Terra Rossa Genesis, Implications for Karst, and Eolian Dust: a Geodynamic Thread

Terra Rossa Genesis, Implications for Karst, and Eolian Dust: A Geodynamic Thread

Enrique Merino and Amlan Banerjee1

Department of Geological Sciences, Indiana University, Bloomington, Indiana 47405, U.S.A. (e-mail: [email protected])

ABSTRACT Although terra rossa has long been thought to form by residual dissolution of limestone and/or by accumulation of detrital mud, ash, or dust on preexisting karst limestones, we present conclusive new ﬁeld and petrographic evidence that terra rossa forms by replacement of limestone by authigenic clay at a moving metasomatic front several centimeters wide. The red clay’s major chemical elements, Al, Fe, and Si, probably come from dissolved eolian dust. The replacement of calcite by clay exhibits a serrated, microstylolitic texture that helps prove that replacement happens not by dissolution-precipitation, as conventional wisdom has it, but by pressure solution of calcite driven by the crystallization stress generated by the growth of clay crystals. The acid produced by the isovolumetric replacement of limestone by clay quickly dissolves out additional porosity/permeability in an adjacent slice of limestone within the front, triggering a reactive-inﬁltration instability that should, theoretically, convert the moving reaction front into a set of wormholes, then funnels, then sinks—the very karst morphology that in nature does contain the terra rossa itself. This beautifully explains why terra rossa and karst are associated.

Introduction Terra rossa clays are red claystones up to several have to be dissolved to yield a significant thickness meters thick and kilometers across that occur at of terra rossa. According to the detrital theory, terra the earth’s surface and are associated with karst rossa forms by accumulation of alluvial mud, vol- carbonates. There are terra rossas across southern canic ash, or eolian dust on limestones. A basic Europe and South Australia (where they support problem of the hypothesis is that it does not ac- vineyards), the Caribbean and surrounding lands, count for the worldwide association of terra rossa southern China, and elsewhere. The terra rossas in with karst carbonate rocks. Jamaica (Comer 1974) and southern France (Guen- The purpose of this article is to propose a new don and Parron 1985) grade into bauxite, an alu- theory of terra rossa formation, by authigenic re- minum ore. The two existing theories about their placement of the underlying limestone at a narrow origin, the residual and the detrital, are discussed reaction front. The new theory is based on field and in detail in “Previous Work on Terra Rossa.” Both petrographic evidence on the terra rossa at Bloom- theories have problems, and both neglect to focus ington, Indiana. Since the clay is authigenic, its on the contact between terra rossa and the under- major elements—Al, Si, and Fe—must come to the lying carbonate. According to the residual origin front as aqueous ions. We propose that these aque- hypothesis, terra rossa is the insoluble residuum ous ions probably result from dissolution of dust left by dissolution of limestone. Its main problem at the surface. Taking account of recent insights (pointed out by nearly every proponent of the de- into the physics of mineral replacement (summa- trital theory) is that limestones contain little or no rized in “Replacement Physics”) and into the dy- clay or other insoluble minerals (Ruhe et al. 1961; namics of moving reaction fronts, we then show Ruhe 1975; Comer 1976; Mee et al. 2004), so un- that the clay-for-carbonate replacement, because it realistically large thicknesses of limestone would generates acid, which dissolves out new porosity, should trigger a morphological reactive-infiltration Manuscript received July 2, 2007; accepted October 12, 2007. instability, which theoretically “specializes” in 1 E-mail: [email protected]. producing wormholes and funnels, precisely the

