Earth and Planetary Science Letters 253 (2007) 159 – 171 www.elsevier.com/locate/epsl
Permo-Pennsylvanian palaeotemperatures from Fe-Oxide and phyllosilicate δ 18 O values Neil J. Tabor ⁎ Department of Geological Sciences, Southern Methodist University, Dallas, TX, 75275-0395, United States Received 13 February 2006; received in revised form 9 October 2006; accepted 11 October 2006 Available online 21 November 2006 Editor: H. Elderfield
Abstract The oxygen isotope composition of fossil roots that have been permineralized by hematite are presented from eight different stratigraphic levels spanning the Upper Pennsylvanian and Lower Permian strata of north-central Texas. Hematite δ18O values range from −0.4% to 3.7%. The most negative δ18O values occur in the upper Pennsylvanian strata, and there is a progressive trend toward more positive δ18O values upward through the lower Permian strata. This stratigraphic pattern is similar in magnitude and style to δ18O values reported for penecontemporaneous authigenic palaeosol phyllosilicates and calcites, suggesting that all three minerals record similar paragenetic histories that are probably attributed to temporal palaeoenvironmental changes across the Late Pennsylvanian and Early Permian landscapes. Palaeotemperature estimates based on paired δ18O values between penecontemporaneous hematite and phyllosilicate samples suggest these minerals co-precipitated at relatively low temperatures that are consistent with a supergene origin in a low-latitude soil-forming environment. Hematite–phyllosilicate δ18O pairs indicate (1) relatively low soil temperatures (∼24 ± 3 °C) during deposition of the upper Pennsylvanian strata followed by (2) a considerable rise in soil temperatures (∼ 35–37 ± 3 °C) during deposition of the lowermost Permian strata. Significantly, δD and δ18O values of contemporaneous phyllosilicates provide single mineral palaeotemperature estimates that are analytically indistinguishable from temperature estimates based on hematite– phyllosilicate oxygen isotope pairs. The results between the two temperature-proxy methods suggest that the inferred large temperature change across the Upper Pennsylvanian–Lower Permian boundary might be taken seriously. If real, such a significant climate change would have undoubtedly had far-reaching ecological effects within this region of Pangaea. Notably, there are important lithological and palaeobotanical changes, such as disappearance of coal and coal swamp floras, across the Upper Pennsylvanian–Early Permian boundary of north-central Texas that may be consistent with major climatic change toward warmer conditions. © 2006 Elsevier B.V. All rights reserved. Keywords: Hematite; Phyllosilicate; Oxygen isotope equilibrium; Palaeotemperatures; Permian
1. Introduction Fe(III) oxyhydroxides commonly form in nearsurface sedimentary environments and are an abundant ⁎ Tel.: +1 214 768 4175. E-mail address:
[email protected]. 0012-821X/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2006.10.024
mineralogical component of ancient sedimentary strata [1]. Many recent studies have demonstrated that the isotopic composition of Fe(III) oxyhydroxides, as well as other minerals that typically form in low temperature environments such as calcite and phyllosilicate, may provide important palaeoenvironmental information related to the conditions of mineral formation in the
160
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
Fig. 1. Pangaean Continental Reconstruction and Permo-Pennsylvanian physiography (∼ 300 Ma BP). The palaeogeographic position of the northcentral Texas study area is marked by the square symbol along western Pangaea, south of the equator. Modified from [47]. (B) Stratigraphic and Geological map of the field area in north central Texas. The stratigraphic column (left) shows epoch and stage boundaries according to conventionalisms for West Texas, whereas the thin dashed lines upon the geological map (right) show stage boundaries according to the International Geological Timescale [48] based upon conodont biostratigraphic zonation in north-central Texas strata (B.Wardlaw, pers. comm., 2005). Bold letters A, B, C, D, E, F, G, H mark approximate locations from which Fe-oxides were sampled in this study. See text for discussion.
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
early burial environment. In particular, these minerals can provide information about the δ18O values of local meteoric water and temperature of crystallization [1–20]. In this regard, pedogenic Fe(III) oxides and phyllosilicates have the potential to provide isotopic records of palaeoclimate, and palaeoclimatic change, through a succession of sedimentary rocks. To date, however, there has been no systematic stratigraphic analysis of oxygen isotope compositions of co-existing Fe-oxides and phyllosilicates preserved in sedimentary strata [17,20]. This study presents mineralogical, chemical and oxygen isotope compositions of Fe(III) oxides from Upper Pennsylvanian through Early Permian sedimentary strata of the eastern Midland basin, north-central Texas, U.S.A. These new Fe-oxide data are compared with previously published oxygen isotope values of penecontemporaneously formed pedogenic phyllosilicates of the eastern Midland basin to investigate (1) isotopic models of mineral formation in the early burial environment, (2) the feasibility of mineral-water oxygen isotope fractionation equations, and (3) PermoPennsylvanian palaeoclimate over western equatorial Pangaea (Fig. 1A).
161
2. Geologic background 2.1. Lithostratigraphy The dominantly terrestrial Upper Pennsylvanian and Lower Permian succession of north-central Texas was deposited in three broad depositional belts (lower and upper coastal plain, and piedmont facies) distributed across the low-sloping eastern shelf of the Midland basin. The succession comprises ∼1100 m of upward fining fluvio-alluvial cycles [21] and contains abundant, welldeveloped palaeosols [17,22]. The study area remained in the western equatorial region of Pangaea, within 5° of the equator, throughout Permo-Pennsylvanian time [M. Steiner, unpublished data; 23]. Fe (III) oxides occur as permineralized plant fossils of in-situ root systems within mudstones and claystones of overbank-floodplain deposits (Fig. 2). Reflected light microscopy of these samples show preservation of original cellular root morphology of the secondary xylem (Fig. 2). This sort of preservation suggests these roots were likely permineralized soon after death, in the very early burial environment, prior to any significant humification of the woody root material [24]. Regionally extensive fluvial sandstone sets [21] are intercalated with marine limestone marker beds throughout
Fig. 2. (A) Photographic image of permineralized fossil root structures (Sample E). Scale is marked in millimeters. (B) Reflected light image of polished transverse section from sample E showing permineralized celluar structure of the root system. Scale bar is 1 mm. (C) Magnified reflected light image of permineralized cells of the secondary xylem in (B). Dark areas define cell walls, whereas light areas define cellular lumen. Field of view is 1 mm. (D) Close up of C. Light circular area is ∼ 0.1 mm across. See Text for discussion.
