dating climatic change in hot deserts using desert varnish on meteorite finds

Dating climatic change in hot deserts using desert varnish onmeteorite ¢nds

M.R. Lee a;�, P.A. Bland b;1

a Division of Earth Sciences, University of Glasgow, Lilybank Gardens, Glasgow G12 8QQ, UKb Planetary and Space Sciences Research Institute, Open University, Milton Keynes MK7 6AA, UK

Received 15 May 2002; received in revised form 5 July 2002; accepted 6 November 2002

Abstract

A thin coating of desert varnish occurs on Forrest 009 and Nurina 004, both equilibrated ordinary chondrite (L6)finds from the Nullarbor Plain, Australia. This finely laminated deposit is chemically and petrographicallycomparable to the varnish found on terrestrial rocks. Forrest 009, which has a terrestrial age of 5.9 kyr, has a 100^130Wm thick coating of desert varnish that has a laterally consistent chemical microstratigraphy comprising a narrow Ba-and Mn-poor lower region, a thick Ba- and Mn-rich central area and a narrow outer zone almost devoid of bothcations. The interior of the meteorite contains Fe-oxide and oxyhydroxide veins that have formed by chemicalweathering of metals and sulphides. As these veins do not cross-cut the varnish, it must have accreted rapidly relativeto the weathering rate of the meteorite. The 9 70 Wm thick varnish on Nurina 004, which has a terrestrial age of 33.4kyr, lacks a consistent chemical microstratigraphy, but it is cross-cut by Fe-oxide and oxyhydroxide veins, some ofwhich have supplied Fe to the varnish. This implies that the chemical weathering rate of Nurina 004’s interior wasslow in comparison to the accretion rate of the varnish. The petrography and chemical composition of varnish onForrest 009 indicates that this meteorite may have resided in a relatively humid environment for most of its 5.9 kyrterrestrial history and that the Nullarbor recently became more arid. This conclusion supports results from an analysisof Fe-bearing weathering products in the interior of the meteorite by Mo«ssbauer spectroscopy, which also indicatethat Forrest 009 experienced an early period of rapid weathering under relatively humid conditions. The petrographyof varnish on Nurina 004 shows that the interior of the meteorite weathered relatively slowly, probably because it fellduring an arid time, which is again in agreement with previous Mo«ssbauer spectroscopy results. Results from bothmeteorites are in agreement with palaeoclimate data derived from a number of other proxies. The implications of thiswork are that the large number of meteorites that have been collected from several hot deserts of the world may be apowerful source of information on climate change over the last 30^35 kyr.8 2002 Elsevier Science B.V. All rights reserved.

Keywords: desert varnish; ordinary chondrite; meteorite; weathering; Nullarbor

0012-821X / 02 / $ ^ see front matter 8 2002 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 2 ) 0 1 0 7 8 - 6

* Corresponding author. Tel. : +44-141-330-2634; Fax: +44-141-330-4817.E-mail address: [email protected] (M.R. Lee).

1 Present address: Department of Earth Science and Engineering, Imperial College, Prince Consort Road, London SW7 2BP, UK.

EPSL 6487 8-1-03

Earth and Planetary Science Letters 206 (2003) 187^198

www.elsevier.com/locate/epsl

mailto:[email protected]

http://www.elsevier.com/locate/epsl

1. Introduction

Determining the nature and chronology of cli-mate change during the Quaternary is currentlyan extremely important area of research that in-corporates many disciplines. Earth scientists haveconcentrated on identifying ¢nely layered sedi-mentary sequences, deposited in terrestrial or ma-rine environments, that record past climates bychemical, isotopic, mineralogical or palaeontolog-ical means. One ¢nely laminated deposit that hasattracted considerable recent interest is desert var-nish (see review in [1]). This is a rock coating thatis predominantly composed of clay minerals ce-mented by Mn^Fe oxyhydroxides [2,3]. Desertvarnish accretes very slowly (6 1^40 Wm/kyr)and rarely exceeds 200 Wm in thickness regardlessof the duration of subaerial exposure [4]. Thepresence of anthropogenic pollutants in recentlyformed varnish indicates that most constituentsare atmospherically derived [5,6] and precipitationplays a major role [1]. Bacteria are also probablyimportant in ¢xing the Mn and Fe [2,3]. Despitethe variability of accretion rates, even betweensamples from the same locality [4], the chemicalmicrostratigraphy of varnish may be correlatedon a regional scale. Variations in Ba and Mnconcentrations, which produce colour banding inultrathin sections examined by transmitted light,are believed to re£ect £uctuations in humidity ofthe environment and so provide a proxy for cli-mate change [1,2,7]. Despite a considerable bodyof research that supports the utility of varnish asa climate proxy, detailed studies of varnish from anumber of sites within a limited geographical areahave shown that the chemical microstratigraphymay vary considerably [8]. A detailed discussionof this issue is beyond the scope of the currentstudy and here we adopt the most recent model[1]. In order for data extracted from varnish to beof any practical use in analysing climate change, itis necessary to study samples derived from rocksurfaces whose duration of subaerial exposure isprecisely known. This necessity of using datedsurfaces has signi¢cantly limited the applicationof desert varnish in climate research. Meteoritesof known terrestrial ages provide dated surfacesthat may be useful in these studies.

