POSITIVE FEEDBACK DRIVES CARBON RELEASE …jcotton/Publications_files/Cotton_et_al...2 to the...

POSITIVE FEEDBACK DRIVES CARBON RELEASE FROM SOILS TOATMOSPHERE DURING PALEOCENE/EOCENE WARMING

JENNIFER M. COTTON*,§, NATHAN D. SHELDON*, MICHAEL T. HREN**,and TIMOTHY M. GALLAGHER*

ABSTRACT. The Paleocene-Eocene Thermal Maximum (PETM) is the most rapidclimatic warming event in the Cenozoic and informs us how the Earth system respondsto large-scale changes to the carbon cycle. Warming was triggered by a massive releaseof 13C depleted carbon to the atmosphere, evidenced by negative carbon isotopeexcursions (CIE) in nearly every carbon pool on Earth. Differences in these CIEs cangive insight into the response of different ecosystems to perturbations in the carboncycle. Here we present records of �13Ccc of pedogenic carbonates and �13Corg frompreserved soil organic matter in corresponding paleosols to understand changes to soilcarbon during the PETM. CIEs during the event are larger in pedogenic carbonatesthan preserved organic matter for corresponding paleosols at three sites across twocontinents. The difference in the CIEs within soil carbon pools can be explained byincreased respiration and carbon turnover rates of near-surface labile soil carbon.Increased rates of labile carbon cycling combined with decreases in the amount ofpreserved organic carbon in soils during the PETM suggests a decrease in the size ofthe soil carbon pool, resulting in a potential increase in atmospheric pCO2 and apositive feedback with warming. The PETM is a model for how the earth systemresponds to warming, and this mechanism would suggest that soils might serve as alarge source for atmospheric CO2 during warming events.

Keywords: PETM, carbon cycle, paleosol, carbon isotopes, organic carbon, pedo-genic carbonate, paleoclimate

introduction

Soils contain the largest stock of carbon in the terrestrial biosphere (Schlesinger,1997), and the primary controls on soil carbon, microbial respiration and plantproductivity, are both sensitive to climate (Andrews and Schlesinger, 2001; Nemaniand others, 2003). However, the response of this reservoir to a rise in global tempera-ture remains poorly understood, and soils have the potential to become a large sourceof CO2 to the atmosphere through positive climate feedbacks (for example Giardinaand others, 2000; Knorr and others, 2005; Liang and Balser, 2012). Modern soil studiesare limited by the length of time they can observe changes in a certain soil, typicallyonly a few years to a decade, which may not be long enough to determine long-termchanges to soil carbon due to warming and increases atmospheric pCO2 becausecarbon stocks can build up or move over larger timescales. Ancient soils formed duringrapid warming events such as the PETM offer a means of identifying how carbon stocksin soils may change in response to future anthropogenically driven climate change.

The Paleocene/Eocene Thermal Maximum (PETM) is the most rapid warmingevent in recent geologic history, and is an analog for future climate change (Zachosand others, 2001). The PETM is characterized by a negative carbon isotope excursion(CIE) in virtually every carbon pool on Earth, both marine and terrestrial (for exampleKennett and Stott, 1991; Koch and others, 1992; McInerney and Wing, 2011). This CIE

* Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, Michigan48109, USA

** Center for Integrative Geoscience, University of Connecticut, Storrs, Connecticut, 06269, USA§ Present Address: University of Utah, Department of Geology and Geophysics, 115 South 1460 East, Salt

Lake City, Utah 84112; E-mail: [email protected]

[American Journal of Science, Vol. 315, April, 2015, P. 337–361, DOI 10.2475/04.2015.03]

337

is thought to be the result of the release of a large amount of a 13C depleted source ofcarbon to the atmosphere, which then was assimilated into marine and terrestrialecosystems (Dickens and others, 1997; Higgins and Schrag, 2006; Zeebe and others,2009). The magnitude of this CIE in different carbon pools on Earth is highly variable(McInerney and Wing, 2011), ranging from an average value of �2.5 permil in benthicforaminifera to an average value of �5.5 permil for pedogenic carbonates. Manyworkers have been interested in determining the true magnitude of the CIE inatmospheric CO2 from these different carbon pools in an effort to quantify the source,amount of CO2, and resulting increase in the concentration of atmospheric CO2during the rapid warming event (Pagani and others, 2006; Zeebe and others, 2009;Diefendorf and others, 2010 and others). In particular, the differences between theCIE recorded by terrestrial and marine carbon pools have been used to quantify themagnitude of carbon release and resulting climatic changes (Bowen and others, 2004;Schubert and Jahren, 2013). However, the differences in these CIEs between differentterrestrial carbon pools also preserve information about the response of differentecosystems to perturbations to the carbon cycle (Bowen and others, 2004; Smith andothers, 2007). Here, we examine changes in carbon isotopes ratios in different soilcarbon reservoirs (pedogenic carbonates and soil organic matter) during the PETM todetermine if soil dynamics and carbon cycling changed during this rapid warmingevent.

Soil organic matter (SOM) turnover typically occurs on timescale of hundreds ofyears (Trumbore and others, 1996). The PETM carbon release occurred over thou-sands of years and was slow enough that changes to the isotopic composition ofatmospheric CO2 were recorded in SOM preserved in paleosols (McInerney and Wing,2011). Pedogenic carbonates precipitate within soils in isotopic equilibrium with totalsoil CO2, which is thought to be predominately derived from the respiration of SOMand root respiration with a smaller contribution from atmospheric CO2 (Cerling, 1984,1991; Kuzyakov, 2006). The carbon isotopic composition of pedogenic carbonates(�13Ccc) is then controlled by the carbon isotopic composition of atmospheric CO2that diffuses into the soil (�13Ca), the carbon isotopic composition of CO2 derived fromplant and microbial respiration (�13Cr), the concentration of CO2 in the soil atmo-sphere derived from both the atmosphere (Ca) and respiration (S(z)), and thetemperature of carbonate precipitation. The concentration of CO2 in the soil derivedfrom respiration is controlled by diffusion, the production rate and the productiondepth, as follows:

S(z)��S

*�0��z2

DS* �1 � e��Z/ �Z�� (1)

where �*s represents the production rate as a function of depth, z represents depth, �zrepresents the characteristic production depth in the soil, and D*S is the bulk diffusioncoefficient for CO2 diffusing through air in soil (Cerling, 1991). Pedogenic carbonatesprecipitate in isotopic equilibrium with total soil CO2, which varies by depth accordingto the following equation:

�13CS � � 1R

PDB� S�z�DS

*

DS13 �13Cr � Ca

*�13Ca

S�z��1 �DS

*

DS13� � Ca

*�1��13Ca�� 1� � 1000 (2)

where DS13 represents the diffusion coefficient for 13CO2 through air in the soil and

RPDB represents the ratio of 13C/12C in the reference standard Pee Dee Belemnite.

338 Jennifer M. Cotton and others—Positive feedback drives carbon release from soils to

Equation (2) can be simplified and rearranged to the following and is often used tocalculate the concentration of CO2 in the atmosphere using pedogenic carbonatespreserved in paleosols.

Ca � S�z��13Cs � 1.0044�13C

r� 4.4

�13Ca

� �13Cs(3)

All other soil forming conditions being equal, given that pedogenic carbonatesare forming from CO2 generated from the respiration of plant derived organicmaterial, a shift in the isotopic composition of atmospheric CO2 should cause a CIE ofsimilar magnitude in both the SOM and pedogenic carbonates in a given soil. Duringthe PETM, the magnitude of the CIE in organic and inorganic soil carbon pools isdramatically different, with preserved SOM recording an average CIE of �3.5 permiland pedogenic carbonates recording an average CIE of �5.5 permil. Both pedogeniccarbonates and soil organic matter are often preserved without isotopic resetting(Koch, 1998; Cotton and others, 2012), and thus the larger magnitude CIE recorded inpedogenic carbonates than in SOM suggests an as of yet unexplored fundamentalchange to the soil carbon pool or soil carbon cycle during the PETM. Here, we explorethe possible causes of these differing CIE magnitudes in soils. To demonstrate thatthe differing excursions are not an artifact of sampling bias, we present matchingrecords of the �13C of both SOM and pedogenic carbonates from three different sitesin North America and Europe.

methods

The data from the three sites were derived from both a literature review andisotopic analysis of new samples, including a new record of the �13Corg of preserved soilorganic carbon from Axhandle Canyon, near Ephraim, Utah. Three vertical transectsthrough the PETM event were measured and correlated to one another through theirstratigraphic position above or below a distinct 5 to 10 m thick conglomerate markerbed. Fresh rock samples for isotopic analysis were collected from the A horizons ofpaleosols. Paleosols were identified based on the presence of root traces, ped structureand mottling. The soils were weakly to moderately developed, were mottled red andgray throughout the Axhandle Canyon section and have been classified mainly asCalcic Inceptisols according to the USDA soil classification system (Soil Survey, 2010).A horizons were identified based on grain size changes and the presence of preservedroot traces. Samples were treated with a 7 percent solution of HCl to remove carbonateand then rinsed, dried, and homogenized according to Cotton and others (2012). Thecarbonate-free soil samples were then weighed into tin capsules and analyzed for theisotopic composition of preserved organic material on a Costech elemental analyzerattached to a ThermoFinnigan Delta V� isotope ratio mass spectrometer. Each samplewas measured either in duplicate or triplicate depending on the variability of themeasured �13Corg in each sample. More variable samples were measured in triplicate.The analytical uncertainty associated with each �13Corg measurement is 0.1 permil.