162
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
the succession and define a high-resolution stratigraphic framework for correlation throughout the study region. Upper Pennsylvanian (and/or Carboniferous) and Lower Permian epoch and stage boundaries have been identified on the basis of conodont and fusilinid biostratigraphy of intercalated marine rocks in the north-central Texas strata (B. Wardlaw, pers. comm., 2005; Fig. 1B). The Pennsylvanian–Permian boundary [301 ± 2 Ma; 25] occurs within the upper Markley Formation of the Bowie Group (Fig. 1B). 2.2. Previous palaeoclimate inferences Palaeosols are an important stratigraphic component of the Upper Pennsylvanian and Lower Permian succession of north-central Texas [22,26]. Tabor and Montañez [17,22] defined eight pedotypes [27] in Permo-Pennsylvanian strata of north-central Texas based on palaeosol macro- and micromorphological characteristics. Those studies interpreted the stratigraphic distribution of pedotypes to record a relatively rapid transition from humid conditions typical of the Late Pennsylvanian to significantly drier conditions in the earliest Permian. A climate trend toward increasingly drier conditions is recorded throughout the Lower Permian by changes in the stratigraphic distribution of palaeosol morphologies [22]. Moreover, the palaeosols proximal to the Pennsylvanian–Permian boundary exhibit morphological characteristics that likely record rapid onset of seasonality in precipitation coincident with the transition to drier conditions [22,26]. In addition to Fe(III) oxides, certain Permo-Pennsylvanian-age pedotypes of the Eastern Shelf of the Midland basin also contain pedogenic phyllosilicates and calcites [17–19]. The oxygen and hydrogen isotope compositions of the pedogenic phyllosilicates were used to estimate changes in meteoric water δ18O values from −5.5% to −3.5% and temperatures from 22 ± 3 °C to 33 ± 3 °C through Permo-Pennsylvanian time in this region of western equatorial Pangea [18,19]. Furthermore, the phyllosilicate δD and δ18O data were used to determine that isotope disequilibrium with coexisting calcite that likely reflects evaporative enrichment of soil water δ18O values prior to calcite precipitation [18]. Collectively, these sedimentological, mineralogical and geochemical studies were interpreted to record isotopically heavier meteoric water δ18O values, temperatures increasing by as much as 10 °C, and a landscape that became progressively drier from Late Pennsylvanian to Early Permian time. Significantly, these lithological and geochemical palaeoenvironmental proxies appear to be consistent with the biological record of changing fossil
Table 1 Name, unit, and position of the Fe-oxide samples, and mineralogy as determined by X-ray diffraction Sample name
Stratigraphic unit
Stratigraphic position (m)1
Mineralogy
H G F E D C
Clear Fork Gp. Clear Fork Gp. Clear Fork Gp. Clear Fork Gp. Nocona Fm. Markley Fm.
850 805 760 710 315 230
B A
Markley Fm. Markley Fm.
205 180
Hematite Hematite Hematite Hematite Hematite Hematite, Goethite, Kaolinite Hematite Hematite
floras across the Permo-Pennsylvanian landscape [26]. Nevertheless, temperature estimates based upon phyllosilicate δ18O and δD values are not without uncertainties. Necessary assumptions that must be made for these palaeotemperature estimates from phyllosilicate include (1) chemical equilibrium with meteoric water during Permo-Pennsylvanian pedogenesis, (2) knowledge of the phyllosilicate-water oxygen isotope fractionation factors and (3) little or no post-pedogenic alteration that would have resulted in isotopic change of original δ18O or δD values of the phyllosilicates. Without an independent method of determining Permo-Pennsylvanian palaeotemperatures, such as oxygen isotope-pair palaeothermometry based on the δ18O values of coexisting palaeopedogenic Fe-oxides and phyllosilicate, it is difficult to evaluate the validity of the aforementioned assumptions and the resulting palaeotemperature estimates. 3. Methods Eight different Fe-oxide samples were taken from the Upper Pennsylvanian Markley Formation and the Lower Permian Archer City, Nocona, and Clear Fork Formations (Table 1). Samples were initially collected in aluminum foil. The chemical pretreatments for the ironoxide samples follow the methods of [6]. Samples were ground in a corundum mortar and pestle under reagentgrade acetone and sized by passage through a 60 μm brass sieve. Only powders from the b60 μm particle size fraction were used in this study. Samples were then kept in ∼ 40 mL of 0.5 N HCl solution over night to remove any admixed carbonates and then rinsed with successive aliquots of deionized H2O until the pH of the solutions associated with the samples was equivalent to the initial pH of deionized H2O. Each sample was subsequently treated over a period of 28–35 days with successive 40 mL aliquots of 30% H2O2 at room temperature to
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
remove organic matter. As the reactivity of the solution diminished and the suspended particles settled, the solution was decanted and replaced by a fresh aliquot of H2O2. After H2O2 treatments, samples were dried in a vacuum desicator at room temperature. In this paper, samples subjected only to the foregoing treatments are designated bulk samples. The mineralogy of the bulk samples was determined by X-ray diffraction (XRD) analysis using Cu–Kα radiation on a Rigaku Ultima III X-ray diffractometer in the Department of Geological Sciences at Southern Methodist University. Samples were back-mounted into an aluminum holder and analyzed under continuous
163
scanning mode from 2–70° 2Θ at 0.01° 2Θ/min, 40 kV and 44 mA. For Fe-oxide samples that contain goethite, the amount of Al substituted for Fe in goethite was determined by the XRD method of Schulze [28]. For chemical analysis, bulk samples were combined with lithium tetraborate to produce a 2:1 mixture on a mass basis. These mixtures were fused in graphite crucibles at temperatures of 1000 °C for one hour and then quenched in deionized water to produce a glass that was subsequently ground to b 60 μm. Approximately 125 mg of fused glass from each sample was then sealed in 15 mL Teflon bombs with 10 mL of concentrated HNO3 and left on a hot plate at 100 °C until all of the
Fig. 3. X-ray diffractograms of powdered permineralized plant material from (A) Sample F and (B) Sample C. Peak positions that were used for mineralogical identification are marked by dashed lines. Spacing (for Cu–Kα, given in angstroms) and the likely mineral responsible for the peak is given above the dashed lines. Peaks near 2.349 Å and 2.034 Å are from the aluminum holder used during analysis, and do not represent the contents of the samples. All of the identifiable peaks in Sample F are attributed to d(hkl) spacings of hematite, whereas Sample C includes peaks with d(hkl) spacings indicating the presence of goethite and phyllosilicate. See text for discussion.