Meteorites recovered from hot deserts are an-other potential source of palaeoclimate data. Ithas been demonstrated in [9,10] that a climatesignal can be extracted from some terrestriallyweathered meteorites (equilibrated ordinary chon-drites) by analysing two properties : (i) the ratiosof di¡erent valence states of iron in weatheringproducts and (ii) the ratios of paramagnetic andmagnetically ordered Fe3þ. The ratio of Fe0 andFe2þ to Fe3þ can be measured using Mo«ssbauerspectroscopy. When equilibrated ordinary chon-drites fall to the Earth all iron is present as Fe0

and Fe2þ but it is progressively oxidised to Fe3þ

during chemical weathering. The weathering ofhot desert meteorites takes place in two distinctstages [9^11]. Prior to terrestrial weathering theequilibrated ordinary chondrites have a high in-tergranular porosity (average of 15% [10]).Weathering agents (atmospheric gas and water)readily gain access to the interior of the meteoritesvia these pores. Fe,Ni metal and troilite is rapidlyconverted to Fe-oxide/oxyhydroxides, which oc-clude the intergranular pores and coat, or ‘ar-mour’, mineral grains in a process termed ‘passi-vation’. Following passivation, weathering ratesslow dramatically [11]. Meteorites that have fallenin hot deserts during relatively humid periods (in-ferred from other palaeoclimate indicators such asformer lake levels and periods of speleothem for-mation) had been more heavily weathered duringthe pre-passivation phase than those that havefallen during more arid times [9,10]. In additionto quantifying Fe3þ concentrations, Mo«ssbauerspectroscopy can also be used to measure theabundance of paramagnetic Fe3þ and magneti-cally ordered Fe3þ. Paramagnetic Fe3þ is associ-ated with the minerals akagene¤ite, goethite or le-pidocrocite whereas the magnetically orderedFe3þ is associated with maghemite and magnetite.These data are expressed as a ratio of paramag-netic Fe3þ/total Fe3þ, where total Fe3þ is para-magnetic Fe3þ plus magnetically ordered Fe3þ

[9]. These authors suggest that weathering duringrelatively humid periods will produce a lower val-ue of paramagnetic Fe3þ/total Fe3þ and this ini-tial mineralogy is preserved through subsequentperiods of weathering.

This paper describes results of a petrographic

EPSL 6487 8-1-03

M.R. Lee, P.A. Bland / Earth and Planetary Science Letters 206 (2003) 187^198188

and chemical study of varnish that coats two me-teorites recovered from the Nullarbor desert, Aus-tralia. The Nullarbor is an excellent location forthis study because the surface has been stable forthe last V30 kyr and during this time the regionhas experienced a number of humid^arid cyclesthat have been studied using proxies includingspeleothems and former lake levels. Results ofthis work show that the varnish and its parentmeteorite are both valuable sources of palaeocli-matic data and that analysing varnish on meteor-ites has a number of advantages over using var-nished terrestrial rocks. In addition, we show thatthe varnish may provide information on the ter-restrial history of the meteorite, for example therelative timing of periods of burial and exhuma-tion.

2. Materials and methods

The two meteorites discussed here, Forrest 009and Nurina 004, were both recovered from theNullarbor Plain, Australia. Five pieces of Forrest009 (L6), totalling 1 kg, were recovered in 1979(latitude S: 30‡09P, longitude E: 128‡05P) [12].One piece of Nurina 004 (L6) weighing 28.3 gwas recovered in 1986 (latitude S: 30‡47P, longi-tude E: 126‡27P) [13]. These two meteorites wereselected for study for a number of reasons.Firstly, the terrestrial ages of these meteoriteshave been determined and weathering productsin their interiors have also been studied in detail[9]. Secondly, both meteorites have a well-pre-served sequence of desert varnish. Lastly, the de-gree of weathering of the interior of Forrest 009is comparable to that of Nurina 004, althoughthe two meteorites have very di¡erent terrestrialages (Fig. 1). Work by Bland et al. [9] suggeststhat this similarity in the degree of weatheringis because Forrest 009 fell during a relativelyhumid period whereas Nurina 004 fell whenthe Nullarbor was much more arid. Thus, thesetwo meteorites provide an excellent test of theutility of desert varnish for describing climatechange.

Double-polished thin sections, prepared tostandard petrographic thickness (V30 Wm), were

made of Forrest 009 and Nurina 004. Backscat-tered electron (BSE) images were acquired using aCambridge Instruments S360 SEM operated at 20kV. X-ray maps of the varnish and underlyingmeteorite were acquired using an Oxford Instru-ments ISIS system attached to the SEM. Quanti-tative chemical analyses of the varnish on Forrest009 were obtained using a Cameca SX50 wave-length-dispersive electron probe operated at 15kV with a 10 nA beam current. The spot wasdefocused to 10 Wm in diameter to limit volatileloss from beam-sensitive constituents and to pro-vide averaged analyses of the micrometre-scalelaminations.