The measured section of �13Corg of preserved organic carbon at Axhandle Canyonwas correlated to the previously published record of �13Ccc from pedogenic carbonatesat the same site (Bowen and Bowen, 2008) using measured meter levels for the�13Corg transect and published meter levels for the �13Ccc transect. The identificationof the PETM was based not only on the appearance of the CIE, but also on themagnetostratigraphy and a biostratigraphic age constraint for the early Eocene abovethe top of the section (Bowen and Bowen, 2008). Paleosols were considered to havecorrelated values if the measured �13Ccc of pedogenic carbonates was within one meterof the measured �13Corg of preserved organic material. Paleosols with published �13Ccc

339atmosphere during Paleocene/Eocene warming

values outside this one-meter range were not used in this study. CIE magnitudes inboth the SOM and the pedogenic carbonates were calculated and compared. Differ-ences in CIE magnitudes within the same soil are expressed as 13C values, which isdefined as:

13Ccc-org � �13Ccc � �13Corg (4)

where a larger magnitude CIE in pedogenic carbonates than in SOM is representedas a smaller 13Ccc-org value during the PETM than before and after the warming event.

We also compare the magnitude of CIEs in SOM and pedogenic carbonates fromtwo other sites from which isotopic data has already been published. The first site isPolecat Bench in the northern Bighorn Basin. Paleosols in the Willwood formation ofthe Bighorn Basin have been previously described by Kraus and Riggins (2007). Thesepaleosols include two different facies, weakly developed paleosols interpreted to beformed on the flood plain during crevassing of the river, and strongly developedpaleosols developed on fine grain, slowly accumulating parent material derived fromoverbank flooding. These paleosols were identified based on the presence of carbon-ate nodules, mottling, rhizoliths, slickensides and gleyed features. Pedogenic carbon-ates and preserved organic matter were analyzed from the strongly developed pa-leosols, which were typically less than one meter thick. The record of the CIE inpedogenic carbonates was published by Bowen and others (2001) and the CIE in thematching organic matter from the same site was published by Magioncalda and others(2004). The age was constrained by mammalian biostratigraphy and the PETM wasidentified based on the CIE (Bowen and others, 2001). Similar to the AxhandleCanyon site, meter levels were considered to have correlated values if the �13Cccmeasurement was within one meter of the paleosol �13Corg measurement. �13Ccc valueswere correlated to each �13Corg value by comparison of meter levels published witheach isotopic measurement and 13Ccc-org values were calculated for each correspond-ing soil.

The second site with data derived from the literature is the Tendruy section in theSouth-central Pyrenees, Lleida, Spain. The PETM paleosols at this site are preserved inthe Tremp formation, which is made up of coastal plain deposits. The Tendruy sectionis located in the upper part of the Tremp formation, and as been interpreted as amegafan formed by interfingering river channel deposits. Schmitz and Pujalte (2007)describe paleosols of the PETM in the nearby Esplugafreda section as yellowishcumulate paleosols with an abundance of calcium carbonate and gypsum nodules.Based on the presence of gypsum, these soils would likely be classified as Aridisolsunder the USDA soils classification scheme (Soil Survey, 2010), however, the megafandeposits in which the paleosols are found have been interpreted as being formedduring increased seasonal and intra-annual precipitation associated with the PETM(Schmitz and Pujalte, 2003, 2007). For this site, the CIE during the PETM is recordedin the �13Ccc of pedogenic carbonates published by Schmitz and Pujalte (2003). The�13Corg from paleosols in the same section was published by Domingo and others(2009). Identification of the PETM was based on microvertebrate biostratigraphy andthe onset of the CIE. Isotopic data was correlated by matching the first appearance andthe duration of the CIE in the isotopic record of the pedogenic carbonate to the CIErecorded in the preserved organic material. At the Tendruy section, the density of�13Corg measurements through the PETM section is far greater than the density of�13Ccc measurements. Therefore, for this site the isotopic data was considered to becorrelated if the �13Ccc measurement was within three meters of the correspondingpaleosol �13Corg measurement. During the PETM event, 13Ccc-org values were calcu-lated for correlated meter level according to the methods described above. Each PETM�13Corg value at all three sites had a correlated �13Ccc value according to the defined


meter level intervals above. Because Schmitz and Pujalte (2003) only published onevalue for the isotopic composition of pedogenic carbonates prior to the PETM, thisvalue was used in the calculation of 13Ccc-org for each paleosol prior to and afterthe PETM event. While this method may not accurately describe the 13Ccc-org values ofmany of the soils through the section, it does however allow us to investigate thechanges to 13Ccc-org just before and during the PETM. Isotopic data from new andpreviously published analyses are located in the Appendix.

resultsFor each site, the CIE recorded in the isotopic composition of pedogenic

carbonates was larger than the CIE recorded in corresponding preserved organicmatter. Each of these localities has slightly different baseline (pre- and post-PETM)13C values, which is likely due to different soil productivity at each site caused byvariable precipitation regimes (Kraus and Riggins, 2007; Bowen and Bowen, 2008;Cotton and Sheldon, 2012). The apparent change in 13Ccc-org during the PETM canbe directly compared at each site by calculating a 13C anomaly (defined here as �)value according to the following equation:

� � ��13Ccc � �13Corg�sample � � (5)

where � represents the average 13C value for soils forming before and after the PETM(�13Ccc – �13Corg)pre-�post PETM). Figure 1 shows the apparent � values at each site.Using equation 5, the maximum negative excursion � value observed at Polecat Benchin the Bighorn Basin is �4.4 permil and the average � value during the PETM is �2.8permil (fig. 1A). At Tendruy, Spain (fig. 1B), the maximum negative excursion � valueobserved during the PETM is �5.7 permil, with an average � of �3.2 permil duringthe event. At Axhandle Canyon, Utah (fig. 1C), the maximum negative excursion �value is �2.6 permil with an average � value of �1.3 permil through the event. Thenegative � excursion occurs at all three sites, and the magnitude of these � values isalso similar. Given the wide distribution of the localities, these � values suggestchanges to soil carbon cycling during the PETM on a global scale.

discussionThere are many factors that control the isotopic composition of pedogenic

carbonates and thus, the calculated 13Ccc-org values for a soil, including the concentra-tion of atmospheric CO2, temperature, productivity, depth at which the carbonatesprecipitate, additional sources of CO2 to the soil and changing soil carbon cycling. Thefollowing six sections will describe each of these factors and whether or not themechanism is able to explain the observed trend in decreasing 13Ccc-org values duringthe PETM warming.

Increased Atmospheric pCO2

The rapid release of carbon to the atmosphere that drove the negative CIE in avariety of materials caused an increase in the concentration of atmospheric CO2(Zachos and others, 2005; McInerney and Wing, 2011), which would have influencedthe isotopic composition of pedogenic carbonates (Cerling, 1991; Ekart and others,1999). As pedogenic carbonates precipitate in equilibrium with soil CO2, changes tothe isotopic composition of that CO2 will influence the �13Ccc of pedogenic carbonate.Soil CO2 is comprised of CO2 derived from root respiration and microbial oxidation ofSOM, as well as atmospheric CO2 diffusing into the soil. The isotopic composition ofpedogenic carbonates is then controlled by the ratio of 13C enriched atmospheric CO2( �6‰) to 13C depleted respired CO2 ( �25‰) in the soil at the time of carbonateformation (Cerling, 1991). During the PETM a massive release of isotopically depletedcarbon shifted the atmosphere from �5 permil (Tipple and others, 2010) to �8 to


�9 permil (Diefendorf and others, 2010; McInerney and Wing, 2011). Despite thedecrease in the �13Ca of the atmosphere during the event, atmospheric CO2 would stillbe far more enriched in 13C compared to soil respired CO2 and an increase inatmospheric pCO2 would increase the amount of CO2 derived from the atmosphere inthe soil. This would increase the ratio of atmospheric CO2 to respired CO2 in thesoil and tend to increase �13Cs. For a small release of very 13C depleted carbon such asbiogenic methane (Whiticar, 1999), the effect on the �13Ccc of pedogenic carbonateswould be minimal. For a large release of carbon such as plant biomass, the �13Ccc ofpedogenic carbonates would increase, causing an increase in the 13C value for soilsduring the PETM rather than the observed decrease. Thus, a release of carbon to theatmosphere during the PETM could not be the cause of the observed 13C anomalies.