164
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
solids dissolved. Dissolution of all of the samples took place within 2 days. Each 10 mL aliquot was then transferred to 100 mL volumetric flasks and the solution was diluted to 2 to 3% HNO3− . Chemical analyses of the dilute HNO3+ sample solutions were performed on an ICP-OES at the DANR Analytical Facility at the University of California, Davis. The relative analytical error of these analyses is b± 2% of the reported value for the oxide component. In order to assess the isotopic composition of noniron oxide constituents within the samples, ∼ 200 mg aliquots of bulk samples were subjected to Citrate– Bicarbonate–Dithionite (CBD) digestion solutions at ∼ 25 °C in order to remove iron-oxides. The remaining non-iron oxide residue was then washed with 5 to 8 successive 50 mL aliquots of deionized H2O and three to four successive 40 mL aliquots of 30% H2O2. After H2O2 treatment, the residues were dried in an oven at 40 °C. In this paper, these non-iron oxide constituents are designated residue samples. Oxygen isotope analyses of the bulk and residue samples were performed using BrF5 reagent following the procedure of [29] to produce O2 gas at the Department of Geological Sciences at the Southern Methodist University. The O2 gas was then quantitatively converted to CO2 and analyzed for oxygen isotope composition on a Finigan MAT 252 IR Mass Spectrometer. Results are reported in per mil notation, where Rsample d18 O ¼ −1 X 1000 Rstandard R = 18O/16O. δ18O values are reported relative the standard mean ocean water (SMOW) [30]. Replicate analyses of NBS 28 (n = 4) indicate analytical errors of b0.2%. Wt.% H2O values of samples containing goethite (as determined by X-ray diffraction analysis) were
determined by measurement of H2O evolved during high-temperature dehydration of the samples. Samples were initially outgassed at low temperature (∼ 120 °C) for ∼ 10 h in order to remove sorbed water. Samples were then heated to ∼ 1100 °C under vacuum to evolve structural hydroxyl water. The liberated structural water was then converted to H2 by passage over hot (∼ 750 °C) U-metal [31]. After complete conversion of evolved H2O to H2, H2 gas yields were measured in a mercury manometer with an uncertainty of ± 1 μmol. 4. Results 4.1. Mineralogy The mineralogical composition of the bulk samples, as deduced from XRD analysis, is presented in Table 1. All of the samples have prominent peaks near 3.004 Å, 2.71 Å, 2.53 Å, 2.21 Å, 2.06 Å, 1.85 Å, 1.70 Å and 1.60 Å, corresponding to the d(hkl) indices of hematite. In addition, sample C (Table 1) exhibits relatively minor peaks near 4.19 Å, 2.45 Å, 2.24 Å, and 1.56 Å, corresponding to the d(hkl) indices of goethite (Fig. 3). Small peaks near 7.21 Å, 3.58 Å and 3.34 Å correspond to the d(00l) and d(002) indices of kaolinite and the d (333) index of quartz, respectively. 4.2. Chemistry Results of the chemical analyses of the bulk samples are reported in Table 2 as the mole fraction of the oxide components. There is some oxygen that is derived from structural hydroxyl groups associated with goethite that must be taken into account in order to calculate the endmember δ18O value for Fe-oxide in Sample C (Fig. 3; Table 1). The mole fraction of oxygen from H2O (X (O)H2O) in sample C, as determined from high-temperature dehydration, is reported in Table 2. The wt.%
Table 2 Mole fraction of oxygen of the various oxide components in each sample and measured δ18O values of bulk and residue samples Sample X(O)P
X(O)K
X(O)Ca X(O)Mg X(O)Na X(O)Mn X(O)Fe X(O)Al X(O)Si X(O)Ti X(O)H2O X(O)Hem δ18Obulk δ18Oresidue (%) (%)
H G F E D C B A
0.0002 0.0004 0.0005 0.0004 0.0001 0.0005 0.0004 0.0009
0.0088 0.0001 0.0005 0.0001 0.0007 0.0001 0.0001 0.0001
0.0220 0.0032 0.0136 0.0001 0.0070 0.0040 0.0044 0.0013
0.0000 0.0001 0.0001 0.0001 0.0000 0.0001 0.0001 0.0001
0.0004 0.0001 0.0001 0.0001 0.0000 0.0001 0.0001 0.0001
0.0063 0.0082 0.0063 0.0144 0.0055 0.0099 0.0100 0.0106
0.9423 0.9780 0.9673 0.9780 0.9566 0.8891 0.9697 0.9506
0.0132 0.0060 0.0051 0.0041 0.0249 0.0122 0.0090 0.0101
0.0068 0.0034 0.0062 0.0025 0.0051 0.0116 0.0062 0.0258
0.0000 0.0006 0.0004 0.0004 0.0002 0.0003 0.0001 0.0005
0.0000 0.0000 0.0000 0.0000 0.0000 0.0722 0.000 0.000
0.9423 0.9780 0.9673 0.9780 0.9566 0.9549 0.9697 0.9506
4.8 2.1 1.9 1.5 1.4 1.3 0.4 1.7
22.0 18.3 16.4 17.7 20.6 32.1 17.6 17.7
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
structural H2O in sample C was determined before (2.6) and after (5.8) Citrate–Dithionite treatment. This corresponds to 90.2% of the oxygen in structural H2O originating from hydroxyl groups in goethite from sample C. The remaining 9.8% is likely derived from structural hydroxyl originating in kaolinite (Table 1, Fig. 3). Therefore, in addition to oxygen derived from Fe–O bonds (X(O)Fe; i.e. only Fe–O bonds in hematite and goethite), sample C has an additional 0.065 mole fraction of oxygen from structural hydroxyl in goethite. Thus, the mole fraction of total oxygen from all Fe-oxides in sample C (X(O)Hem) is 0.9451. All of the other Fe-oxide samples are hematite and their X(O)Fe values are equivalent to their X(O)Hem values, because all of the oxygen in this mineral occupies Fe–O bonds (i.e., Fe2O3).
