplemental Data accompanying the online version of this article). The taxa that define the boundaries are shown in Table 1.
|
The meteorological data within the overlapping area provide the following characterization of the Shanwang Miocene climate throughout all 19 units combined: MAT 10.9–14.5°C, MART 21.1–22.7°C, CMMT (–0.5)–3.3°C, WMMT 21.9–25.0°C, and MAP 1107.3–1880.0 mm (Appendix S2, see Supplemental Data accompanying the online version of this article).
LMA was also used to estimate the MAT of the Shanwang Miocene basin. When the leaves of 113 woody dicot species (note here that more taxa can be used because the NLRs do not have to be identified and their distributions determined) were used, the MAT of the Shanwang Miocene basin was estimated as 12.0°C with an error of ±1.4°C. The MAT estimated by the LMA is within the range estimated by the ODA.
Climate estimated within and between individual horizons
Palaeoclimate parameters derived using ODA from six subunits (4, 5, 7, 13, 14, and 15) are listed next (Appendix S3, see Supplemental Data accompanying the online version of this article).
MAT
MAT values range from 8.5 to 15.3°C (Tables 2 and 3), which suggest some real climate fluctuations during the period of time represented by the six subunits (Fig. 2). Because of the uncertainties associated with the MAT estimates in each subunit, some of the MAT fluctuations could be ignored. However, there does appear to be a warming from subunit 7 (10.811.4°C) to 13 (11.914.5°C).
|
|
|
), but distinctly lower than those derived from CA using either leaf or palynofloral data. The uncertainties in MAT yielded by ODA are similar to those of LMA, but more narrowly defined.
CMMT, WMMT, and MART
Perhaps unsurprisingly the fluctuations of CMMT and WMMT values parallel those of the MAT. Most CMMT and WMMT values in each subunit estimated by ODA are much lower than those by CA based on both leaf and palynofloral data. However, the CMMT and WMMT values suggested by ODA are close to those obtained by CLAMP (Figs. 3--4). LMA is incapable of providing these or the following climate parameters.
|
|
|
Figure 6 shows that in subunits 4 and 7 MAP values obtained by the ODA based on the megaflora alone are higher than those from the CA approach based on a combination of foliar material together with the palynoflora. However, in subunit 14 ODA values are lower than those from CA based on the palynoflora alone. In the current configuration of CLAMP, MAP is not normally determined because in any relatively wet regime where water availability is not limiting to plant growth, leaf physiognomy is not constrained by precipitation. This results in poor precision in CLAMP precipitation estimates. In areas that experience drought, however, precipitation uncertainties in CLAMP become less. This is particularly so regarding precipitation estimates for the three driest months. Poor MAP estimates in humid or wet regimes are further degraded in vegetation where dormancy occurs because during the period of no growth plants are largely insensitive to precipitation (unlike temperature, Spicer et al., 2004
).
|
As early as 1937, Skvortzov (1937)
described six species of diatom in the Shanwang Formation, Melosira distans (Ehr.) Kutz., Fragilariaria virescens Ralfs var. elliptica Hust., Eunotia clevei Grun., E. submonodon Hust., Pinnularia episcopalis Cleve, and Tetracyclus lacustris (Ralfs). These species are still living in alpine lakes in southern China and at lower elevation lakes in northern Asia. This implies that the Shanwang flora may have been deposited in an alpine lake in Miocene times. Hu and Chaney (1940, p. 86) recognized this and stated, "It seems reasonable to conclude that during the Miocene the Shanwang Basin, whose altitude is now 250 meters, was located on a highland not less than a thousand meters above its present level, and possibly much higher." They also suggested that the taxonomic richness may reflect topographic diversity. The Miocene Shanwang flora is very similar to the montane floras of the middle reaches of the Yangtze River (Hu and Chaney, 1940
; Liu and Leopold, 1992
; Yang et al., 2002
). About 80% of the genera in Shanwang Miocene now live in Shennongjia (Hubei Province) (31°21'20''–31°36'20'' N, 110°03'05''–110°33'50'' E). More than half (51.8%) of the NLR species of the Shanwang Miocene flora are found in this area, and most of these species live from 1000 to 1200 m (Zhu and Song, 1999
). The fact that 82% of the NLRs (90/127) from China occur in the maximum overlap area is reassuring. It is unrealistic to expect complete congruence because as Mosbrugger and Utescher (1997, p. 66) pointed out, "In fossil floras, however, it turns out that in most cases there is no climatic interval in which all nearest living relatives of the fossil taxa can coexist. This apparent "climatic inconsistency" of fossil floras may be due to various reasons: (1) the climatic tolerances defined in the data base could be wrong or inadequate for some of the nearest living relatives and for some of the climate parameters: (2) some of the fossil taxa could have climatic tolerances different from those of their nearest living relatives; (3) the assignment of the nearest living relative could be wrong or inadequate for one or a few fossil taxa; (4) the determination of one or a few fossil taxa could be wrong." Added to this is the fact that neither evolutionary pressure, nor climate change, are ever likely to result in taxonomic or biogeographic stasis. In this context the exclusion of anomalous taxa is a strength, not a weakness, of the method.