Increased Temperature and ProductivityIf an increase in atmospheric pCO2 increases 13Ccc-om values, then an increase in

soil-respired CO2 at the time of pedogenic carbonate formation will cause a decrease in13Ccc-om values. The amount of respired CO2 is controlled by soil productivity, and isinfluenced by both temperature and precipitation (Brook and others, 1983; Raich andSchlesinger, 1992; Cotton and Sheldon, 2012). With an increase in temperature, onemight expect to observe an increase in soil productivity during the PETM, whichcould lower �13Ccc and 13Ccc-om values. The temperature of the soil also influences

Fig. 1. � values for Polecat Bench (A), Tendruy (B), and Axhandle Canyon (C) calculated using eq. 5.The gray shaded areas highlight the PETM event. Error bars represent the combined analytical errors for the�13Ccc and �13Corg measurements. The anomalies are similar in magnitude for each site. The �13Cvalue used in eq. 5 for Polecat bench was calculated using only post-PETM soils because the pre-PETM soilsare likely recording a brief negative atmospheric CIE prior to the PETM (Domingo and others, 2009).


the isotopic composition of pedogenic carbonates, because the fractionation betweensoil CO2 and calcite during carbonate precipitation is temperature dependent (smallerat higher temperatures; Romanek and others, 1992). Therefore, a temperatureincrease during the PETM would decrease the fractionation between soil CO2 andpedogenic carbonates and lower 13Ccc-om values for the soil. It is possible to quantifythe effects of temperature and a possible increase in productivity on 13Ccc-om byrearranging the pedogenic paleobarometer equation (eq. 3) to:

Ca

S�z��

�13Cs � 1.0044�13Cr� 4.4

�13Ca

� �13Cs(6)

where �13Cr represents the isotopic composition of respired CO2 which is recorded bypreserved SOM and �13Ca represents the carbon isotopic composition of CO2 in theatmosphere. �13Cs represents the carbon isotopic composition of total soil CO2 whichis recorded by pedogenic carbonates, and which is offset from the �13Ccc according to atemperature dependent fractionation factor (Romanek and others, 1992) that de-creases as temperatures increase. S(z) represents the amount of respired CO2 in the soilat the time of pedogenic carbonate formation, and is proportional to soil productivity.Because the concentration of CO2 in the atmosphere during the PETM is not preciselyknown, it is necessary to look at possible changes in productivity as a ratio ofatmospheric CO2/S(z) in the soil. For a low productivity ecosystem, this value might beas high as 1 (that is, S(z) equal to atmospheric CO2), for a higher productivityecosystem, this value might drop to as low as 0.05 (Brook and others, 1983; Cotton andothers, 2013), and if there was a productivity increase during the PETM, one wouldexpect to observe a decrease in the atmospheric CO2/S(z) values during the event.

We can calculate the hypothetical change in CO2/S(z) values for each of these sitesto determine if a combined productivity and temperature increase can explain theobserved � values. These calculations of atmospheric CO2/S(z) factor in a temperatureincrease of 5 °C to 10 °C during the PETM, consistent with changes predicted bypaleotemperature proxies (Zachos and others, 2001; Wing and others, 2005) andstudies that show a larger increase in summer temperatures than mean annualtemperatures (MAT) during warming events (Fricke and Wing, 2004; Snell and others,2013; VanDeVelde and others, 2013). These calculations also assume a decrease in�13Ca from �5 permil to �8 permil (Tipple and others, 2010). For Polecat Bench (fig.2A), there is a decrease in atmospheric CO2/S(z) values through the PETM, from anaverage value of 0.1 for before and after the PETM to negative values during the PETMfor both a 5° and 10 °C increase in temperature. While the ratios drop during thewarming event, negative results from equation 6 imply negative atmospheric CO2contribution to soil CO2, an impossible result that indicates that a temperature andproductivity increase alone cannot account for the 13Ccc-om values. At Tendruy (fig.2B), the atmospheric CO2/S(z) ratios decrease from 0.3 prior to the PETM to valuesvery close to zero during the warming event. With a temperature increase of 5 °C, manyof these values are again negative. For a 10 °C temperature increase, most of theatmospheric CO2/S(z) ratios for soils during the PETM are slightly positive, with anaverage value of 0.02. However, these ratios are so low that the concentration ofrespired CO2 would be 50x more than the concentration of CO2 derived from theatmosphere in the soil. These conditions are similar to those found in rainforest soils(Brook and others, 1983; Cotton and others, 2013), which have S(z) values an order ofmagnitude larger than those found in soils precipitating pedogenic carbonates (Cot-ton and Sheldon, 2012) and would likely be too acidic for carbonate formation. Apartfrom root and microbial respiration, diffusivity also controls S(z) (eq. 1), with decreas-ing diffusivity increasing S(z). However, regardless of the cause of a potential increase


in S(z), modern carbonates typically form in soils with CO2/S(z) values greater than .06 (Cotton and Sheldon, 2012), so plant productivity or diffusivity changes couldpartially but not fully explain the � values at Polecat Bench and Tendruy.

At Axhandle Canyon (fig. 2C), the atmospheric CO2/S(z) ratios are highlyvariable, but show a slight decrease from 0.4 to 0.2. These changes suggest anincrease in productivity during the PETM equating to a doubling of S(z) during thetime of carbonate formation through the warming event. Therefore, an increase inboth temperature and productivity could explain the � values observed at AxhandleCanyon, but not at Polecat Bench and Tendruy.

Changing Depth to Bk HorizonThe carbon isotopic composition of total soil CO2 is not constant in a soil, and

decreases with increasing depth in a soil profile (Cerling, 1991; Amundson and others,1998). This trend is due to the mixing of atmospheric CO2 diffusing in from thesurface of the soil and the production of CO2 that typically occurs at depth (fig. 3). Thedepth at which �13Cs becomes relatively steady is a function of the production rate ofCO2 in the soil and the concentration of CO2 in the atmosphere. Increasing atmo-spheric pCO2 will increase the depth at which �13C of CO2 in the soil reaches a steadystate, and increasing the production rate of CO2 (that is increasing soil productivity)will decrease this depth in the soil. The depth at which pedogenic carbonate formswithin a soil is proportional to mean annual precipitation (Retallack, 2005), and a

Fig. 2. Atmospheric CO2/S(z) ratios for each site calculated using eq. 6. Dark gray circles are calculatedchanges in atmosphericCO2/S(z) for a 5 °C increase in temperature during the PETM. Light gray circles arefor a 10 °C increase in temperature. Decreasing values are caused by increasing productivity under warming,but at Tendruy and Polecat Bench, many atmospheric CO2/S(z) values are negative, implying negativeatmospheric CO2 concentrations.


hypothetical increase in precipitation at each of these sites during the PETM couldincrease the depth at which carbonates are precipitated in a soil such that thecarbonates are forming in equilibrium with soil CO2 that is more 13C depleted than soilCO2 higher in the soil profile.

For example, following the diffusion-production equation for �13Cs with depthdescribed by Cerling (1991; eq. 2), a pedogenic carbonate forming at PolecatBench under pre-PETM conditions with a �13Cr of �23.5 permil (Magionalda andothers, 2004), an atmospheric pCO2 concentration of 1000 ppm, and precipitating at50 cm depth with a production rate of 3 mmol m�2 hr�1 would theoreticallybe expected to have a 13Ccc-om of 16.2 permil (fig. 3A). For a similar soil at PolecatBench during the PETM, temperature would increase by the observed 5 to 10 °C(Fricke and Wing, 2004; Wing and others, 2005) the �13C of respired CO2 woulddecrease to �27.5 permil, atmospheric pCO2 could increase to 1500 ppm. Now we willassume that with warming there was also an increase in mean annual precipitation andpedogenic carbonates were now forming at 150 cm depth in the soil. The 13Ccc-om forthis soil, incorporating a 10 °C temperature increase would be 14. permil. Themaximum � based on sampling depth would be 1.7 permil. An increase in produc-tion rate of CO2 in the soil, which would be expected with an increase in precipitation(Cotton and Sheldon, 2012) would increase the predicted �, but even a doubling ofproduction rate would bring the � value only to 2.1 permil (fig. 3B). Therefore,changing the depth at which pedogenic carbonates are precipitated is an unlikely

Fig. 3. The isotopic composition of soil CO2 with depth in a hypothetical soil. Because the isotopiccomposition of soil CO2 is not steady with depth in a soil, a negative � value could be observed by changingthe depth of carbonate formation. A pre-PETM soil is plotted with open circles and dashed line, with a�13Cr value of �23.5‰, a �13Ca value of �5‰, and an atmospheric CO2 concentration of 1000 ppm.The solid circles with solid line is an example of PETM soil, with a �13Cr value of �27.5‰, a �13Ca value of�8‰, an atmospheric CO2 concentration of 1500 ppm (A). In panel B, the same conditions apply exceptthat productivity has increased from 3 mmol m�2 hr�1 to 6 mmol m�2 hr�1.


explanation for the full 3 permil � observed at Polecat Bench and Tendruy. Inaddition, to increase the depth to carbonate bearing (Bk) horizon by 1 meter, therewould need to be an increase in precipitation on the order of 500 mm yr-1 (Retallack,2005) during the PETM. At Polecat Bench, there is evidence for drying during theevent (Kraus and Riggins, 2007; Kraus and others, 2013), which should decrease thedepth to the Bk horizon, further showing that the depth at which carbonates aresampled cannot account for the observed 13C anomalies.

methane cyclingUnder anaerobic conditions, microbial acetoclastic methanogenesis and meth-

anotrophy within soils can produce CO2. Biogenic methane is isotopically very de-pleted, and has a �13C value ranging from �45 to �80 permil (Whiticar, 1999).Oxidation of that methane by other microbes would contribute CO2 to the soilatmosphere that is isotopically more depleted than respired CO2. Even a smallproportion of total soil CO2 coming from oxidation of methane could shift the �13Cssuch that pedogenic carbonates precipitating in equilibrium with that CO2 would beisotopically more depleted than expected. Highly negative �13Ccc values from non-pedogenic floodplain limestones in the Bighorn basin have been attributed to oxida-tion of methane in ponded water and palustrine environments (Bowen and Bloch,2002). The anoxic or low oxygen environments necessary for methane productionrequire water logged soil conditions. Various continental paleoprecipitation recordsshow increases in seasonality of precipitation in many regions during the PETM(Schmitz and Pujalte, 2007; Handley and others, 2012), which could cause seasonalwater logging of soils. However, pedogenic carbonate precipitation requires well-drained soil conditions, which typically occurs during the dry months in a seasonalclimate (Breecker and others, 2009, 2010). Therefore, the soil conditions required forboth the production of methane and the precipitation of pedogenic carbonates shouldoccur at different times of the year in a climate with highly seasonal precipitation or insoils with completely different drainage conditions, and we would not expect CO2derived from methane to contribute to soil CO2 during the time of carbonateformation.