165
The mole fraction of oxygen from Fe-oxide minerals (X (O)Hem) in all samples analyzed represents ∼94% to 98% of the total oxygen derived from the sample during BrF5 reaction. Measured δ18O values for the bulk and residue samples are presented in Table 2. These δ18O values range from 0.4% to 4.8%. In all cases, the residues remaining after complete dissolution of Fe-oxides have more positive δ18O values than the bulk (hematite-rich) fractions. This probably reflects the generally more positive range of δ18O values of other common minerals (e.g., phyllosilicates and quartz) compared to Fe (III) oxides in these samples [4,7]. In this regard, the difference between the δ18O of X(O)Fe values measured for the bulk (hematite-rich) and residue (CD-treated)
Table 3 Stratigraphic position and δ18O values measured for paleosol hematite and phyllosilicate samples Position (m) a
Hematite 103ln18α =
Hematite sample
Hematite δ18O (%)
Phyllosilicate 103ln18α =
Phyllo δ18O (%)
T (°C) Phyl-Hem Oxygen isotope pair
850
1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d 1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d 1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d 1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d – – – – – – 1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d
H
3.7
–
–
–
G
1.7
–
–
–
F
1.4
–
–
–
E
1.1
2.82 ⁎ 106/T 2 − 5.06
22
D
– – – – – – 0.5
2.83 ⁎ 106/T 2 − 4.73 2.83 ⁎ 106/T 2 − 4.78 2.85 ⁎ 106/T 2 − 4.85 2.83 ⁎ 106/T 2 − 5.04 2.82 ⁎ 106/T 2 − 5.00 2.83 ⁎ 106/T 2 − 5.13 2.84 ⁎ 106/T 2 − 5.08
25 ± 1 49 ± 1 54 ± 1
22.6 22.7 21.2 21.8 21.1 21.1 20.6
1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d 1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d 1.63 ⁎ 106/T 2 − 12.3 b 0.413 ⁎ 106/T 2 − 2.56 c 0.773 ⁎ 106/T 2 − 6.194 d –
C
− 0.4
2.83 ⁎ 106/T 2 − 4.87 2.83 ⁎ 106/T 2 − 5.23
20.8 19.5
B
− 0.1
2.83 ⁎ 106/T 2 − 6.20
19.6
A
0.9
2.76 ⁎ 106/T 2 − 6.75
19.6
–
2.78 ⁎ 106/T 2 − 6.11
20.4
800
750
710
620 525 475 415 395 365 315
265 230
205
180
135 a b c d
Stratigraphic height above the basal Markley Formation boundary. Yapp [3]. Synthetic hematites, pH = 1–2. Clayton and Epstein [2]. Natural assemblage formed from alkaline solution. Bao and Koch [1]. Synthetic hematites, pH = 8–9.
35 56 62 37 56 60 26 50 54 24 50 54 –
T (°C) Phyllosilicate δD and δ18O
27
28 29 34 26 33 34 35
31 34
33
24
22
166
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
sample fractions represents two points upon a two-endmember mixing line, where: d18 Obulk ¼ XðOÞHem ⁎d18 OHem þ XðOÞresidue ⁎d18 Oresidue
time [34]. However, in order for these minerals to be utilized as palaeothermomters, the temperature-dependent oxygen isotope fractionation between the various minerals and coexisting water must be known. 5.1. Mineral-water oxygen isotope fractionation factors
and
1 ¼ XðOÞHem þ XðOÞresidue
These two end-member mixing relationships result in calculated oxygen isotopic compositions for the endmember Fe-oxide minerals from −0.4% to 3.7% (Table 3; Fig. 4). Specifically, the most negative δ18O values occur within the Late Pennsylvanian strata of the Eastern Midland basin, whereas the Early Permian strata exhibit heavier δ18O values (Fig. 4). Although portions of the Early Permian strata are not represented by hematite δ18O data, oxygen isotopic compositions among Fe-oxides, phyllosilicates and calcites exhibit similar stratigraphic patterns through Upper Pennsylvanian and Lower Permian strata (Fig. 4).
As discussed by Tabor and Montañez [18], the phyllosilicate-water oxygen isotope fractionation equations for the Permo-Pennsylvanian samples vary according to the mole fraction of oxygen contributed from the 1:1 and 1:2 phyllosilicate minerals present within the samples. Furthermore, the chemical composition of the 2:1 phyllosilicates within each sample has an effect upon phyllosilicate-water oxygen isotope fractionation. The calculated phyllosilicate-water oxygen isotope fractionation factors for the Permo-Pennsylvanian phyllosilicates are presented in Table 3. Hematite-water oxygen isotope fractionation equations have been presented by several authors at different
5. Discussion Similar intrabasinal stratigraphic trends have been documented for δ18O values of phyllosilicates and calcites [11,19], phyllosilicates and iron-oxides [16,32] as well as calcites and iron oxides [33]. These studies assert that similar stratigraphic δ18O trends would only be expected between two minerals if they independently record similar temporal changes in environmental factors such as δ18O values of regional waters (i.e., rainfall and/or groundwater) and Earth-surface temperatures. As mentioned earlier, Permo-Pennsylvanian calcites and phyllosilicates from the eastern Midland basin exhibit similar stratigraphic trends indicative of a regional transition to isotopically heavier soil water from Pennsylvanian through Permian time. However, the paleopedogenic calcite and phyllosilicate data also exhibit oxygen isotope disequilibrium, because the calcite samples crystallized from evaporatively modified soil waters that were enriched in O18 [18,19]. Therefore, the Δ18O values between contemporaneously-formed pedogenic calcite and phyllosilicate δ18O values are not permissive of oxygen isotope pair paleothermometry. Nevertheless, The Permo-Pennsylvanian phyllosilicates and Fe-oxide δ18O values also exhibit similar stratigraphic trends (Fig. 4), suggesting they record similar environmental conditions at or near isotope equilibrium. If so, paired oxygen isotope values of the phyllosilicate and hematite minerals may provide geothermometers to indicate temperatures of mineral formation at Earth's surface during Permo-Pennsylvanian
Fig. 4. Plot showing the stratigraphic position (in meters above the base of the Markley Fm.) versus the measured δ18O values of hematites (filled circles; Table 3), Phyllosilicates (filled squares; [17,18]; Table 3) and calcites [17,18] sampled from palaeosols in the Upper Pennsylvanian and Lower Permian strata of north-central Texas. The horizontal dashed line corresponds to the approximate location of the Pennsylvanian– Permian boundary in the study area. All three minerals show a general stratigraphic trend toward heavier δ18O values from Pennsylvanian through Permian strata, suggesting a change toward generally heavier δ18O values of soil moisture and rainfall through this time. See text.