Fig. 1. Diagram illustrating data from nine L and LL ordi-nary chondrites ¢nds from the Nullarbor studied in [9]. Foreach ¢nd the degree of oxidation and ratio of paramagneticFe3þ to total Fe3þ (quanti¢ed by Mo«ssbauer spectroscopy) isshown, together with the 14C terrestrial age. The shadedareas represent humid periods identi¢ed in [9] from otherproxies including lake levels and periods of speleothemgrowth. Error bars on each datapoint represent the errors in14C age determinations (typically T 1.3 kyr).

EPSL 6487 8-1-03

M.R. Lee, P.A. Bland / Earth and Planetary Science Letters 206 (2003) 187^198 189

Fig. 2. BSE images of Forrest 009. Olivine and pyroxene grains are both light grey in these images and are di⁄cult to distinguish owing to their similar meanatomic number, but feldspar grains are dark grey. (a) The interior of the meteorite showing primary silicate grains (mainly olivine and pyroxene) cut by twoFe-oxide/oxyhydroxide veins (oriented NE-SW in the image). Silicates between the two veins have been partially dissolved, leaving a network of small pores(arrowed). (b) Desert varnish (D) within a microbasin in the external surface of the meteorite. Note how the varnish thickens into the microbasin, althoughparts on the right and left hand sides may have been partially removed during thin section preparation. (c) Desert varnish (D) coating the external surface ofthe meteorite. Note how many of the Fe-oxide/oxyhydroxide veins (white) are oriented roughly perpendicular to the former outer surface of the meteorite andterminate abruptly at the base of the varnish. (d) The interface between an olivine grain (medium grey, upper right) and heavily oxidised Fe,Ni metal (white,lower half of image). Much of the interior of the olivine has been dissolved to leave pores (black). These pores are bridged by narrow Fe-oxide/oxyhydroxideveins, which produce a ‘skeletal’ texture and demonstrate that silicate dissolution postdates at least one stage of vein formation.

EPSL

64878-1-03

M.R.Lee,

P.A.Bland

/Earth

andPlanetary

Science

Letters

206(2003)

187^198190

3. Chemical and mineralogical composition ofmeteorite interiors

Both meteorites are L6 equilibrated ordinarychondrites. The average mineralogy of L class or-dinary chondrites is 44.8% olivine, 24.2% ortho-pyroxene, 5.0% clinopyroxene, 10.3% plagioclase8.4% Fe,Ni metal, 5.8% troilite [14]. Forrest 009

has a 14C terrestrial age of 5.9T 1.3 kyr (B.P.) andis relatively highly weathered (oxidation of38.4%), whereas Nurina 004 has a 14C age of33 400T 2300 kyr (B.P.) but has undergone acomparable degree of weathering (oxidation of40.4%; visual weathering grade ‘C’) [10,13] (Fig.1). The values of paramagnetic Fe3þ/total Fe3þ

for Forrest 009 and Nurina 004 are 0.54 and

Fig. 3. Image and corresponding X-ray maps of the outer part of Nurina 004. Lighter grey tones in the X-ray maps correspondto greater numbers of counts. (a) BSE image showing the interior of the meteorite, which contains Fe-oxide/oxyhydroxide veins(V) and pseudomorphs of metal and sulphide grains (P). This image features a large ‘pocket’ ¢lled with rounded grains (de-lineated by arrows). The outer surface of the meteorite (top edge of the image) is coated with desert varnish. (b) Al KK X-raymap showing that with the exception of small grains of feldspar in the interior of the meteorite most Al occurs within the var-nish that coats its external surface. Some very Al-rich areas within the pocket (arrowed) are discrete grains of varnish. Desertvarnish also occurs between rounded grains within the ‘pocket’. (c) Si KK X-ray map. In this image grains of quartz within thepocket can be recognised as small areas with high counts (white). This X-ray map also allows grains of olivine (Ol) in the interi-or of the meteorite to be distinguished from orthopyroxene (Px). (d) Fe KK X-ray map illustrating the abundance of Fe-oxide/oxyhydroxide veins in the interior of the meteorite (white/light grey). Fe-oxide/oxyhydroxide also occurs between rounded grainsin the ‘pocket’.

EPSL 6487 8-1-03


0.70, respectively [9] (Fig. 1). More data on thedistribution of total Fe between various primaryand secondary minerals and valence states arelisted in [9].