MATs estimated by the CLAMP, LMA, and ODA are similar to each other and close to that of present Shanwang (12.4°C). Nevertheless, we noticed that the Shanwang Miocene flora is a mixed one with subtropical (e.g., Cinnamoumum, Ficus, and Keteleeria) and temperate elements (e.g., Betula, Salix, and Ulmus). The local flora today in Shanwang region is typically temperate without any subtropical elements. Despite the MAT of Miocene Shanwang basin apparently being similar to today, Miocene Shanwang had a warmer winter (CMMT (–0.5)–3.3°C) that was more humid (MAP 1107.3–1880.0 mm) than present (CMMT –3.1°C; MAP 709.4 mm). The milder winter and overall wetter regime could explain the presence of subtropical elements. The closest modern analogue for the palaeoclimate of Shanwang is in the mountain region at the middle reaches of Yangtse River, where many subtropical and temperate elements match those of Miocene Shanwang flora at the levels of genera or even species (Hu and Chaney, 1940
; Liu and Leopold, 1992
; Yang et al., 2002
).
Sun et al. (2002)
suggested that lower temperature estimates obtained from CLAMP and LMA compared to CA in the small Miocene Shanwang basin might be due to preferential selection of leaves from plants proximal to the lake margin with physiognomies adapted to the lakeside microclimate. Here it is worth bearing in mind that both Mosbrugger and Utescher (1997)
and Wolfe (1993) deliberately chose not to factor in relative abundance in their methodologies. In a lacustrine situation such as Shanwang where there is an inferred pronounced highland region surrounding the lake margins, the inclusion of even a small number of leaves from higher elevations away from the immediate lakeside microclimate will be detected in the analyses and given equal weight to those leaves growing proximal to the lake. However, long distance downslope transport of leaves is highly destructive (Spicer and Wolfe, 1987
), so the number of leaves entering the lake from significantly higher altitudes will be small. Few taxa (or morphotypes) from higher elevations will be captured in the analyses, and their signal will be diluted by the more proximal contributions. Because there is also a taxonomic difference between lake-margin taxa and those growing on higher, better-drained substrates, any affect of differential transport should affect all techniques used here to some degree. Both CA and ODA analyses will include more taxa growing around the lake margin than at those occurring at higher altitudes, and similarly LMA and CLAMP include more lake microclimate adapted leaf forms than those reflecting climates distal to the lake.
The ODA analysis detects climate fluctuations in the Shanwang sequences (Figs. 2--6), and this differs from the conclusion of Liang et al. (2003)
who suggested that no overall climate change was apparent through the sequence when uncertainties are taken into account.
Considering the MAT trends estimated by the ODA, CLAMP, and LMA, it is clear that they are same (from warming to cooling) from the subunit 4 to subunit 7 (Fig. 2). But they are different from the subunit 7 to subunit 15: from warming to cooling to warming by ODA, from warming to warming to cooling by CLAMP, and from cooling to warming to warming by LMA (Fig. 2).
Given the congruencies in physiognomic and NLR climate proxies and their importance for quantifying locally based climate parameters in warmer global regime contexts, there is a clear and urgent need for developing high quality calibration data using modern vegetation taking into account biogeographic context and quality of climate data. Both physiognomic and NLR approaches have strengths and weaknesses, and neither is superior to the other. Both have their individual applications, and where relevant, both should be applied to fossil sites to provide important cross validation of results.
FOOTNOTES
1 The authors thank Drs. J. J. Wòjcicki, K. N. Paudayal, A. D'Rozario, and S. Bera for providing data on plant distributions in Europe, Nepal, and India, Prof. S. Syabryay for providing valuable advice on an earlier draft and two anonymous reviewers for helpful comments on this manuscript. This work is supported by the projects of the Natural Science Foundation of China (30530050, 30470117 and 30100012), the National Basic Research Program (2004CB720205), the Chinese Academy of Sciences Project (KSCX2-YW-Z-065), and the Royal Society in Great Britain/Chinese Academy of Sciences (RS/CAS) joint project. 