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
167
ranges of temperatures and pH's for both natural and synthetic samples [1–3], Table 3:
values results in a calculated temperature for equilibrium coprecipitation:
1000 ln18 a ¼ 0:413⁎106 =T 2 −2:56ðnatural assemblage½2Þ
103 ln18 aphyllo ¼ 2:82⁎106 =TK2 −5:06½18
1000 ln a ¼ 0:542⁎10 =T −5:221ðpH ¼ 1 2; ½1Þ 18
6
2
ð1Þ
ð2Þ
1000 ln18 a ¼ 0:733⁎106 =T 2 −6:194ðpH ¼ 8 9; ½1Þ
ð3Þ
1000 ln18 a ¼ 1:63⁎106 =T 2 −12:3ðpH ¼ 1 2; ½3Þ
ð4Þ
These studies indicate significantly different hematite-water oxygen isotope fractionation at low temperatures (Fig. 6). In particular, it appears that hematitewater oxygen isotope fractionation factors may be sensitive to solution pH [1,35]. There are very few data for hematite and coexisting water δ18O values from young, natural environments where temperatures and pH levels are known [35]. However, Girard et al. [16] published hematite and water δ18O values from lateritic soil profiles in Yaou, French Guinea, where soil pH's are relatively acid (b 4.5) and soil mean annual temperatures are ∼ 25 °C. Oxygen isotope fractionation between those hematites and local meteoric water exhibits a large range of 103lnα values from 5.8% to 11.0% at 25 °C. (Fig. 6). However, 91% of the Yaou samples exhibit less variation, with hematite-water 103lnα values ranging from 6.8 ± 0.8% (1 s). Such 103lnα values compare most favorably with the hematite-water oxygen isotope fractionation value of 6.0% at 25 °C proposed in Yapp ([35]; 1990; Eq. (4); Fig. 5). Based upon this example (and several others discussed in [35]), the hematitewater oxygen isotope fractionation equation of Yapp [3] is adopted here. As mentioned, if Permo-Pennsylvanian hematites and phyllosilicates from north central Texas record conditions of oxygen isotope equilibrium, then the isotopic difference between penecontemporaneously formed hematites and phyllosilicates may provide an estimate of temperature at the time of mineral precipitation (Fig. 6). For example, at a stratigraphic level of ∼ 710 m (Table 3), phyllosilicate and hematite samples have δ18O values of 22.0 ± 0.2% and 1.1 ± 0.2%, respectively. Combination of the hematite (Eq. (4) above; [3]) and phyllosilicate (Table 3; [18]) oxygen isotope fractionation factors with the measured δ18O
103 ln18 aHem ¼ 1:63⁎106 =TK2 −12:3½3 103 ln18 aphyllo −103 ln18 aHem ¼ 2:82⁎106 =TK2 −5:06 −1:63⁎106 =TK2 −12:3 3 Let; 10 lnð1000 þ ðd18 Ophyllo −d18 OHem Þ=1000Þ ¼ 103 ln18 aphyllo −103 ln18 aHem
ð5aÞ
103 lnð1:0209Þ ¼ ð2:82−1:63Þ⁎106 =TK2 þ ð−5:06 þ 12:3Þ
ð5bÞ
20:68 ¼ 1:19⁎106 =TK2 þ ð7:24Þ
ð5cÞ
1:19⁎106 =13:44 ¼ TK2
ð5dÞ
MTK2 ¼ M88541
ð5eÞ
TK ¼ 298
ð5f Þ
T ð-CÞ ¼ 25F3:
ð5gÞ
This approach was used to calculate oxygen isotope pair estimates of palaeotemperature between δ18 O values of penecontemporaneous hematite and phyllosilicate samples. The reported uncertainty of the calculated temperature of crystallization reflects the sum of errors associated with the fractionation equations, the mole fractions of the oxide components and the δ18O measurement. The results of these calculations, including different hematite-water oxygen-isotope-fractionation equations discussed above, are reported in Table 3. For the Yapp [3] Hematite-water oxygen isotope fractionation equation, hematite-phyllosilicate oxygen isotope pairs appear to be consistent with isotope equilibrium at Earth-surface conditions, with temperatures of co-precipitation ranging from 24 to 35 °C (Table 3). When the Clayton and Epstein [2] and Bao and Koch [1] hematite-water oxygen isotope fractionation equations are considered for the same penecontemporaneous phyllosilicate-hematite oxygen isotope pairs, apparent equilibrium temperatures are too high for Earth-surface conditions, ranging from 50–62 °C.
168
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
It is difficult, probably even impossible, to assess whether ancient minerals truly record isotope equilibrium, or whether they reflect different environments of formation that coincidentally provide reasonable palaeotemperature estimates. As mentioned, Tabor and Montañez [18] estimated the Permo-Pennsylvanian palaeotemperatures from palaeosol profiles across the Permo-Pennsylvanian landscape using the combined δ18O and δD values of Permo-Pennsylvanian palaeopedogenic phyllosilicates. These palaeotemperature estimates are given in Table 3. There are five stratigraphic horizons where hematite-phyllosilicate oxygen isotope pairs and phyllosilicate δ18O and δD values are available for independent estimates of palaeotemperature (Table 3). The hematitephyllosilicate oxygen-isotope pair temperature estimates based on the hematite-water oxygen isotope fractionation equations of Clayton and Epstein [2] and Bao and Koch [1] are consistently ∼20–25 °C higher than stratigraphically equivalent palaeotemperature estimates based on phyllosilicate δ18O and δD values. However, when the hematite-water fractionation equation of Yapp [3] is used
Fig. 6. Plot of the oxygen isotope fractionation values (α-values) for Hematite and Phyllosilicate vs. temperature in degrees Kelvin. α values for hematite follow the oxygen isotope fractionation equation of Yapp (1990), whereas α values for phyllosilicate follow the oxygen isotope fractionation equation for smectite with the chemical composition collected from 710 m (Table 3; [18]). The Δ values measured between oxygen isotope values of co-precipitated hematite and phyllosilicate correspond to equilibrium temperature of crystallization. The Δ value between paleopedogenic hematite and phyllosilicate from 710 m is 0.0210, which corresponds to an equilibrium temperature of crystallization of 25±3 °C. See text for details.