Almost all of the original Fe,Ni metal in Forrest009 has been replaced by Fe-oxide/oxyhydrox-ides, although considerable volumes of troilite re-main. Ramifying veins of Fe-oxide/oxyhydroxides

also cross-cut the interior of the meteorite (Fig.2a^c). Olivine and pyroxene grains adjacent to theFe-oxide/oxyhydroxide pseudomorphs and veinshave been partially dissolved to leave skeletal re-licts (Fig. 2d). The interior of Nurina 004 shows acomparable degree of terrestrial weathering toForrest 009 and again contains pseudomorphsand veins of Fe-oxide/oxyhydroxides (Fig. 3). Sig-

Fig. 4. BSE images of desert varnish from Nurina 004. (a) Finely lamiated varnish overlying heavily weathered clinopyroxene(CPX) grains. (b) A complex interface between varnish (upper part of the image, base and top delineated by arrows) and the me-teorite substrate (lower part). Most of the interior of the meteorite is composed of compositionally zoned olivine (Ol) that hasbeen partially replaced by euhedral Fe-oxide/oxyhydroxide crystals that are concentrated just beneath the varnish. A Fe-oxide/oxyhydroxide vein (V) can be traced from the interior of the meteorite into the varnish and may be associated with a Fe-richpatch in upper parts of the varnish (Fe). (c) A vein of Fe-oxide/oxyhydroxide (V) extends from the interior of the meteorite(base of the image) into the varnish (D) and connects with a Fe-rich band in the interior of the varnish (Fe). (d) Area of varnish(D, upper part of image) overlying a ‘pocket’ in the outer part of the meteorite that is ¢lled with rounded grains and cementedby Fe-oxide/oxyhydroxide. Most of the rounded grains are quartz, but some are of desert varnish. Varnish also ¢lls a ¢ssure atthe left hand side edge of the pocket and a Fe-oxide/oxyhydroxide vein (V) runs up the centre of the ¢ssure. Note that this vein,as well as other lower parts of the varnish, is truncated at an ‘unconformity’ (arrowed).

EPSL 6487 8-1-03


ni¢cantly, some veins are partially occluded by aniron sulphide (?pyrite or marcasite), demonstrat-ing that hydrous Fe-oxides and oxyhydroxides arenot the only weathering products of troilite. Thereis again evidence for dissolution of silicates toproducing skeletal relicts (Fig. 4a). Outer partsof the meteorite contain ‘pockets’ up to 0.5 mmdeep that are ¢lled with rounded grains V40^3Wm in diameter and cemented by Fe-oxide/oxy-hydroxides and desert varnish (Fig. 3). Therounded grains include desert varnish (Fig. 3b)and quartz (Fig. 3c).

4. Structure and composition of the desert varnish

Light microscope and BSE images show thatthe varnish on outer surfaces of Forrest 009 is100^130 Wm thick (Fig. 2b,c) and is thickest inlocal depressions, or ‘microbasins’, in the outersurfaces of the meteorite (Fig. 2b). Another sur-face of Forrest 009, probably that which rested onthe ground, is encrusted with a layer of mixedcarbonate and quartz. As the varnish directly

overlies olivine, pyroxene, Fe,Ni metal and troilitegrains, it has clearly formed directly on the interiorof the meteorite rather than on the fusion crust.Signi¢cantly, none of the Fe-oxide/oxyhydroxideveins that pervade the interior of the stone canbe traced into the varnish and they terminateabruptly at its base (Fig. 2b,c). A few quartzgrains occur at the interface of the outer meteoritesurface and the base of the varnish, but quartz isabsent from within the varnish itself. Some of theouter layers of the varnish have been exfoliated bydisplacive growth of Ca-carbonate within frac-tures (Fig. 5). These fractures are oriented bothparallel and perpendicular to layering in the var-nish. Those fractures that cut through lower levelsof the varnish appear to have been partially‘healed’, although contain Ca-carbonate in theupper levels (Fig. 5). Electron probe data showthat the varnish is dominated by Al, Si, Mn andFe (Table 1). Correlation coe⁄cients between allof the elements analysed are also listed in thetable. These data demonstrate two related elemen-tal groups: Si^Al^Mg^K and Ba^Mn. Data forFe are less clear, but this element is most strongly

Fig. 5. BSE image of a complete sequence of desert varnish from the centre of the ‘microbasin’ in Forrest 009. The black andwhite line marks the path of an electron probe traverse. Compositional data (right hand side) indicate that the centre of the var-nish layer is Ba- and Mn-rich whereas the outer edges are Ba- and Mn-poor. Note that the outer layers of the varnish are beingexfoliated by displacive growth of Ca-carbonate. Other fractures, which are oriented perpendicular to the layering, contain Ca-carbonate in upper parts of the varnish but appear to have been partially ‘healed’ lower down.

EPSL 6487 8-1-03


associated with the Si^Al^Mg^K group. Throughmost of its thickness the varnish is opaque intransmitted light apart from the outermost part,which is dark orange. BSE images show that thevarnish is laminated on the micrometre scale andthe outermost part, which is orange in transmittedlight, has a lower mean atomic number (Fig. 5).Where thickest, within local microbasins (Fig. 2c),the varnish has a narrow and discontinuous lowerzone that is relatively poor in Ba and Mn, a widemiddle zone that is Ba- and Mn-rich and a nar-row uppermost zone that is almost devoid of bothcations (Fig. 5).