7 Authors for correspondence (e-mail: Wang, wangyf{at}ibcas.ac.cn
; Li, lics{at}ibcas.ac.cn
)
LITERATURE CITED
Ackerly D. D. Dudley S. A. Sultan S. E. Schmitt J. Coleman J. S. Linder C. R. Sandquist D. R. Geber M. A. Evans A. S. Dawson T. E. Lechowicz M. J.. 2000. The evolution of plant ecophysiological traits: recent advances and future directions. BioScience 50: 979-995..[CrossRef][ISI]
Bailey I. W. Sinnott E. W.. 1915. A botanical index of Cretaceous and Tertiary climates. Science 41: 831-834..
Bailey I. W. Sinnott E. W.. 1916. The climatic distribution of certain types of angiosperm leaves. American Journal of Botany 3: 24-39..[CrossRef][ISI]
Berggren W. A. Kent D. V. Swisher C. C. III Aubry M. P.. 1995. A revised Cenozoic geochronology and chronostratigraphy. In W. A. Berggren, D. V. Kent, M. P. Aubry, J. Hardenbol, [eds.], Geochronology, time scales and global stratigraphic correlation, SEPM special publication 54, 129-212. SEPM, Tulsa, Oklahoma, USA..
Brandis D.. 1971. Indian trees (fifth impression). Bishen Singh Mahendra Pal Singh, Dehradun, India..
Bureau of Geology and Mineral Resources of Shandong Province. 1991. Regional geology of Shandong Province. People's Republic of China Ministry of Geology and Mineral Resources Geological Memoirs, series 1, no. 26. Geological Publishing House, Beijing, China..
Chen D. G. Pen Z. C.. 1985. K-Ar ages and Pb, Sr isotopic characteristics of Cenozoic volcanic rocks in Shandong, China. Geochemica 4: 293-303..
Delectis Florae Reipublicae Popularis Sinicae Aqendae Academiae Sinicae Edita. 1959–1998. Flora Reipublicae Popularis Sinicae, vols. 1–80. Science Press, Beijing, Beijing, China..
Deng L. H.. 1976. A review of the ‘bamboo shoot' fossils at Yenzhou recorded in ‘Dream Pool Essays' with notes on Shen Kuo's contribution to the development of palaeontology. Acta Palaeobotanica Sinica 15: 1-16..
Forest C. E. Molnar P. Emanuel K. A.. 1995. Palaeoaltimetry from energy conservation principles. Nature 374: 347-350..[CrossRef]
Gamble J. S.. 1902. A manual of Indian timbers. Bishen Singh Mahendra Pal Singh, Dehra Dun, India..
Grubov V. I.. 2001. Key to the vascular plants of Mongolia (with an atlas), vols. I–II. Science Publishers, Enfield, New Hampshire, USA..
Guiot J.. 1987. Late quaternary climatic change in France estimated from multivariate pollen time series. Quaternary Research 28: 100-118..
Hara H. Stearn W. T. Williams L. H. J.. 1978–1982. An enumeration of the flowering plants of Nepal, vols. I–III. British Museum of Natural History, London, UK..
Hooker J. D.. 1876–1896. The flora of British India, vols. I–VII. Bishen Singh Mahendra Pal Singh, Dehradun, India..
Hu H. H. Chaney R. W.. 1940. A Miocene flora from Shantung Province, China. Carnegie Institute of Washington Publication 507: 1-147..
Hultén E. Fries M.. 1986. Atlas of North European vascular plants, vol. I. Koeltz Scientific Books, Königstein, Germany..
Information Department of Beijing Meteorological Center. 1984. Land climate data of China (part I–VI) (1951–1980). Meteorology Press, Beijing, Beijing, China..
Kennedy E. M. Spicer R. A. Rees P. M.. 2002. Quantitative paleoclimate estimates from Late Cretaceous and Paleocene leaf floras in the northwest of the South Island, New Zealand. Palaeogeography, Palaeoclimatology, Palaeoecology 184: 321-345..[CrossRef]
Komarov V. L., [ed.],. 1934–1960. Flora of the USSR, vols. 1–30. Academy of Sciences of USSR, Moscow, Russia..