to estimate palaeotemperatures, all five horizons have phyllosilicate-hematite oxygen isotope pairs and phyllosilicate δ18O and δD palaeotemperature estimates that are analytically indistinguishable from one another (Fig. 7, Table 3). 5.2. Palaeoenvironmental implications of PermoCarboniferous temperature change
Fig. 5. Plot of oxygen isotope fractionation (103ln18α) versus temperature (as 106/T2) values with calculated hematite-water oxygen isotope fractionation curves from [1–3]. Open squares represent 103ln18α values for natural soil hematites from 3 different units collected in Yaou Province, French Guinea [16]. The modern meteoric δ18O value, 3.0‰, was used to calculate 103ln18α values for Unit 1. Meteoric water δ18O values used for hematite samples in units 2 and 3 are −4.0% and −5.0%, respectively, which are the interpreted meteoric water δ18O values at the time unit 2 and unit 3 hematites formed [16].
Unless these results are coincidental, the similar estimates for temperatures of crystallization that are provided from δ18O and δD of the Permo-Carboniferous palaeopedogenic phyllosilicates [18] to those of oxygen isotope pairs between coexisting palaeopedogenic hematite and phyllosilicate (using equation of [3]; Fig. 7) have several important implications: (1) hematite and phyllosilicate samples taken from the palaeosol profiles likely preserve a record of oxygen isotope equilibrium, (2) the singlemineral palaeotemperature estimates based upon δ18O and δD of phyllosilicate values assume equilibrium conditions with waters that lie upon the meteoric water line of Craig ([36]; δD = 8 ⁎ δ18O+10). Given the agreement between the single-mineral and oxygen-isotope pair estimates for palaeotemperature, both hematite δ18O and phyllosilicate δ18O and δD likely preserve isotope values that represent equilibrium with meteoric water. This strongly suggests that controls upon the isotope composition of Permo-Pennsylvanian meteoric precipitation (i.e, oceanic isotope composition, kinetic and equilibrium fractionation in meteoric precipitation) are similar to
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
Fig. 7. Plot of the calculated temperatures based on (1) hematite– phyllosilicate oxygen isotope pairs (large gray filled circles) and (2) phyllosilicate δ18O and δD values assuming equilibrium with meteoric waters (small open circles) relative to their stratigraphic position (in meters) above the base of the Late Pennsylvanian Markley Fm. The horizontal dashed line corresponds to the approximate location of the Pennsylvanian–Permian boundary in the study area. See text for discussion.
modern controls upon the meteoric hydrological cycle [36,37]. (3) The isotope composition of the studied hematites and phyllosilicates retain their original PermoPennsylvanian values. (4) The significant temperature change (∼6–10 °C) through Permo-Carboniferous time that is indicated from these data may be taken seriously. Nevertheless, some of the Permian paleotemperature estimates presented here may seem implausible, because they are significantly higher than modern tropical temperatures, and such conditions may be incapable of supporting plant ecosystems. However, it is noteworthy that soil subsurface horizons are buffered against seasonal swings in temperature, and mean annual subsurface soil temperatures are typically 1° to 2 °C warmer than mean annual surface air temperature [41]. The paleotemperature estimates provided here are interpreted to represent the soil subsurface and, as such, should be regarded as maximum, or slight overestimates, of mean annual surface atmospheric temperatures. In addition, soil paleotemperature estimates from the oxygen isotope pair analyses range from 24 ± 3 °C to 35± 3 °C. Due to the analytical uncertainty associated with these estimates, paleotempera-
169
tures can only be confidently discussed as a temperature rise that is minimally 5 °C (from 27 °C to 32 °C) from Pennsylvanian to Permian time. Coincidentally, a rise of 5 °C is thought to be the range of Pleistocene and Holocene temperature change in the tropical latitudes between glacial and interglacial states (although the modern tropics never got quite so hot as those reported for the Permian) [42]. Finally, although some of the paleotemperature estimates from Permian strata may seem too high to support an ecosystem and sustain plant life, it has been known for nearly a century [e.g., 43,44] that the temperature at which photosynthesis is most rapid ranges between 29 and 32 °C, and there is an exponential decline in plant growth away from that optimum range of temperatures. Within analytical uncertainty, the most warm paleotemperature estimates overlap with this range of optimum temperatures for photosynthesis. Therefore, the real limitation to plant growth and paleoecosystem health at these estimated paleteomperatures was likely the availability of water for photosynthesis, rather than the temperature itself. Upon consideration of these issues, the reported range of PermoPennsylvanian paleotemperature estimates is considered reasonable, and they are within the range of permissible values for modern tropical ecosystems. Significant palaeotemperature change has been predicted from Permo-Carboniferous General Circulation Models if atmospheric PCO2 increased from near present atmospheric levels (PAL) during Carboniferous time to 4X to 8X PAL during Permian time [38], and references therein. Significantly, some evidence does exist for increasing atmospheric PCO2 change, from ∼1 to 10XPAL, from Carboniferous to Permian time [39,40], which appears to be temporally and mechanistically linked with the Late Palaeozoic Gondwanan deglaciation [40]. If such a temperature change occurred, it would have likely had a profound impact upon the ecology of this region. For example, elevated atmospheric temperature results in greater potential evapotranspiration and loss of plant-accessible soil moisture [45,46]. Considering the general observation that potential evapotranspiration increases by 0.2 mm/d for every 1 °C increase in temperature [46], a 6–10 °C temperature change would correspond to an increase in potential evapotranspiration of ∼ 400–700 mm/yr. Therefore, in the absence of a concomitant increase in rainfall to offset increased evapotranspiration, soil moisture content and soil water storage would have become significantly drier and less, respectively, from Pennsylvanian to Permian time. Furthermore, this palaeotemperature change would have resulted in less surplus water, or through-flow in soil profiles, thus reducing (1) leaching and chemical weathering in the soil profile and (2) recharge to local