The varnish on Nurina 004 is 9 70 Wm thick(Figs. 3 and 4) and has a complex relationship tothe parent meteorite. The interface between theouter surface of the meteorite and the varnishcan be abrupt (Fig. 3a), but veins of Fe-oxide/oxyhydroxide may also penetrate from the interi-or of the meteorite into the varnish (Fig. 4b,c).In some places these veins are abruptly trun-cated along an internal ‘unconformity’ (Fig. 4d),whereas elsewhere the vein connects with a Fe-rich layer within the varnish (Fig. 4b,c). Thechemical stratigraphy of the varnish, as revealedby BSE images, X-ray maps and linescans, iscomplex and highly variable so that a consistentpattern could not be recognised. Most of the var-nish is Ba- and Mn-rich although small Ba- andMn-poor areas occur in the lower parts of micro-basins.

5. Discussion

5.1. Previous work on varnished meteorites andvarnish from Australia

A number of varnished meteorite ¢nds havebeen previously described, including the Nami-bian ¢nd Gobabeb [15] and meteorites collectedfrom the Gold Basin strewn ¢eld (Mojave Desert)by Kring et al. [16,17]. Considerable work hasbeen undertaken on varnish in arid regions ofAustralia by Dragovich, predominantly in westernNew South Wales [18^22]. This author has notedthe ubiquity of varnished rock surfaces in aridAustralia. The varnish ranges from 50 to 150Wm in thickness and commonly shows a layeringof Mn-rich and Mn-poor zones when studied incross-section [19].

5.2. Accumulation of varnish on the Nullarbormeteorites

The presence of varnish on both meteoritesstudied and the absence of layers of coarse detri-tal material within the varnish support the asser-tion of Bland et al. [10] that the Nullarbor mete-orites have been subaerially exposed for most oftheir terrestrial history. As Bland et al. [10] alsonote, most of the Nullarbor ¢nds are too small tohave formed a crater on impact, which could alsohave led to burial. The absence of a fusion crust

Table 1Mean chemical composition of the Forrest 009 varnish and correlation coe⁄cients of the elements

Wt% Correlation coe⁄cients

SiO2 Al2O3 Fe2O3 NiO MnO MgO BaO CaO Na2O K2O

SiO2 30.8 1.00 0.77 0.45 30.04 30.12 0.60 30.11 30.29 0.04 0.78Al2O3 19.8 0.77 1.00 0.60 0.00 0.00 0.49 0.00 30.42 0.03 0.57Fe2O3 9.1 0.45 0.60 1.00 30.01 0.01 0.47 30.10 0.33 0.03 0.55NiO 0.2 30.04 0.00 30.01 1.00 0.22 30.05 0.27 0.00 0.00 30.08MnO 12.4 30.12 0.00 0.01 0.22 1.00 30.04 0.94 30.14 0.00 30.16MgO 1.9 0.60 0.49 0.47 30.05 30.04 1.00 30.10 30.18 0.01 0.58BaO 1.0 30.11 0.00 30.10 0.27 0.94 30.10 1.00 30.14 n.a. 30.43CaO 4.1 30.29 30.42 0.33 0.00 30.14 30.18 30.14 1.00 30.07 30.27Na2O 0.1 0.04 0.03 0.03 0.00 0.00 0.01 n.a. 30.07 1.00 0.03K2O 2.1 0.78 0.57 0.55 30.08 30.16 0.58 30.43 30.27 0.03 1.00Total 81.5

Number of analyses was 124 for all elements apart from Na2O (78 analyses) and BaO (45 analyses). n.a. denotes not applicablebecause Na2O and BaO were analysed in separate runs. Data in bold are signi¢cant correlations (R2 s 0.55).

EPSL 6487 8-1-03


beneath the varnish of both ¢nds suggests that ithas either weathered away or spalled o¡ soonafter impact. If the crust had persisted for tensor hundreds of years prior to being eroded fromthe outer surface of the meteorite some of theearliest layers of desert varnish may have beenlost. This possibility is di⁄cult to evaluate withthe information currently available.

The small rounded grains of quartz and desertvarnish within ‘pockets’ in the outermost parts ofNurina 004 are interpreted to be wind-blowngrains that accumulated within depressions inthe surface of the meteorite. Some early varnishoccurs between the detrital grains and has prob-ably cemented them (Fig. 3b). The presence ofdetrital grains of desert varnish within these pock-ets demonstrates that it was being actively erodedfrom rock surfaces elsewhere shortly after Nurina004 fell. Although the varnish on Forrest 009shows no evidence of abrasion, indicating that itmay have been accreting continuously since thefall of the meteorite, ‘unconformities’, presumablyproduced by localised erosion, occur within var-nish on Nurina 004. Signi¢cantly, the Fe-oxide/oxyhydroxide veins that are very common in theinterior of Forrest 009 do not cross-cut the var-nish and so must predate it. By contrast, Fe-ox-ide/oxyhydroxide veins within Nurina 004 com-monly penetrate into the varnish, showing thatthe interior of the meteorite was still undergoingweathering after the varnish had started to form.The potential palaeoclimatic signi¢cance of theseobservation is discussed later. The fact that someof the Fe-oxide/oxyhydroxide veins within Nurina004 connect with Fe-rich layers within the varnish(Fig. 4b,c), indicates that some components of thevarnish have been derived from within the parentrock.