Additionally, soils experiencing seasonal water saturated conditions often exhibitgleying and redoximorphic features. If soils at these sites were water saturated andproducing methane during the PETM, but not before and after, we would expect toobserve these redoximorphic features within the PETM time interval. Precise soildescriptions for the Tendruy site are unavailable from the previously published papers,but we can assess these pedogenic features at Polecat Bench and Axhandle Canyon. AtPolecat Bench, some soils did exhibit mottling and gleying features (Kraus andRiggins, 2007) but these features were neither consistent throughout nor only foundduring the PETM interval as would be expected if methane production drove thedecrease in 13Ccc-om values. At Axhandle Canyon, many of the soils exhibit mottlingfeatures, but this mottling is not confined to only the PETM interval. These soilfeatures are inconsistent with water logged conditions and methane cycling during thePETM causing the � values observed at these sites.

Additional Carbon SourceSediment formation and transport is sensitive to climate, and many models show

that increased sedimentation results from climatic changes such as increased run offand vegetation change (Tucker and Slingerland, 1997; Armitage and others, 2011).Depending on the source, sediment may contain previously buried carbon that couldbe incorporated into soils formed from fluvial deposits. Foreman and others (2012)find that the Piceance Creek Basin in western Colorado experienced increases insediment flux and discharge during the PETM. Schmidt and Pujalte (2003) also find


increased erosion rates during the PETM in the Basque Basin in northern Spain,including the Tendruy site. If increased sedimentation rates delivered an allochtho-nous source of 13C enriched pre-PETM carbon to soils during the PETM, then themagnitude of the CIE could be masked by the mixed carbon sources. The new carbonwould be respired to produce CO2 and the allochthonous recalcitrant carbon would berespired at a slower rate and preserved in the paleosols as soils aggraded. In thissituation, the true magnitude of the atmospheric CIE would be recorded by thepedogenic carbonates, while the full CIE in the bulk SOM would be suppressed.

The presence of older, likely Mesozoic-aged refractory carbon has been proposedfor PETM sediments in the Bighorn Basin (Wing and others, 2005; Bataille and others,2013). Baczynski and others (2013) observe varying magnitudes of the CIE in pre-served organic material along lateral transects of PETM exposures in the BighornBasin, which they attribute to variations in sedimentation rate and delivery of allochtho-nous carbon to different depositional environments. We can perform a first orderassessment of the possibility of allochthonous carbon muting the full expression of theCIE at each site by comparing the percent C preserved in each soil in relation to its �value. All other factors held constant, if increased carbon-rich sedimentation weredelivering older recalcitrant carbon to the PETM soils, one would expect to observe atrend of increasing � values with increasing percent C in a soil. Figure 4 shows the �value vs. percent C for soils at each site. There is no trend of increasing magnitude of �with amount of preserved SOM in each soil, and at Axhandle Canyon and Tenduy, soilsforming during the PETM tend to have lower percent C than before and after theevent, but this argument relies on the assumption that the total amount of carbonpresent in the pre-PETM and PETM soils was similar. As Baczynski and others (2013)describe, if there were a decrease in soil carbon in the PETM soils, an increase inallochthonous carbon delivered by sedimentation would more substantially impact thetotal isotopic composition of that soil carbon and � values without changing thepercent carbon. However, because there was only C3 vegetation during the PETM andthe isotopic composition of carbon-rich sediment and soils would likely have been atmost a few permil different, a substantial amount of carbon would need to have beendelivered to the soils in order to change the bulk �13Csom value. Additionally, there is aspike in percent C observed in each Bighorn Basin section at the onset of the PETM,followed by a decrease in percent C back to pre-PETM values (Appendix fig. A1). If thisinitial rise in percent C were caused by delivery of allochthonous carbon one would

Fig. 4. � vs. %C for each PETM site. Solid circle data points indicate pre- and post-PETM samples andopen circle data points indicate PETM samples. If increased sedimentation delivered an older source ofcarbon to the soils that muted the CIE in bulk soil organic material, we would expect to see a trend ofincreasing %C preserved with increasing magnitude of � values. This trend is not visible in the data, andmany soils with the highest magnitude � values have very low %C, especially for Tendruy and AxhandleCanyon.


expect sustained high percent C in the soils through out the warming event as erosionrates remained high for an extended period of time. As sustained high percent C is notobserved in these sections, it is instead more likely that the spike in percent C in theBighorn Basin was caused by an initial increase in soil productivity and the drawdownof carbon into the soils at the onset of the PETM. While we cannot rule out thepossibility that the � values at Polecat Bench were caused by increased delivery ofallochthonous carbon during sedimentation, we do not believe this is the most likelyscenario for the decreases in 13C during the PETM.

Additionally, at Tendruy, Schmitz and Pujalte (2003) show a rise in weatheringand erosion during the PETM, and that erosion may have increased the delivery ofsediment to soils, potentially increasing the proportion of allochthonous carbon inthose soils. However, Domingo and others (2009) observe no relationship between the�13C of organic matter and percent C in soils, which they argue demonstrates that the�13C of preserved organic matter is a primary signal of vegetation. Furthermore, basedupon the regional geology near Tendruy, which is dominated by platform carbonatesand organic-poor marine marls (Lopez-Martınez and Pelaez-Campomanes, 1999;Teixell and Munoz, 2000), there is no clear external source for recycled organicmatter.

If increased delivery of allochthonous carbon is the explanation for the 13Canomalies at all three sites, then increased erosion and sedimentation rates were not aregional occurrence during the PETM but more likely a global phenomenon, possiblydue to increased hydrological cycling (Bowen and others, 2004; Pagani and others,2006; Samanta and others, 2013). This mechanism would also suggest that rapidwarming events recorded in paleosol dispersed organic carbon may not record the fullexpression of any associated CIE and that pedogenic carbonates more faithfully recordthe magnitude of CIEs than preserved organic carbon. While the evidence presentedhere does not support this mechanism, we cannot confirm or reject the possibility thatan additional recalcitrant source of carbon to soils caused the 13C anomalies atPolecat Bench in the Bighorn Basin and this idea warrants further investigation.

Increased Labile Soil Carbon CyclingIf allochthonous carbon is not the mechanism driving the � at these PETM sites,

then it is possible to shift the isotopic composition of pedogenic carbonates bychanging depth or the source of respired CO2 in the soil. The characteristic produc-tion depth indirectly controls the isotopic composition of soil respired CO2, withincreasing production depth increasing S(z) values and subsequently lowering �13Csvalues. With increased productivity, one might expect to observe an increase in rootingdepth and thus an increase in production depth (eq. 2, Cerling, 1991). �13Cs isrelatively insensitive to large changes in S(z), with even a 4-fold increase only decreasing�13Cs values by just over 1 permil. Changes to rooting and CO2 production depth alonecannot explain the full � value at Polecat Bench and Tendruy. If the preserved SOM isaccurately recording the isotopic composition of plant material, then the isotopiccomposition of total soil CO2 (from which the pedogenic carbonates are precipitating)is not tracking respired SOM, which suggests a different isotopically depleted source ofcarbon for respired CO2.

The �13Corg of SOM in a soil profile increases with depth due to the higher rates ofdegradation of isotopically depleted labile carbon compared to recalcitrant SOM(Wynn, 2007) or soil CO2-C incorporation into microbial biomass (Ehleringer andothers, 2000; Breecker and others, 2014). The degree of isotopic enrichment at depthis correlated with MAT (Garten and others, 2000) as well as grain size (Wynn andothers, 2005; Wynn, 2007). The difference between the �13Corg of input leaf litter and�13Corg of SOM at depth increases with increasing MAT and decreasing grain size. Themajority of respired CO2 in a soil is generated from the microbial decomposition of


labile dead plant material and SOM, as well as photorespiration by roots and respirationof root exudates (Hanson and others, 2000; Kuzyakov, 2006), and the proportion ofrespiration from these different pools is dependent on a variety of factors includingvegetation type (Hanson and others, 2000) and climate (Chen and others, 2000;Taneva and others, 2006). Changing temperature and precipitation regimes couldchange the types of carbon respired in soils and may increase carbon turnover rates insoils (Schimel and others, 1994; Knorr and others, 2005; Karhu and others, 2014), andalso alter the rates at which certain pools of soil carbon are respired (Knorr and others,2005; Carillo and others, 2011; Liang and Balser, 2012; Karhu and others, 2014). Labilecarbon and carbon derived from root respiration is depleted in 13C compared to SOMat depth, and increased rates of root respiration or respiration of this labile carbonnear the surface of the soil as compared to recalcitrant SOM at depth in the soil couldcontribute more 13C depleted CO2 to total soil CO2 and change the �13C of total soilCO2 (�13Cs) without changing �13Corg of SOM. In this situation the pedogeniccarbonates that precipitate from this soil CO2 would not track the �13Corg of respiredSOM at depth, but would instead record a mixture of CO2, with a larger proportion ofCO2 originating from respiration of root or dead plant material rather than SOM atdepth in the soil.