170
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171
and regional water tables. Significantly, the stratigraphic interval over which the estimated Permo-Pennsylvanian temperature change occurred preserves the stratigraphic transition from poorly drained swampland deposits with organic-rich soils (Coals, or Histosols) and deeply weathered, well-drained palaeosols (Ultisols) typical of Upper Pennsylvanian rocks to moderately weathered palaeosols (Vertisols, Alfisols, Inceptisols) characterized by abundant calcium carbonate nodules and rhizoliths typical of Lower Permian rocks [22]. Perhaps this lithological transition from deeply weathered to incompletely leached palaeosol profiles records the effects of increased evapotranspiration, rather than decreased rainfall, across the Permo-Pennsylvanian boundary, whereas the disappearance of coal deposits may record the effects of reduced recharge to the regional water tables and drainage of extensive swamplands as a result of temperature change across the Permo-Pennsylvanian boundary. More detailed comparisons of this palaeotemperature change to regional and global scale species-and genera-level extirpation and extinction may provide further insight to the magnitude of climate change across the Permo-Pennsylvanian boundary. Acknowledgements My most sincere thanks to Bill DiMichele and Dan Chaney, Dept. of Palaeobiology, Smithsonian Institution, for providing an introduction to the study area and a stratigraphic and palaeoecologic framework. Thanks also to Isabel Montañez and Crayton Yapp for helpful conversations that significantly improved the quality of this manuscript. This research was funded by NSF grant EAR-0447381 and EAR-0545654. References [1] H. Bao, P.L. Koch, Oxygen isotope fractionation in ferric oxidewater systems; low temperature synthesis, Geochim. Cosmochim. Acta 63 (1999) 599–613. [2] R.N. Clayton, S. Epstein, The use of oxygen isotopes in hightemperature geological thermometry, J. Geol. 69 (1961) 447–452. [3] C.J. Yapp, Oxygen isotopes in Fe(III) oxides: 2, Possible constraints on the depositional environment of a Precambrian quartz–hematite banded information, Chem. Geol. 85 (1990) 337–344. [4] C.J. Yapp, The stable isotope geochemistry of low temperature Fe(III) and Al "oxides" with iimplications for continental paleoclimates, Geophys. Monogr. 78 (1993) 285–294. [5] C.J. Yapp, An assessment of isotopic equilibrium in goethites from a bog iron deposit and a lateritic regolith, Chem. Geol. 135 (1997) 159–171. [6] C.J. Yapp, Paleoenvironmental interpretations of oxygen isotope ratios in oolitic ironstones, Geochim. Cosmochim. Acta 62 (1998) 2409–2420.
[7] C.J. Yapp, Climatic implications of surface domains in arrays of dD and d18O from hydroxyl minerals: Goethite as an example, Geochim. Cosmochim. Acta 64 (2000) 2009–2025. [8] C.J. Yapp, H. Poths, 13C/12C ratios of the Fe(III) carbonate component in natural goethites, Geochem. Soc. Spec. Pub. 3 (1991) 257–270. [9] M.I. Bird, Geomorphic and palaeoclimatic implications of an oxygen-isotope chronology for Australian deeply weathered profiles, Aust. J. Earth Sci. 40 (1993) 345–358. [10] J.C.C. Hsieh, 1996. An Isotopic Study of Soil Water and Pedogenic Clays in Hawaii. Ph.D. Thesis, Cal. Instit. Tech. 181 p. [11] L.A. Stern, C.P. Chamberlain, R.C. Reynolds, G.D. Johnson, Oxygen isotope evidence of climate change from pedogenic clay minerals in the Himalayan molasse, Geochim. Cosmochim. Acta 61 (1997) 731–744. [12] S.M. Savin, J.C.C. Hsieh, The hydrogen and oxygen isotope geochemistry of pedogenic clay minerals: principles and theoretical background, Geoderma 82 (1998) 227–253. [13] M.A. Poage, D.J. Sjostrom, J. Goldberg, C.P. Chamberlain, G. Furniss, Isotopic evidence for Holocene climate change in the northern Rockies from a goethite-rich ferricrete chronosequence, Chem. Geol. 166 (2000) 327–340. [14] C.P. Chamberlain, M.A. Poage, Reconstructing the paleotopography of mountain belts from the isotopic composition of authigenic minerals, Geology 28 (2000) 115–118. [15] F. Vitali, F.J. Longstaffe, P.J. McCarthy, A.G. Plint, W.G.E. Caldwell, Stable isotopic investigation of clay minerals and pedogenesis in an interfluve paleosol from the Cenomanian Dunvegan Formation, N.E. British Columbia, Canada, Chem. Geol. 192 (2002) 269–287. [16] J.P. Girard, P. Freyssinet, A.C. Morillon, Oxygen isotope study of Cayenne duricrust paleosurfaces; implications for past climate and laterization processes over French Guiana, Chem. Geol. 191 (2002) 329–343. [17] N.J. Tabor, I.P. Montañez, Shifts in late Paleozoic atmospheric circulation over western equatorial Pangaea: insights from pedogenic mineral δ18O compositions, Geology 30 (2002) 1127–1130. [18] N.J. Tabor, I.P. Montañez, Oxygen and hydrogen isotope compositions of Permian pedogenic phyllosilicates: development of modern surface domain arrays and implications for paleotemperature reconstructions, Palaeogeogr. Palaeoclimatol. Palaeoecol. 223 (2005) 127–146. [19] N.J. Tabor, I.P. Montañez, R.J. Southard, Paleoenvironmental reconstructions from chemical and isotopic compositions of Permo-Pennsylvanian pedogenic minerals, Geochim. Cosmochim. Acta 51 (2002) 851–884. [20] N.J. Tabor, I.P. Montañez, R. Zierenberg, B.S. Currie, Mineralogical and geochemical evolution of a basalt-hosted fossil soil (Late Triassic, Ischigualasto Formation, northwest Argentina): potential for paleoenvironmental reconstruction, Geol. Soc. Amer. Bull. 116 (2004) 1280–1293. [21] T.F. Hentz, Lithostratigraphy and paleoenvironments of upper Paleozoic continental red beds, North-Central Texas; Bowie (new) and Wichita (revised) Groups: The University of Texas at Austin, Bur. of Econ. Geol. Rep. Invest. 170 (1988) 48pp. [22] N.J. Tabor, I.P. Montañez, Morphology and distribution of fossil soils in the Permo-Pennsylvanian Wichita and Bowie Groups, north-central Texas, USA: implications for western equatorial Pangaean paleaoclimate during icehouse–greenhouse transition, Sedimentology 51 (2004) 851–884. [23] C.R. Scotese, C.R. Golonka, PALEOMAP paleogeographic atlas. University of Texas at Arlington: Arlington, TX (1984).