The varnish on Forrest 009 and Nurina 004 ischemically comparable to that found on terrestrialrocks from hot deserts in the USA [8] and aridAustralia [20]. From electron probe analyses ofvarnish on terrestrial volcanic rocks, Reneau etal. [8] likewise identi¢ed the presence of a silicatephase (containing Si, Al, K and Mg), a Mn-bear-ing phase and a Fe-bearing phase. The 6 70 Wmand 100^130 Wm thicknesses of the Nurina 004and Forrest 009 varnishes, respectively, give ac-

cretion rates of 9 2^22 Wm/kyr. This is in goodagreement with the 3^15 Wm/kyr mean growthrates of varnish on young (6 30 kyr) substratesin the USA [4].

5.3. Palaeoclimate data from meteorite weatheringproducts

Forrest 009 and Nurina 004 are among a suiteof nine L and LL class ordinary chondrites fromthe Nullarbor whose terrestrial ages and degreesof oxidation were quanti¢ed by Bland et al. [9,10](Fig. 1). They found that when the degree of ox-idation of these ¢nds was compared with theirterrestrial age (2.2^33.4 kyr), some meteoritesshowed greater rates of weathering than others.The three ¢nds with terrestrial ages of 5.9^7.6kyr, which include Forrest 009, stand out as hav-ing been relatively highly weathered (Fig. 1) andalso have low values of paramagnetic Fe3þ/totalFe3þ (0.47^0.57) (Fig. 1). These data have beeninterpreted to indicate that the three meteoritesfell during a humid period [9]. By contrast someof the older ¢nds, including Nullarbor 012 andNurina 004, have relatively low degrees of weath-ering for their age and higher values of paramag-netic Fe3þ/total Fe3þ (Fig. 1), suggesting that theyfell during a more arid time [9].

Data from a number of sources con¢rms thatthe climate in the Nullarbor has changed over thelast 30 kyr, although it must be emphasised thatthroughout this time the climate has been essen-tially arid (Alex Bevan, personal communication,2002). Data from lake levels indicate that the Null-arbor was relatively humid from 25 to 20 kyr,then became more arid, with a minimum level at15 kyr, and returned to humid conditions withanother peak in lake levels at V7 kyr (B.P.)[23^25]. In addition, botanical evidence suggeststhat e¡ective precipitation in the Nullarbor in-creased from V180 to 250 mm from 10^8 to 5^4 kyr (Alex Bevan, personal communication,2002) [26]. A period of halite speleothem growthat 2.5 T 1.2 kyr has also been identi¢ed in [27].Humid periods identi¢ed by these independentclimate proxies are indicated in Fig. 1 (shadedareas) and show a good correspondence with thehumid periods identi¢ed by rapid rates of weath-

EPSL 6487 8-1-03


ering and low paramagnetic Fe3þ/total Fe3þ val-ues in meteorites that fell approximately at thosetimes [9].

5.4. Palaeoclimate data from the varnish

Palaeoclimate information derived from weath-ering products within Forrest 009 can be indepen-dently assessed by examining the chemical micro-stratigraphy of its varnish. The information thatvarnish may contain regarding past climates hasbeen the subject of a number of previous studiesand is still the source of considerable controversy.The most recent work [1] has concluded that var-iations in the concentration of Ba and Mn thougha sequence of desert varnish record changes in theambient humidity (or ‘wetness’) of the local envi-ronment. As these authors stress, one of the prin-cipal di⁄culties in using the technique morewidely, especially in areas outside the USA, isthe paucity of reliable dates for the duration ofsubaerial exposure of rock surfaces.

Most of the varnish on Forrest 009 is relativelyBa- and Mn-rich, but the innermost and especiallyoutermost parts are poor in both cations (Fig. 5).Interpreting these data using the model of [1]would suggest that Forrest 009 spent most of itsterrestrial history in a relatively humid environ-ment and conditions became more arid relativelyrecently. The relatively high degree of oxidationand abundance of magnetically ordered Fe3þ inForrest 009, and the two other meteorites thatfell at a similar time (Fig. 1), certainly do supportthe idea that Forrest 009 experienced humid con-ditions early in its terrestrial history. The fact thatFe-oxide/oxyhydroxide veins do not cut the var-nish (Fig. 2b,c) also con¢rms that the weatheringproducts formed rapidly in comparison to the ac-cretion rate of the varnish. Although there is aqualitative agreement between palaeoclimaticdata from the interior of the meteorite and fromthe chemical stratigraphy of the varnish, we fullyrecognise the limited value in drawing conclusionsfrom a single sample. One way in which theseconclusions could be strengthened is by datingthe carbonates that ¢ll fractures in the outer partsof the varnish (Alex Bevan, personal communica-tion, 2002).