Lande R. Arnold S. J.. 1983. The measurement of selection on correlated characters. Evolution 36: 1210-1226..[CrossRef]
Lecomte M. H.. 1907–1951. Flore Generale de L'Indo-Chine, vols. I–VII. Paris Masson et Cie, Editeurs, Paris, France..
Li C. S. Wang Y. F. Sun Q. G. Li F. L. Zhang J. P. Wang X. Z. Li J. D. Chen P. F.. 2000. Stratigraphical sequence of diatomaceous beds within Shanwang Formation, Linqu Country, Shandong Province, China. Chinese Bulletin of Botany 17: 247-251..
Li F. L.. 1991. Reconsideration of the Shanwang Formation, Linqu, Shandong. Journal of Stratigraphy 15: 123-129..
Li F. L. Zhang J. P. Li M. L.. 2000. A study on Miocene Shanwang formation, Linqu, Shandong, China—with a brief review of research history in this locality. Chinese Bulletin of Botany 17: 237-246..
Liang M. M. Bruch A. Collinson M. Mosbrugger V. Li C. S. Sun Q. G. Hilton J.. 2003. Testing the climatic estimates from different palaeobotanical methods: an example from the Middle Miocene Shanwang flora of China. Palaeogeography, Palaeoclimatology, Palaeoecology 198: 279-301..[CrossRef]
Liu G. W. Leopold E. B.. 1992. Paleoecology of a Miocene flora from the Shanwang Formation, Shandong Province, northern East China. Palynology 16: 187-212..[Abstract]
Mauseth J. D.. 1988. Plant anatomy. Adison Wesley/Benjamin Cummings, San Francisco, California, USA..
Mori T.. 1921. An enumeration of plants hitherto known from Corea. The Government of Chosen, Seoul, Korea..
Mosbrugger V. Utescher T.. 1997. The coexistence approach—a method for quantitative reconstructions of Tertiary terrestrial palaeoclimate data using plant fossils. Palaeogeography, Palaeoclimatology, Palaeoecology 134: 61-86..[CrossRef]
Ni J.. 1997. Development of Kira's indices and its application to vegetation–climate interaction study of China. Chinese Journal of Applied Ecology 8: 161-170..
Ni J. Song Y. C.. 1998. A discussion on the relationship between Penman's indices and distribution of some tree species of subtropical evergreen broadleaved forest in China. Acta Botanical Sinica 40: 647-656..
Ohwi J. Kitagawa M.. 1992. New flora of Japan. Shibundo Co., Publishers, Tokyo, Japan..
Press J. R. Shrestha K. K. Sutton D. A.. 2000. Annotated checklist of flowering plants of Nepal. British Museum of Natural History, London, UK..
Qiu Z. X.. 1990. The Chinese Neogene mammalian biochronology. Its correlation with the European Neogene mammalian zonation. In E. H. Lindsay, V. Fahlbusch, P. Mein, [eds.], European Neogene mammal chronology, 527-556. Plenum, New York, New York, USA..
Royer D. L. Wilf P.. 2006. Why do toothed leaves correlate with cold climates? Gas exchange at leaf margins provides new insights into a classic palaeotemperature proxy. International Journal of Plant Sciences 167: 11-18..[CrossRef][ISI]
Schuepp P. H.. 1993. Leaf boundary layers. New Phytologist 125: 477-507..[CrossRef][ISI]
Skvortzov B. V.. 1937. Neogene diatoms from eastern Shantung. Bulletin of the Geological Society of China 17: 193-208..
Song Z. C. Cao L. Li M. Y.. 1964. Tertiary spore–pollen complexes of Shandong. Memoir of Institute of Geology and Palaeontology of Academia Sinica 3: 179-190..
Spicer R. A. Harris N. B. W. Widdowson M. Herman A. B. Guo S. X. Valdes P. J. Wolfe J. A. Kelley S. P.. 2003. Constant elevation of southern Tibet over the past 15 million years. Nature 421: 622-624..[CrossRef]
Spicer R. A. Herman A. B. Kennedy E. M.. 2004. Foliar physiognomic record of climatic conditions during dormancy: climate leaf analysis multivariate program (CLAMP) and the cold month mean temperature. Journal of Geology 112: 685-702..[CrossRef][ISI]
Spicer R. A. Herman A. B. Kennedy E. M.. 2005. The sensitivity of CLAMP to the taphonomic loss of foliar physiognomic characters. Palaios 20: 429-438..