We can compare the �13Cs recorded by the pedogenic carbonates at PolecatBench and Tendruy to the predicted �13Cs for soils respiring only SOM at depth as wellas respired CO2 derived from varying amounts of root or labile carbon respiration todetermine if higher rates of respiration of isotopically depleted labile carbon thanrecalcitrant SOM could explain the observed 13C anomalies. Figure 5 plots the �13Csrecorded by the pedogenic carbonates at Polecat Bench and Tendruy accounting for a5° (fig. 5A) or 10 °C (fig. 5B) MAT increase in the temperature dependent fraction-ation between �13Ccc and �13Cs (Romanek and others, 1992). The shaded areas showthe predicted �13Cs for pre/post-PETM soils respiring only SOM at depth, as well as thepredicted �13Cs for PETM soils with up to 100 percent contribution of respired labileand root carbon to total soil CO2. The �13Corg of surface litter carbon was calculatedusing the temperature dependent offset between �13Corg of SOM and �13Corg ofleaf litter published by Garten and others (2000), and we have assumed that the�13Corg of leaf litter equals that of labile and root carbon. Table 1 displays the predicted�13Corg of labile carbon during the PETM at Polecat Bench and Tenduy for both a 5 °Cand 10 °C warming across the event. The predicted �13Cs at depth is then calculatedfollowing the Cerling (1991) soil CO2 diffusion-production model, where �13Cadecreases from �5 permil (Tipple and others, 2010) to �8 permil (McInerney andWing, 2011) and atmospheric pCO2 increases from 500 to 1500 ppm during the PETM(McInerney and Wing, 2011) and temperature warms by 5° and 10 °C. Predicted �13Csvalues are reported for 100 cm depth.

Figure 5 shows that the �13Cs values recorded by pedogenic carbonates have somecontribution of respired labile and root carbon to soil CO2, but a large portion ofrespired carbon for pre- and post-PETM soils is from SOM at depth. The AxhandleCanyon 13C anomalies can be explained by an increase in productivity and tempera-ture, or by an allochthonous carbon source and data for that site are not displayed. Forthe PETM soils at Polecat Bench, the majority of pedogenic carbonates are recordingsoil CO2 that is comprised of 25 to 75 percent respired labile or root carbon. For a 5 °C(fig. 5A) warming during the PETM, three of these carbonates show greater than 100percent contribution of labile or root carbon to soil CO2, however with a 10 °Cwarming (fig. 5B), all carbonates fall within a reasonable range (below 100% contribu-tion of respired labile carbon to soil CO2) of �13Cs values. Many the pre/post-PETM carbonates at Polecat Bench also record soil CO2 that is depleted in 13Ccompared to preserved SOM. These carbonates are from the rising limb of the CIE


(meter level 1498–1500) as well as one carbonate that may be recording a pre-PETMcarbon release event that has been documented in �13Corg records from multiple sitesin the Bighorn Basin as well as at Tendruy (Domingo and others, 2009). The Tendruycarbonates exhibit a similar but slightly smaller contribution of labile or root carbon tosoil CO2 than Polecat Bench. For a 5 °C temperature increase, there is one carbonatethat records a greater than 100 percent contribution of labile carbon to soil CO2,which is remedied when temperature is increased by 10 °C. This plot shows that

Fig. 5. Predicted �13Cs and �13Cs recorded by the isotopic composition of pedogenic carbonates fromPolecat Bench and Tendruy for the pre/post PETM. The predicted �13Cs is based on a change inatmospheric pCO2 of 500 ppm to 1500 ppm, a change in �13C of atmospheric CO2 of �5 to �8‰, andno overall productivity change during the PETM. Each predicted �13Cs value is for 100 cm depth in thesoil. �13Corg of labile carbon values were calculated using the dependence of �13Clitter – �13Csom on MAT(Garten and others, 2000). These values are summarized in table 1. Also plotted is the �13Cs recorded bypedogenic carbonates. Each �13Ccc value is converted to �13Cs using the temperature dependent fraction-ation between CO2 and calcite (Romanek and others, 1992). The top panel (A) assumes T�5 °C and thebottom (B) assumes T�10 °C during the PETM.


increased respiration and carbon cycling of a near-surface labile litter or root carbon isa viable mechanism to explain the 13C anomalies. Given that pedogenic carbonatesform over hundreds to thousands of years (Retallack, 2005), a prolonged change in thetype of carbon respired would have occurred to sustain 13C anomalies throughout thePETM. These results also imply that summer (time of carbonate formation; Breeckerand others, 2009) temperature increases during the PETM may have been closer to 10°than 5 °C. This observation is supported by multiple temperature reconstructions fromthe Bighorn Basin and Axhandle Canyon that show larger increases in summertemperatures than in MAT during warming events (Fricke and Wing, 2004; Snell andothers, 2013; VanDeVelde and others, 2013).

Modern soil studies demonstrate that warming and increased atmospheric pCO2could increase rates of carbon cycling (for example, Schimel and others, 1994;Trumbore and others, 1996; Chen and others, 2000; Knorr and others, 2005; Karhuand others, 2014) and that these increases may be restricted to only labile carbon(Cardon and others, 2001; Carillo and others, 2011). Therefore, one could expect toobserve increased rates of labile carbon cycling and increased respiration of labilecarbon near the surface of soils under climatic warming and increased atmosphericpCO2 during the PETM.

implications for carbon cycle

Increased rates of respiration and turnover in labile carbon may change the size ofthe terrestrial carbon pool and consequently affect the global carbon cycle. Trumboreand others (1996) suggest that changing soil carbon dynamics due to anthropogenicclimate change could alter the amount of carbon stored in soils, causing soils tobecome a source of atmospheric CO2 through a decrease in carbon storage. In order tochange the amount of carbon buried in soils, an increase in overall respiration is notnecessary. Because soils aggrade and labile carbon is eventually buried as SOM,increasing respiration at the surface of the soil removes labile carbon at a faster rateand decreases amount of surface carbon available to be buried in the stable pool ofSOM at depth (fig. 7). This decrease in the amount of carbon entering the SOM poolhas been observed under warming and elevated atmospheric pCO2 in moderngrassland soils (Cardon and others, 2001), and extrapolated over hundreds to thou-sands of years, decreases in amount of carbon buried as soils aggrade would eventuallydecrease the amount of carbon stored on land in soils.

Given the length of the PETM warming event (ca. 100 Kyr), these increasedturnover rates could have caused a decrease in the size of the soil carbon pool.Evidence for such a decrease in terrestrial carbon storage could be recorded in therock record as a decrease in the amount of carbon preserved at depth in the PETMpaleosols. This decrease in amount of SOM is in fact what is observed in many paleosolsduring the PETM, including the paleosols at Axhandle Canyon and Tenduy, and also

Table 1

�13Corg values used to calculate �13Cs with varying contributions of respiration fromsurface litter. �13Corg of surface litter is dependent on MAT, and was calculated using the

following equation derived from Garten and others (2000)


at Claret, Spain (Domingo and others, 2009) and Cabin Fork in the Bighorn Basin(Wing and others, 2005) as shown in figure 6. Preliminary evidence from the BighornBasin Coring Project also shows decreases in the abundance of preserved n-alkanes(Baczynksi and others, 2012) as well as pollen (Harrington and Jardine, 2012) duringthe PETM. Measured 13C anomalies and decreases in preserved organic carbonabundances occur at multiple sites with differing ecosystems. Axhandle Canyon was anarid, low productivity ecosystem (Bowen and Bowen, 2008), whereas Polecat Benchand Tendruy were temperate forests (Wing and others, 2005; Domingo and others,2009), suggesting that these changes to soils occurred on a global scale.

A decrease in terrestrial carbon stored in soils translates to an increase in the sizeof the atmospheric carbon reservoir (Trumbore and others, 1996). This process ofchanging respiration rates and its effect on the carbon cycle is summarized in figure7, where the thickness of the line is proportional to the relative size of the fluxes.Because the CIE in different carbon pools on Earth remains stable during thePETM (McInerney and Wing, 2011), it has been suggested that the warming eventwas caused by a continued release of carbon to the atmosphere and that this carbonmay have come from more than one source (Zeebe and others, 2009). It is possiblethat a first pulse of warming from a release of carbon to the atmosphere frommethane clathrates or another source initiated a positive feedback in soils, causingincreased carbon turnover rates in the surface of soils. This increase in respirationreduced the amount of carbon able to be buried as stable SOM, and therebyincreased the flux of carbon back to the atmosphere (fig. 7). The release of thiscarbon back to the atmosphere would contribute to the atmospheric CIE, asrespired CO2 is isotopically more depleted than atmospheric CO2. As the PETM isbroadly analogous to future climate change, a positive feedback system throughwhich increased carbon turnover rates decreases carbon burial in soils andincreases the concentration of atmospheric.

Fig. 6. Percent carbon preserved in paleosols during the PETM from three different sites, PolecatBench (Magioncalda and others, 2004) Tendruy, Spain and Claret, Spain (Domingo and others, 2009).Shaded regions represent the PETM interval at each site.


conclusionsThe carbon isotope excursion during the Paleocene-Eocene Thermal Maximum

is larger in pedogenic carbonates than in corresponding soil organic matter for thesame soils. These different CIEs result in decreased 13C values at three sites aroundthe world. The differing CIEs and resulting � values at Axhandle Canyon are bestexplained by a temperature and soil productivity increase. The � values at PolecatBench and Tendruy cannot be explained by an atmospheric CO2 increase, a rise intemperature and soil productivity, changes to the depth of the Bk horizon, or methanecycling, and are best explained by one of two different mechanisms. We cannot ruleout the possibility that increased sedimentation caused by an enhanced hydrologiccycle could have delivered a larger amount of older recalcitrant carbon to soils thanbefore and after the PETM, which could have muted the signal of the CIE in preserveddispersed organic matter. However, the lack of a relationship between � values and

Fig. 7. Schematic diagram of changing soil respiration during the PETM. The arrows represent fluxesof carbon into terrestrial pools and fluxes of respired carbon back to the atmosphere. The thickness andlength of the lines and size of the arrows represent the relative magnitude of the fluxes of carbon. During thePETM, an increase in respiration rates of labile carbon reduced the amount of carbon buried as stable SOMas the soils aggraded, which over time could have caused an increase in the net flux of CO2 to theatmosphere.


percent C at all three sites, as well as the rapid increase in percent C followed by adecrease throughout the PETM suggests that this mechanism is not the best explana-tion for the larger CIE in pedogenic carbonates than preserved organic matter.