Nurina 004 lacks a clearly developed chemicalmicrostratigraphy to use for palaeoclimate recon-struction and also has evidence for abrasion (Fig.4d), again mitigating against ¢rm conclusionsbeing drawn from the varnish. However, theabundance of Fe-oxide/oxyhydroxide veins thatcross-cut the varnish does demonstrate that themeteorite was still undergoing chemical weather-ing even after accretion of tens of micrometres ofvarnish. This could indicate that the weatheringrate of the meteorite was relatively slow, suggest-ing that it fell during an arid period. This conclu-sion is supported by Mo«ssbauer spectroscopydata from meteorite weathering products [9] andagrees with independent climate records (Fig. 1).

6. Implications

Results of this study show that meteorite ¢ndshave a number of advantages over terrestrial rocksubstrates for the purpose of using desert varnishto study climate change. Firstly, a large numberof ¢nds have already been collected from hot des-erts, principally the Sahara (1508 ¢nds), Nullar-bor (s 280 ¢nds), and Roosevelt County, USA[28]. Secondly, many of these meteorites havebeen dated using 14C techniques, which have anerror of TV1^2 kyr. It is absolutely certain thatvarnish accretion started after the fall of the me-teorite; sampling geomorphologic sites can havemore uncertainties. Finds in the Nullarbor spanthe last 33 kyr and Saharan ¢nds range over acomparable time period (V35 kyr). As the £uxof meteorites is continuous, so long as a su⁄cientnumber of varnished ¢nds are available from onearea a complete chronology can be established. Bycontrast, ¢nding suitable terrestrial rocks in hotdeserts for radiocarbon dating is di⁄cult [1]. An-other advantage is that the varnished surface ofall meteorite ¢nds will come from an identicalmicroenvironment, a few centimetres above theground surface, and so di¡erences in topographywill not be a signi¢cant variable in the nature orrate of varnish accumulation between sites. Ordi-nary chondrites of a given type also vary little inoriginal mineralogy or bulk chemical composi-tion, regardless of where they fall. Thus, di¡er-

EPSL 6487 8-1-03


ences in rock substrate can be discounted as apotential factor a¡ecting the chemistry of varnishfrom di¡erent localities. Lastly, as the volume andmineralogy of weathering products within theequilibrated ordinary chondrites is determinedby climate soon after the fall of the meteorite,these properties can be used to cross-check pa-laeoclimate data derived from overlying varnish.

We recognise that studying desert varnish onmeteorite ¢nds has some inherent disadvantagesin comparison to terrestrial rocks. Not all ¢ndswill have a varnish and that on some may beunsuitable for the purpose (for example it mayhave been partially eroded, as is the case in Nur-ina 004). As they are low-lying, some meteoritesmay have been partially covered by soil or sedi-ment for a portion of their terrestrial residence,although this does not appear to have happenedwith regard to the two meteorites studied hereand may in fact be of use. For example Ashwal[29] interpreted the absence of varnish from thesurface of Korra Korrabes, a Namibian ¢nd, toindicate that it may have been covered by sandfor much of its terrestrial history. Lastly, mostauthors emphasise the importance of selectingsamples for study that have the most completevarnish record. Given the relatively small size ofindividual meteorites and the limited numbers of¢nds available, the potential for such selectivity isrestricted. However, as the outer surface of mete-orites is typically pitted, the result of ablationduring atmospheric entry, many microbasins willbe available for study even on a small sample.

7. Conclusions

This work has shown that desert varnish onextraterrestrial rocks can provide detailed infor-mation on changes in the climate of terrestrialenvironments. The advantages of using meteoritesfor understanding climate change over terrestrialrocks are the ease of dating their duration of res-idence on the Earth’s surface and the presence ofweathering products containing a climate signalwithin the meteorites themselves. As large num-bers of meteorites have already been recoveredfrom hot deserts in Africa, America and Austra-

lia, this approach has worldwide applicability.Further work is currently in progress to test thismodel, using ¢nds from the Acfer region of theSahara.

Acknowledgements

We would like to thank Robert MacDonald forassistance with electron probe analyses and JohnGileece for making the thin sections. We are alsovery grateful to Alex Bevan (Department of Earthand Planetary Sciences, Western Australian Mu-seum of Natural Science) for supplying the sampleof Nurina 004. This manuscript was signi¢cantlyimproved by helpful reviews from Alex Bevan,John Bridges and an anonymous reviewer. Thiswork was funded by the UK Natural Environ-ment Research Council.[BW]

References

[1] W. Broecker, T. Liu, Rock varnish: Recorder of desertwetness?, GSA Today 11 (2001) 4^9.

[2] R.S. Perry, J.B. Adams, Desert varnish evidence for cyclicdeposition of manganese, Nature 276 (1978) 489^491.

[3] D. Krinsley, Models of rock varnish formation con-strained by high resolution transmission electron micros-copy, Sedimentology 45 (1998) 711^725.

[4] T. Liu, W.S. Broecker, How fast does rock varnish grow?,Geology 28 (2000) 183^186.

[5] M. Fleischer, T. Liu, W. Broecker, A clue regarding theorigin of rock varnish, Geophys. Res. Lett. 26 (1999) 103^106.

[6] H. Bao, G.M. Michalski, M.H. Thiemens, Sulfate oxygen-17 anomalies in desert varnishes, Geochim. Cosmochim.Acta 65 (2001) 2029^2036.