Spicer R. A. Wolfe J. A.. 1987. Taphonomy of Holocene deposits in Trinity (Clair Engle) Lake, northern California. Paleobiology 13: 227-245..[Abstract]
Steininger F. F. Berggren W. A. Kent D. V. Bernor R. L. Sen S. Agust J.. 1996. Circum-Mediterranean Neogene (Miocene and Pliocene) marine–continental chronologic correlations of European mammal units. In R. L. Bernor, V. Fahlbusch, H. W. Mittmann, [eds.], The evolution of eastern Eurasian Neogene mammal faunas, 7-46. Columbia University Press, New York, New York, USA..
Sun Q. G.. 2000. Quantitative reconstruction of palaeoclimate from the Middle Miocene Shanwang leaf flora, Shandong Province, eastern China. Ph.D. dissertation, Institute of Botany, Chinese Academy of Sciences, Beijing, Beijing, China..
Sun Q. G. Collinson M. E. Li C. S. Wang Y. F. Beerling D. J.. 2002. Quantitative reconstruction of palaeoclimate from the Middle Miocene Shanwang flora, eastern China. Palaeogeography, Palaeoclimatology, Palaeoecology 180: 315-329..[CrossRef]
ter Braak C. J. F.. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 1167-1179..[CrossRef][ISI]
Tutin T. G. Heywood V. H. Burges N. A. Valentine D. H. Walters S. M. Webb D. A.. 1964. Flora Europaea, vol. 1. Cambridge at the University, England, UK..
Wang F. Z. Jin L. Y.. 1986. Petrological and geo-chemistry characteristics of Cenozic basalts from Shanwang area, Shandong. Collections of mineralogy and petrology, vol. II, 43-65. Geological Publishing House, Beijing, Beijing, China..
Wang X. Z.. 1981. An introductory survey of the palaeo-surroundings of Miocene Shanwang Lake in Linqu, Shandong Province. Acta Scientiarum Naturalium Universitatis Pekinensis 4: 100-111..
Wang Y. F. Li C. S. Collinson M. E. Lin J. Sun Q. G.. 2003. Eucommia (Eucommiaceae), a potential biothermometer for the reconstruction of paleoenvironments. American Journal of Botany 90: 1-7..
Wilf P.. 1997. When are leaves good thermometers? A new case for leaf margin analysis. Paleobiology 23: 373-390..[Abstract]
Wing S. L. Greenwood D. R.. 1993. Fossils and fossil climate: the case for equable continental interiors in the Eocene. Philosophical Transactions of the Royal Society of London, B, Biological Sciences 341: 243-252..[CrossRef]
Wolfe J. A.. 1979. Temperature parameters of the humid to mesic forests of eastern Asia and their relation to forests of other regions of the Northern Hemisphere and Australasia. United States Geological Survey Professional Paper 1106: 1-37..
Wolfe J. A.. 1993. A method of obtaining climatic parameters from leaf assemblages. United States Geological Survey Bulletin 2040: 1-71..
Wolfe J. A. Forest C. E. Molnar P.. 1998. Paleobotanical evidence of Eocene and Oligocene paleoaltitudes in midlatitude western North America. Geological Society of America Bulletin 110: 664-678..
Writing Group of Cenozoic Plants of China. 1978. Cenozoic plants from China (fossil plants of China, vol. 3). Science Press, Beijing, Beijing, China..
Wu Z. Y. Ding T. Y.. 1999. Seed plants of China. Yunnan Science and Technology Press, Kunming, Yunan, China..
Yang H. Yang S. P.. 1994. The Shanwang fossil biota in eastern China: a Miocene Konservat-Lagerstätte in lacustrine deposits. Lethaia 27: 345-354..[ISI]
Yang J. Wang Y. F. Sun Q. G. Li C. S.. 2002. Quantitative studies on paleoelevation and palaeoclimate of Shanwang Miocene basin, East China. Earth Science Frontiers 9: 183-188..
Zachos J. Pagani M. Sloan L. Thomas E. Billups K.. 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292: 686-693..[CrossRef][ISI][Medline]
Zhang J. C. Lin Z. G.. 1985. Climate of China. Shanghai Science and Technology Press, Shanghai, Shanghai, China..
Zhu M. Hu G. H. Zhao D. Z. Liu S. L. Hu X. Z. Ma Z. Q. Jian W. Y.. 1985. Potassium argon dating of Neogene basalt in Shanwang area, Shandong Province. Petrological Research 5: 47-59..
Zhu Z. Q. Song Z. S.. 1999. Scientific survey of Shennongjia Nature Reserve. Forestry Publishing House, Beijing, Beijing, China..
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