The second and more likely explanation for the � values is increased carbonturnover rates for labile and root derived carbon. In this scenario, increased labilecarbon turnover rates could have initiated a positive feedback system in whichdecreased carbon stored in soils lead to increased concentrations of atmospheric CO2.It is also possible that a combination of these two mechanisms is responsible for the �values, as a decrease in soil carbon (which is predicted by increased surface littercycling) is necessary for an additional older carbon source to substantially raise the�13C of total soil carbon. Modern soil studies support the conclusion that warmingincreases respiration rates for labile carbon pools, however due to length limitations ofthese experiments it is not possible to determine how respiration changes influencethe carbon cycle over hundreds to thousands of years from modern studies. Isotopicevidence from PETM paleosols may show long term changes to soil respiration as aresult of climatic warming and suggest that these changes can decrease the amount ofcarbon buried in soils and affect the global carbon cycle. Therefore, it is possible thatfuture climate warming will alter soil carbon cycling rates in a similar manner, resultingin increasing fluxes of carbon to the atmosphere.

ACKNOWLEDGMENTSThe authors would like to acknowledge the American Association for Petroleum

Geologists, the Society for Sedimentary Geology, the University of Michigan ScottTurner Award and the University of Michigan Rackham Graduate School and NSFAward 1024535 to NDS for funding this research. The authors would also like to thankLora Wingate for sample analysis and Ethan G. Hyland for helpful discussions. Theauthors would also like to thank reviewers Richard Pancost, Justin VanDe Velde andClaudia Mora for their numerous helpful suggestions that significantly improved themanuscript.

APPENDIX

Fig. A1. Percent carbon through the Bighorn Basin PETM sections published by Baczynski and others(2013). The gray shaded areas highlight the PETM interval.


Table A1

Isotopic compositions of preserved organic material and pedogenic carbonates from paleosolsat Axhandle Canyon, Utah

The �13Corg measurements are new and were analyzed at the University of Michigan Stable Isotope Lab-oratory for this study. The �13Ccc measurements were previously published by Bowen and Bowen (2008). Theblack line outlines the data that has been identified as the PETM.


Table A2

Carbon isotopic data from preserved organic material and pedogenic carbonates in paleosolsfrom Polecat Bench in the Bighorn Basin, Wyoming

The �13Ccc measurements of pedogenic carbonates were published by Bowen and others (2001) and the�13Corg measurements were published by Magioncalda and others (2004). The black line outlinesthe data identified as the PETM.


Table A3

Carbon isotopic data from preserved organic material and pedogenic carbonates in paleosolsfrom Tendruy in the Southeastern Pyrenees, Spain

The �13Ccc measurements from pedogenic carbonates were published by Schmitz and Pujalte (2003)and the �13Corg measurements were published by Domingo and others (2009). The black line outlines thedata identified as the PETM.


References

Amundson, R., Stern, L., Braisden, T., and Wang, Y., 1998, The isotopic composition of soil and soil-respiredCO2: Geoderma, v. 82, n. 1–3, p. 83–114, http://dx.doi.org/10.1016/S0016-7061(97)00098-0

Andrews, J. A., and Schlesinger, W. H., 2001, Soil CO2 dynamics, acidification, and chemical weathering in atemperate forest with experimental CO2 enrichment: Global Biogeochemical Cycles, v. 15, n. 1, p.149–162, http://dx.doi.org/10.1029/2000GB001278

Armitage, J. J., Duller, R. A., Whittaker, A. C., and Allen, P. A., 2011, Transformation of tectonic and climaticsignals from source to sedimentary archive: Nature Geoscience, v. 4, p. 231–235, http://dx.doi.org/10.1038/ngeo1087

Baczynski, A. A., McInerney, F. A., and Wing, S. L., 2012, Preliminary n-alkane results from Basin Substation,BBCP: Abstract 1499156 presented at 2012 Fall Meeting, AGU, San Francisco.

Baczynski, A. A., McInerney, F. A., Wing, S. L., Kraus, M. J., Bloch, J. I., Boyer, D. M., Secord, R., Morse, R. E.,and Fricke, H. C., 2013, Chemostratigraphic implications of spatial variation in the Paleocene-EoceneThermal Maximum carbon isotope excursion, SE Bighorn Basin, Wyoming: Geochemistry, Geophysics,Geosystems, v. 14, n. 10, p. 4133–4152, http://dx.doi.org/10.1002/ggge.20265

Battaille, C. P., Mastalerz, M., Tipple, B. J., and Bowen, G. J, 201 3, Influence of provenance and preservationon the carbon isotope variations of dispersed organic matter in ancient floodplain sediments: Geochem-istry, Geophysics, Geosystems, v. 14, n. 11, p. 4874–4891, http://dx.doi.org/10.1002/ggge.20294

Bowen, G. J., and Bloch, J. I., 2002, Petrogrpahy and geochemistry of floodplain limestones from the ClarksFork Basin, Wyoming, U.S.A.: Carbonation deposition and floodplain accumulation on a Paleocene-Eocene floodplain: Journal of Sedimentary Research, v. 72, n. 1, p. 46–58, http://dx.doi.org/10.1306/061901720046

Bowen, G. J., and Bowen, B. B., 2008, Mechanisms of PETM global change constrained by a new record fromcentral Utah: Geology, v. 36, n. 5, p. 379–382, http://dx.doi.org/10.1130/G24597A.1

Bowen, G. J., Koch, P. L., Gingerich, P. D., Norris, R. D., Bains, S., and Corfield, R. M., 2001, Refined isotopestratigraphy across the continental Paleocene-Eocene boundary on polecat bench in the northernBighorn Basin, in Gingerich, P. D., editor, Paleocene-Eocene Stratigraphy and Biotic Change in theBighorn and Clarks Fork Basin, Wyoming: University of Michigan Papers on Paleontology, v. 33, p.73–88.

Bowen, G. J., Beerling, D. J., Koch, P. L., Zachos, J. C., and Quattlebaum, T., 2004, A humid climate stateduring the Palaeocene/Eocene thermal maximum: Nature, v. 432, p. 495–499, http://dx.doi.org/10.1038/nature03115

Breecker, D. O., Sharp, Z. D., and McFadden, L. D., 2009, Seasonal bias in the formation and stable isotopiccomposition of pedogenic carbonate in modern soils from central New Mexico, USA: GeologicalSociety of America Bulletin, v. 121, n. 3–4, p. 630–640, http://dx.doi.org/10.1130/B26413.1

–––––– 2010, Atmospheric CO2 concentrations during ancient greenhouse climates were similar to thosepredicted for A.D. 2100: Proceedings of the National Academy of Sciences of the United States ofAmerica, v. 107, n. 2, p. 576–580, http://dx.doi.org/10.1073/pnas.0902323106

Breecker, D. O., Bergel, S., Nadel, M., Tremblay, M. M., Osuna-Orozco, R., Larson, T. E., and Sharp, Z. D.,2014, Minor stable carbon isotope fractionation between respired carbon dioxide and bulk soil organicmatter during laboratory incubation of topsoil: Biogeochemist, http://dx.doi.org 10.1007/s10533-014-0054-3.

Brook, G. A., Folkoff, M. E., and Box, E. O., 1983, A world model of soil carbon dioxide: Earth SurfaceProcesses and Landforms, v. 8, n. 1, p. 79–88, http://dx.doi.org/10.1002/esp.3290080108

Cardon, Z. G., Hungate, B. A., Cambardella, C. A., Chapin, F. S., III, Field, C. B., Holland, E. A., and Mooney,H. A., 2001, Contrasting effects of elevated CO2 on old and new soil carbon pools: Soil Biology andBiochemistry, v. 33, n. 3, p. 365–373, http://dx.doi.org/10.1016/S0038-0717(00)00151-6

Carillo, Y., Pendall, E., Dijkstra, F. A., Morgan, J. A., and Newcomb, J. M., 2011, Response of soil organicmatter pools to elevated CO2 and warming in a semi-arid grassland: Plant and Soil, v. 347, n. 1–2, p.339–350, http://dx.doi.org/10.1007/s11104-011-0853-4

Cerling, T. E., 1984, The stable isotopic composition of modern soil carbonate and its relationship toclimate: Earth and Planetary Science Letters, v. 271, n. 2, p. 229–240, http://dx.doi.org/10.1016/0012-821X(84)90089-X

–––––– 1991, Carbon dioxide in the atmosphere: Evidence from Cenozoic and Mesozoic paleosols: AmericanJournal of Science, v. 291, n. 4, p. 377–400, http://dx.doi.org/10.2475/ajs.291.4.377

–––––– 1999, Stable carbon isotopes in paleosol carbonates, in Thiry, M., and Simon-Coinco, R., editors,Palaeoweathering, Palaeosurfaces and Continental Deposits: International Association for Sedimentol-ogy Special Publication 27, p. 43–60.