N.J. Tabor / Earth and Planetary Science Letters 253 (2007) 159–171 [24] K.A. Dunn, R.J.C. McLean, G.R. Upchurch Jr., R.L. Folk, Enhancement of leaf fossilization potential by bacterial biofilms, Geology 25 (1997) 1199–1202. [25] T.E. Rasbury, G.N. Hanson, W.J. Meyers, W.E. Holt, R.H. Goldstein, A.H. Saller, U–Pb dates of Paleosols; constraints on late Paleozoic cycle durations and boundary ages, Geology 26 (1998) 403–406. [26] W.A. DiMichele, N.J. Tabor, D.S. Chaney, in press. From wetlands to wetspots: the fate and significance of Carboniferous elements in Early Permian coastal plain floras of north central Texas. Geol. Soc. Am. Mem. [27] G.J. Retallack, A pedotype approach to latest Cretaceous and earliest Tertiary paleosols in eastern Montana, Geol. Soc. Amer. Bull. 106 (1994) 1377–1397. [28] D.G. Schulze, The influence of aluminum on iron oxides; VII, Unit-cell dimensions of Al-substituted goethites and estimation of Al from them, Clays Clay Miner. 32 (1984) 36–44. [29] R.N. Clayton, T.K. Mayeda, The use of bromine pentafluoride in the extraction of oxygen from oxides and silicates for isotopic analysis, Geochim. Cosmochim. Acta 27 (1963) 43–52. [30] R. Gonfiantini, Standards for stable isotope measurements in natural compounds, Nature 271 (1978) 534–536. [31] J. Bigeleisen, M.L. Perlman, H.C. Prosser, Conversion of hydrogenic materials to hydrogen for isotopic analysis, Anal. Chem. 24 (1952) 1356–1357. [32] M.I. Bird, A.R. Chivas, Stable-isotope evidence for lowtemperature kaolinitic weathering and post-formational hydrogen-isotope exchange in Permian kaolinites, Chem. Geol. 72 (1988) 249–265. [33] H. Bao, P.L. Koch, D. Rumble, Paleocene–Eocene climatic variation in western North America; evidence from the d18O of pedogenic hematite, Geol. Soc. Amer. Bull. 111 (1999) 1405–1415. [34] H.C. Urey, The thermodynamic properties of isotopic substances, J. Chem. Soc. (1947) 562–581. [35] C.J. Yapp, Rusty relics of Earth history; iron (III) oxides, isotopes, and surficial environments, Annu. Rev. Earth Planet. Sci. 29 (2001) 165–199. [36] H. Craig, Isotopic variations in meteoric waters, Science 133 (1961) 1702–1703. [37] K. Rozanski, L. Araguas-Araguas, R. Gonfiantini, Isotopic patterns in modern global precipitation, Geophys. Monogr. 178 (1993) 1–36.
171
[38] P.M. Rees, A.M. Ziegler, M.T. Gibbs, J.E. Kutzbach, P.J. Behling, D.B. Rowley, Permian phytogeographic patterns and climate data/model comparisons, J. Geol. 110 (2002) 1–31. [39] D.D. Ekart, T.E. Cerling, I.P. Montanez, N.J. Tabor, A 400 million year carbon isotope record of pedogenic carbonate: implications for paleoatmospheric carbon dioxide, Am. J. Sci. 299 (1999) 805–827. [40] I.P. Montañez, N.J. Tabor, D. Niemeier, W. DiMichele, T.D. Frank, C.R. Fielding, J.L. Isbell, CO2-forced climate and vegetation instability during Late Paleozoic deglaciation, Science (in press). [41] S.W. Buol, F.D. Hole, R.J. McCracken, R.J. Southard, Soil Genesis and Classification, Iowa State UniversityPress, Ames, IA, 1997, 527 pp. [42] P.A. Collinvaux, P.E. de Oliveira, J.E. Moreno, M.C. Miller, M.B. Bush, A long pollen record from lowland Amazonia: forest and cooling in glacial times, Science 274 (1996) 85–88. [43] P.A. Lehenbauer, Growth of Maize seedlings in relation to temperature, Physiol. Res. 1 (1914) 247–288. [44] F.M. Wadley, Development-Temperature correlation in the Green Bug, Toxoptera Graminum, J. Agric. Res. (1936) 259–266. [45] C.W. Thornthwaite, An approach toward a rational classification of climate, Geogr. Rev. 38 (1948) 55–89. [46] J.E. Oliver, Evapotranspiration, in: J.E. Oliver, R.W. Fairbridge (Eds.), The Encyclopedia of Climatology, Van Nostrand Reinhold, New York, 1987, pp. 449–456. [47] A.M. Ziegler, M.L. Hulver, D.B. Rowley, Permian world topography and climate, in: I.P. Martini (Ed.), Late glacial and post glacial environmental changes; Quaternary, Carboniferous-Periman, and Proterozoic, Oxford University Press, New York, 1997, pp. 111–142. [48] F.M. Gradstein, J.G. Ogg, A.G. Smith, F.P. Agterberg, W. Bleeker, R.A. Cooper, V. Davydov, P. Gibbard, L. Hinnov, M.R. House, L. Lourens, H.P. Luterbacher, J. McArthur, M.J. Melchin, L.J. Robb, J. Shergold, M. Villeneuve, B. Wardlaw, J. Ali, H. Brinkhuis, F.J. Hilgen, J. Hooker, R.J. Howarth, A.H. Knoll, J. Laskar, S. Monechi, K.A. Plumb, J. Powell, I. Raffi, U. Roehl, A. Sanfilippo, B. Schmitz, N.J. Shackleton, G.A. Shields, H. Strauss, J. van Dam, T. van Kolfschoten, J. Veizer, D. Wilson, A geological time scale 2004, Cambridge University Press, Cambridge, New York, 2004, 384 pp.