[7] T. Liu, R.I. Dorn, Understanding the spatial variability ofenvironmental change in drylands with rock varnish mi-crolaminations, Ann. Assoc. Am. Geogr. 86 (1996) 187^212.

[8] S. Reneau, R. Raymond, C. Harrington, Elemental rela-tionships in rock varnish stratigraphic layers, Cima vol-canic ¢eld, California: Implications for varnish develop-ment and the interpretation of varnish chemistry, Am. J.Sci. 292 (1992) 684^723.

[9] P.A. Bland, A.S. Sexton, A.J.T. Jull, A.W.R. Bevan, F.J.Berry, D.M. Thornley, T.R. Astin, D.T. Britt, C.T. Pil-linger, Climate and rock weathering: A study of terrestrialage dated ordinary chondritic meteorites from hot desertregions, Geochim. Cosmochim. Acta 62 (1998) 3169^3184.

EPSL 6487 8-1-03


[10] P.A. Bland, A.W.R. Bevan, A.J.T. Jull, Ancient meteorite¢nds and the Earth surface environment, Quat. Res. 53(2000) 131^142.

[11] P.A. Bland, F.J. Berry, C.T. Pillinger, Rapid weatheringin Holbrook: An iron-57 Mo«ssbauer spectroscopy study,Meteor. Planet. Sci. 33 (1998) 127^129.

[12] A.L. Graham, The meteoritical bulletin, Meteoritics 25(1990) 59^70.

[13] F. Wlotzka, The meteoritical bulletin, no. 77, Meteoritics29 (1994) 891^897.

[14] H.Y. McSween, M.E. Bennett III, E. Jarosewich, Themineralogy of ordinary chondrites and implications forasteroid spectrophotometry, Icarus 90 (1991) 107^116.

[15] R.F. Fudali, F. Noonan, Gobabeb, a new chondrite: Thecoexistence of equilibrated silicates and unequilibratedspinels, Meteoritics 10 (1975) 31^39.

[16] D.A. Kring, A.J.T. Jull, P.A. Bland, The Gold Basinstrewn ¢eld, Mojave Desert, and its survival from thelate Pleistocene to the present, Workshop on Extraterr.Materials from Hot and Cold Deserts, LPI Contrib.997, 1999, pp. 44^45.

[17] D.A. Kring, A.J.T. Jull, L.R. McHargue, P.A. Bland,D.H. Hill, F.J. Berry, Gold Basin meteorite strewn ¢eld,Mojave Desert, northwest Arizona: Relic of a small latePleistocene impact event, Meteor. Planet. Sci. 36 (2001)1057^1066.

[18] D. Dragovich, The survival of desert varnish in subsur-face positions, western New South Wales, Australia,Earth Surf. Proc. Land. 9 (1984) 425^434.

[19] D. Dragovich, Desert varnish and environmental changenear Broken Hill, western New South Wales, Earth Sci.Rev. 25 (1988) 399^407.

[20] D. Dragovich, A preliminary electron probe study of mi-

crochemical variations in desert varnish in Western NewSouth Wales, Australia, Earth Surf. Proc. Land. 13 (1988)259^270.

[21] D. Dragovich, Fire, climate and the persistence of desertvarnish near Dampier, western Australia, Palaeogeogr.Palaeoclimatol. Palaeoecol. 111 (1994) 279^288.

[22] D. Dragovich, Microchemistry of small desert varnishsamples, western New South Wales, Australia, EarthSurf. Proc. Land. 23 (1998) 445^453.

[23] F.A. Street, A.T. Grove, Global maps of lake level £uc-tuation since 30,000 yr. BP, Quat. Res. 12 (1979) 83^118.

[24] J.M. Bowler, Glacial age aeolian events at high and lowlatitudes: A southern hemisphere perspective, in: E.M.Van Zinderen Bakker (Ed.), Antarctic Glacial Historyand World Palaeoenvironments, Balkema, 1978, pp.149^172.

[25] J.M. Bowler, G.S. Hope, J.N. Jennings, G. Singh, D.Walker, Late Quaternary climates of Australia and NewGuinea, Quat. Res. 6 (1976) 359^394.

[26] H.A. Martin, Palynology and historical ecology of somecave excavations in the Australian Nullarbor, Aust. J.Bot. 21 (1973) 283^316.

[27] A. Goede, R.S. Harmon, T.C. Atkinson, P.J. Rowe, Pleis-tocene climatic change in southern Australia and its ef-fects on speleothem deposition in some Nullarbor caves,J. Quat. Sci. 5 (1990) 29^38.

[28] M.M. Grady, Meteorites from cold and hot deserts: Howmany, how big and what sort? Workshop on Extraterr.Materials from Hot and Cold Deserts, LPI Contrib. 997,1999, pp. 36^40.

[29] L.D. Ashwal, Korra Korrabes, A new, large 3 chondritebreccia from Namibia, Meteor. Planet. Sci. 36 (2001)1027^1038.

EPSL 6487 8-1-03


dating climatic change in hot deserts using desert varnish on meteorite finds

Documents