Chen, H., Harmon, M. E., Griffiths, R. P., and Hicks, W., 2000, Effects of temperature and moisture oncarbon respired from recomposing woody roots: Forest Ecology and Management, v. 138, n. 1–3, p.51–64, http://dx.doi.org/10.1016/S0378-1127(00)00411-4

Cotton, J. M., and Sheldon, N. D., 2012, New constraints on using paleosols to reconstruct atmosphericpCO2: Geological Society of America Bulletin, v. 124, n. 9–10, p. 1411–1423, http://dx.doi.org/10.1130/B30607.1

Cotton, J. M., Sheldon, N. D., and Stromberg, C. A. E., 2012, High resolution isotopic record of C4photosynthesis in a Miocene grassland: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 337–338, p. 88–98, http://dx.doi.org/10.1016/j.palaeo.2012.03.035

Cotton, J. M., Jeffery, M. L., and Sheldon, N. D., 2013, Climate controls on soil-respired CO2 in the UnitedStates: Implications for 21st century chemical weathering rates in temperate and arid grasslands:Chemical Geology, v. 358, p. 37–45, http://dx.doi.org/10.1016/j.chemgeo.2013.08.048

Dickens, G. R., Castillo, M. M., and Walker, J. C. G., 1997, A blast of gas in the latest Paleocene: simulating


first-order effects of massive dissociation of oceanic methane hydrate: Geology, v. 25, n. 3, p. 259–62,http://dx.doi.org/10.1130/0091-7613(1997)0250259:ABOGIT�2.3.CO;2

Diefendorf, A. F., Mueller, K. E., Wing, S. L., Koch, P. L., and Freeman, K. H., 2010, Global patterns in leaf13C discrimination and implications for studies of past and future climate: Proceedings of the NationalAcademy of Sciences of the United States of America, v. 107, n. 13, p. 5738–5743, http://dx.doi.org/10.1073/pnas.0910513107

Domingo, L., Lopez-Martınez, N., Leng, M. L., and Grimes, S. T., 2009, The Paleocene-Eocene ThermalMaximum record in the organic matter of the Claret and Tendruy continental sections (South-centralPyrenees, Lleida, Spain): Earth and Planetary Science Letters, v. 281, n. 3–4, p. 226–237, http://dx.doi.org/10.1016/j.epsl.2009.02.025

Ehleringer, J. R., Buchmann, N., and Flanagan, L. B., 2000, Carbon isotope ratios in belowground carboncycle processes: Ecological Applications, v. 10, n. 2, p. 412–422, http://dx.doi.org/10.1890/1051-0761(2000)010[0412:CIRIBC]2.0.CO;2

Ekart, D. D., Cerling, T. E., Montanez, I. P., and Tabor, N. J., 1999, A 400 million year carbon isotope recordof pedogenic carbonate: Implications for paleoatmospheric carbon dioxide: American Journal ofScience, v. 299, n. 10, p. 805–827, http://dx.doi.org/10.2475/ajs.299.10.805

Foreman, B. Z., Heller, P. L., and Clementz, M .T., 2012, Fluvial response to abrupt global warming at thePaleocene/Eocene boundary: Nature, v. 491, p. 92–95, http://dx.doi.org/10.1038/nature11513

Fricke, H. C., and Wing, S. L., 2004, Oxygen isotope and paleobotanical estimates of temperature and18O-latitude gradients over North America during the early Eocene: American Journal of Science, v.304, n. 7, p. 612–635, http://dx.doi.org/10.2475/ajs.304.7.612

Garten, C. T., Jr., Cooper, L. W., Post, W. M., III, and Hanson, P. J., 2000, Climate controls on forest soil Cisotope ratios in the southern Appalachian Mountains: Ecology, v. 81, p. 1108–1119, http://dx.doi.org/10.1890/0012-9658(2000)081[1108:CCOFSC]2.0.CO;2

Giardina, C. P., and Ryan, M. G., 2000, Evidence that decomposition rates of organic carbon in mineral soildo not vary with temperature: Nature, v. 404, p. 858–861, http://dx.doi.org/10.1038/35009076

Handley, L., O’Halloran, A., Pearson, P. N., Hawkins, E., Nicholas, C. J., Schouten, S., McMillan, I. K., andPancost, R. D. 2012, Changes in the hydrologic cycle in tropical East Africa during the Paleocene-Eocene Thermal Maximum: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 329–330, p. 10–21,http://dx.doi.org/10.1016/j.palaeo.2012.02.002

Hanson, P. J., Edwards, N. T., Garten, C. T., and Andrews, J. A., 2000, Separating root and soil microbialcontributions to soil respiration: A review of methods and observations: Biogeochemistry, v. 48, n. 1, p.115–146, http://dx.doi.org/10.1023/A:1006244819642

Harrington, G., and Jardine, P., 2012, Bighorn Basin Coring Project: Palynofloral changes and taphonomythrough the Paleocene-Eocene Thermal Maximum in the Bighorn Basin, Wyoming, USA: San Fran-cisco, AGU, Abstract 1469447 presented at 2012 Fall Meeting.

Higgins, J. A., and Schrag, D. P., 2006, Beyond methane: Toward a theory for the Paleocene-Eocene ThermalMaximum: Earth and Planetary Science Letters, v. 245, n. 3–4, p. 523–537, http://dx.doi.org/10.1016/j.epsl.2006.03.009

Karhu, K., Auffret, M. D., Dungait, J. A., Hopkins, D. W., Prosser, J. I., Singh, B. K., Subke, J.-A., Wookey,P. A., Årgen, G. I., Sebastia, M.-T., Gouriveau, F., Bergkvist, G., Meir, P., Nottingham, A. T., Salinas, N.,and Hartley, I. P., 2014, Temperature sensitivity of soil respiration rates enhanced by microbialcommunity response: Nature, v. 513, p. 81–84, http://dx.doi.org/10.1038/nature13604

Kennett, J. P., and Stott, L. D., 1991, Abrupt deep-sea warming, palaeoceanographic changes and benthicextinctions at the end of the Palaeocene: Nature, v. 353, p. 225–229, http://dx.doi.org/10.1038/353225a0

Knorr, W., Prentice, I. C., House, J. I., and Holland, E. A., 2005, Long-term sensitivity of soil carbon turnoverto warming: Nature, v. 433, p. 298–301, http://dx.doi.org/10.1038/nature03226

Koch, P. L., 1998, Isotopic reconstruction of past continental environments: Annual Review of Earth andPlanetary Sciences, v. 26, p. 573–613, http://dx.doi.org/10.1146/annurev.earth.26.1.573

Koch, P. L., Zachos, J. C., and Gingerich, P. D., 1992, Correlation between istope record in marine andcontinental resevoirs near the Paleocene/Eocene boundary: Nature, v. 358, p. 319–322, http://dx.doi.org/10.1038/358319a0

Kraus, M. J., and Riggins, S., 2007, Transient drying during the Paleocene-Eocene Thermal Maximum(PETM): Analysis of paleosols in the bighorn basin, Wyoming: Palaeogeography, Palaeoclimatology,Palaeoecology, p. 245, n. 3–4, v. 444–461, http://dx.doi.org/10.1016/j.palaeo.2006.09.011

Kraus, M. J., McInerney, F. A., Wing, S. L., Secord, R., Bacynscki, A. A., and Bloch, J. I., 2013, Paleohydrologicresponse to continental warming during the Paleocene-Eocene Thermal Maximum: Palaeogeography,Palaeoclimatology, Palaeoecology, v. 370, p. 196–208, http://dx.doi.org/10.1016/j.palaeo.2012.12.008

Kuzyakov, Y., 2006, Sources of CO2 efflux from soil and review of partitioning methods: Soil Biology andBiogeochemistry, v. 38, n. 3, p. 425–448 http://dx.doi.org/10.1016/j.soilbio.2005.08.020

Liang, C., and Balser, T. C., 2012, Warming and nitrogen deposition lessen microbial residue contribution tosoil carbon pool: Nature Communications, v. 3, p. 1222, http://dx.doi.org/10.1038/ncomms2224

Lopez-Martınez, N., and Pelaez-Campomanes, P., 1999, New mammals from south-central Pyrenees (TrempFormation, Spain) and their bearing on late Paleocene marine-continental correlations: Bulletin de alSociete Geologique de France, v. 170, n. 5, p. 681–696.

Magioncalda, R., Dupuis, C., Smith, T., Steurbat, E., and Gingerich, P. D., 2004, Paleocene-Eocene carbonisotope excursion in organic carbon and pedogenic carbonate: Direct comparison in a continentalstratigraphic section: Geology, v. 32, n. 7, p. 553–556, http://dx.doi.org/10.1130/G20476.1

McInerney, F. A., and Wing, S. L., 2011, The Paleocene-Eocene Thermal Maximum: A perturbation ofcarbon cycle, climate, and biosphere with implications for the future: Annual Review of Earth andPlanetary Sciences, v. 39, p. 489–516, http://dx.doi.org/10.1146/annurev-earth-040610-133431


Nemani, R. R., Keeling, C. D., Hashimoto, H., Jolly, W. M., Piper, S. C., Tucker, C. J., Myneni, R. B., andRunning S. W., 2003, Climate driven increases in global terrestrial net primary production from1982-1999: Science, v. 300, n. 5625, p. 1560–1563, http://dx.doi.org/10.1126/science.1082750

Pagani, M., Caldiera, K., Archer, D., and Zachos, J. C., 2006, An ancient carbon mystery: Science, v. 314, n.5805, p. 1556–1557, http://dx.doi.org/10.1126/science.1136110

Pagani, M., Pedentchouk, N., Huber, M., Sluijs, A., Schouten, S., Brinkhuis, H., Sinninghe Damaste, J. S.,Dickens, G. R., Expedition 302 Scientists, Backman, J., Clemens, S., Cronin, T., Eynaud, F., Gattacceca,J., Jakobsson, M., Jordan, R., Kaminski, M., King, J., Koc, N., Martinez, N. C., McInroy, D., Moore, T. C.,Jr., O’Regan, M., Onodera, J., Palike, H., Rea, B., Rio, D., Sakamoto, T., Smith, D. C., St. John, K. E. K.,Suto, I., Suzuki, N., Takahashi, K., Watanabe, M., and Yamamoto, M., 2006, Arctic hydrology duringglobal warming at the Paleocene/Eocene thermal maximum: Nature, v. 442, p. 671– 675,http://dx.doi.org/10.1038/nature05043

Raich, J. W., and Schlesinger, W. H., 1992, The global carbon dioxide flux in soil respiration and itsrelationship to vegetation and climate: Tellus B, v. 44, n. 2, p. 81–99, http://dx.doi.org/10.1034/j.1600-0889.1992.t01-1-00001.x

Retallack, G. J., 2005, Pedogenic carbonate proxies for amount and seasonality of precipitation in paleosols:Geology, v. 33, n. 4, p. 333–336, http://dx.doi.org/10.1130/G21263.1

Romanek, C. S., Grossman, E. L., and Morse, J. W., 1992, Carbon isotopic fractionation in synthetic aragoniteand calcite: Effects of temperature and precipitation rate: Geochimica et Cosmochimica Acta, v. 56, n. 1,p. 419–430, http://dx.doi.org/10.1016/0016-7037(92)90142-6

Samanta, A., Bera, M. K., Ghosh, R., Bera, S., Filley, T., Pande, K., Rathore, S. S., Rai, J., and Sarkar, A., 2013,Do the large carbon isotopic excursions in terrestrial organic matter across the Paleocene-Eoceneboundary in India indicate intensification of tropical precipitation?: Palaeogeography, Palaeoclimatol-ogy, Palaeoecology, v. 387, p. 91–103, http://dx.doi.org/10.1016/j.palaeo.2013.07.008

Schimel, D. S., Braswell, B. H., Holland, E. A., McKeown, R., Ojima, D. S., Painter, T. H., Parton, W. J., andTownsend, A. R., 1994, Climatic, edaphic and biotic controls over storage and turnover of carbon insoils: Global Biogeochemical Cycles, v. 8, n. 3, p. 279–293, http://dx.doi.org/10.1029/94GB00993

Schlesinger, W. H., 1997, Biogeochemistry: An Analysis of Global Change: San Diego, California, AcademicPress, 588 p.

Schmitz, B., and Pujalte, V., 2003, Sea-Level, humidity and land-erosion records across the initial Eocenethermal maximum from a continental-marine transect in northern Spain: Geology, v. 31, n. 8, p.689–69, http://dx.doi.org/10.1130/G19527.1

–––––– 2007, Abrupt increase in seasonal extreme precipitation at the Paleocene-Eocene boundary: Geology,v. 35, n. 3, p. 215–218, http://dx.doi.org/10.1130/G23261A.1

Schubert, B. A., and Jahren, A. H., 2013, Reconciliation of marine and terrestrial carbon isotope excursionsbased on changing atmospheric CO2 levels: Nature Communications, v. 4, p. 1653, http://dx.doi.org/10.1038/ncomms2659

Smith, F. A., Wing, S. L., and Freeman, K. H., 2007, Magnitude of the carbon isotope excursion at thePaleocene-Eocene Thermal Maximum: The role of plant community change: Earth and PlanetaryScience Letters, v. 262, n. 1–2, p. 50–65, http://dx.doi.org/10.1016/j.epsl.2007.07.021

Snell, K. E., Thrasher, B. L., Eiler, J. M., Koch, P. L., Sloan, L. C., and Tabor, N. J., 2013, Hot summers in theBighorn Basin during the early Paleogene: Geology, v. 41, n. 1, p. 55–58, http://dx.doi.org/10.1130/G33567.1

Soil Survey Staff, 2010, Keys to Soil Taxonomy, 11th edition: Washington, D. C., USDA-Natural ResourcesConservation Service, 346 p.

Taneva, L., Pippen, J. S., Schlesinger, W. H., and Gonzalez-Meler, M. A., 2006, The turnover of carbon poolscontributing to soil CO2 and soil respiration in a temperature forest exposed to elevated CO2concentration: Global Change Biology, v. 12, n. 6, p. 983–994, http://dx.doi.org/10.1111/j.1365-2486.2006.01147.x

Teixell, A., and Munoz, J. A., 2000, Evolucion tectono-sedimentaria del Pirieno merdional durante elTerciario: Una sıntesis basada en la transversal del rıo Noguera Ribagorcana: Revista de la SociedadGeologica de Espana, v. 13, n. 2, p. 251–264.

Tipple, B. J., Meyers, S. R., and Pagani, M., 2010, Carbon isotope ratio of Cenozoic CO2: A comparativeevaluation of available geochemical proxies: Paleoceanography, v. 25, n. 3, PA001851, http://dx.doi.org/10.1029/2009PA001851

Trumbore, S. E., Chadwick, O. A., and Amundson, R., 1996, Rapid exchange between soil carbon andatmospheric carbon dioxide driven by temperature change: Science, v. 272, n. 5260, p. 393–396,http://dx.doi.org/10.1126/science.272.5260.393

Tucker, G. E., and Slingerland R., 1997, Drainage basin responses to climate change: Water ResourcesResearch, v. 33, n. 8, p. 2031–2047, http://dx.doi.org/10.1029/97WR00409

VanDeVelde, J. H., Bowen, G. J., Passey, B. H., and Bowen, B. B., 2013, Climatic and diagenetic signals in thestable isotope geochemistry of dolomitic paleosols spanning the Paleocene-Eocene boundary: Geo-chemica et Cosmochimica Acta, v. 109, p. 254–276, http://dx.doi.org/10.1016/j.gca.2013.02.005

Vitousek, P. M., and Farrington H., 1997, Nutrient limitation and soil development: Experimental test of abiogeochemical theory: Biogeochemistry, v. 37, n. 1, p. 63–75, http://dx.doi.org/10.1023/A:1005757218475

Vitousek, P. M., and Howarth, R. W., 1991, Nitrogen limitation on land and in the sea: How can it occur?:Biogeochemistry, v. 13, n. 2, p. 87–115, http://dx.doi.org/10.1007/BF00002772

Whiticar, M. J., 1999, Carbon and hydrogen isotope systematics of bacterial formation and oxidation ofmethane: Chemical Geology, v. 161, n. 1–3, p. 291–314, http://dx.doi.org/10.1016/S0009-2541(99)00092-3

Wing, S. L., Harrington, G. J., Smith, F. A., Bloch, J. I., Boyer, D. M., and Freeman, K. H., 2005, Transient


http://dx.doi.org/10.1038/nature05043

floral change and rapid global warming at the Paleocene-Eocene boundary: Science, v. 310, n. 5750, p.993–996, http://dx.doi.org/10.1126/science.1116913

Wynn, J. G., 2007, Carbon isotope fractionation during decomposition of organic matter in soils andpaleosols: Implications for paleoecological interpretation of paleosols: Palaeogeography, Palaeoclima-tology, Palaeoecology, v. 251, n. 3–4, p. 437–448, http://dx.doi.org/10.1016/j.palaeo.2007.04.009

Wynn, J. G., Bird, M. I., and Wong, V. N. L., 2005, Rayleigh distillation and the depth profile of 13C/12C ratiosof soil organic carbon from soils of disparate textures in Iron Range National Park, Far NorthQueensland, Australia: Geochimica et Cosmochimica Acta, v. 69, n. 8, p. 1961–1973, http://dx.doi.org/10.1016/j.gca.2004.09.003

Zachos, J. C., Pagani, M., Sloan, L., Thomas, E., and Billups, K., 2001, Trends, rhythms and aberrations inglobal climate 65 Ma to present: Science, v. 292, p. 686–693, http://dx.doi.org/10.1126/sci-ence.1059412

Zachos, J. C., Rohl, U., Schellenberg, S. A., Sluijs, A., Hodell, D. A., Kelly, D. C., Thomas, E., Nicolo, M., Rafii,I., Lourens, L. J., McCarren, H., and Kroon, D., 2005, Rapid acidification of the ocean during thePaleocene-Eocene Thermal Maximum: Science, v. 308, n. 5728, p. 1611–1615, http://dx.doi.org/10.1126/science.1109004

Zeebe, R. E., Zachos, J. C., and Dickens, G. R., 2009, Carbon dioxide forcing alone insufficient to explainPalaeocene- Eocene Thermal Maximum warming: Nature Geoscience, v. 2, p. 576–580, http://dx.doi.org/10.1038/ngeo578


POSITIVE FEEDBACK DRIVES CARBON RELEASE …jcotton/Publications_files/Cotton_et_al...2 to the...

Documents

Transcript of POSITIVE FEEDBACK DRIVES CARBON RELEASE …jcotton/Publications_files/Cotton_et_al...2 to the...