62 Journal of Geology TERRA ROSSA GENESIS 63 morphology characteristic of karst. That is, terra are the major elements of terra rossa clay. This rossa genesis and karst weathering appear to be cou- mass deficit, pointed out by Ruhe et al. (1961), Ruhe pled and mutually reinforcing phenomena that oc- (1975, p. 17–19), Comer (1974), Hall (1976), Olson cur at the base of the terra rossa, driven by the et al. (1980), Herwitz et al. (1996), Donovan (2002), aqueous solutes resulting from dissolution of de- Foster and Chittleborough (2003), Mee et al. (2004), trital dust by rain and soil water at the surface of Muhs et al. (2007), and others, is what gave rise to the terra rossa. the detrital theory. Second, the dissolution of lime- In the second article of this series (A. Banerjee stone into sets of funnels and sinkholes that char- and E. Merino, unpub. manuscript), we will present acterize karst weathering is attributed to the fo- a quantitative reaction-transport model of terra cusing of dissolution at supposed intersecting rossa formation based on the qualitative geochem- subvertical fractures (e.g., Thornbury 1954; White ical dynamics discovered and described here. The 1988, p. 111; Donovan 2002; Twidale 2004), but this quantitative model yields predictions of the front fails to explain why at least some holes (such as width, bizonality, and velocity; the predictions the one in the “floater” of fig. 6C; see “Discus- compare reasonably well to field, petrographic, and sion”), funnels, or sinks have no associated frac- paleomagnetic observations. Paleomagnetic obser- tures, and it does not help explain the association vations made on the Bloomington terra rossa, of terra rossa with karst. which allow us to date it and to estimate its rate The Detrital Origin. According to the detrital of formation, are presented in a third article (J. G. theory, terra rossa is an accumulation of detrital Meert, F. Pruett, and E. Merino, unpub. manu- material—alluvial mud (Ruhe et al. 1961; Hall script). 1976; Olson et al. 1980), volcanic ash (Comer 1974), or eolian dust (Yaalon and Ganor 1973; Herwitz et al. 1996; Yaalon 1997; Durn et al. 1999; Durn 2003; Previous Work on Terra Rossa Foster and Chittleborough 2003; Frumkin and Stein The Residual Origin. This theory was proposed 2004; Muhs et al. 2007)—on an (assumed preexist- early (de Lapparent 1930; Thornbury 1954) and sur- ing) karst limestone. Although photos in National vives in geomorphology textbooks. Of anecdotal in- Geographic (January 2004) of dust storms settling terest is Bardossy’s uncritical description of terra Saharan dust on southern Europe, satellite photos rossa research since the nineteenth century (Bar- of dust plumes traveling from the Sahara to the dossy 1982, p. 329–338, 350–351). Thornbury (1954, Caribbean, and extensive evidence obtained in the p. 319) simply took it for granted that terra rossa past 25 years (e.g., Yaalon 1997; Prospero et al. 2001, is the residual product of limestone dissolution by 2002; Muhs et al. 2007), plus evidence of eolian dust “descending groundwater,” the very dissolution supply to weathering profiles on silicate rocks (e.g., that is assumed to produce the depressions typical Brimhall et al. 1988; Kurtz et al. 2001), leave no of karst weathering. Moresi and Mongelli (1988), doubt that eolian dust settles on all surface rocks, after comparing chemical analyses of the insoluble including carbonates, in many parts of the world, residue left by dissolving Apulia limestones to the idea that the terra rossa itself is a detrital ac- chemical analyses of the terra rossa that occurs on cumulation (with or without later alteration) also the limestones, thought that the terra rossa is prob- has problems. First, it implicitly assumes that dust ably in part a residue from limestone dissolution. accumulates only on already karstified carbonates, Other recent authors (Durn et al. 1999; Delgado et which not only converts the terra rossa/karst as- al. 2003; Durn 2003), also on the basis of comparing sociation into a coincidence but also neglects that chemical analyses and comparisons of particle size Guendon and Parron (1985), based on excellent analyses, believe that only minor portions of spe- field evidence, demonstrated that the karst under cific terra rossa formations in southeastern Spain the terra rossa/bauxite at the bauxite type locality and Istria (Croatia) are residual products of lime- of Les Baux, southern France, had developed si- stone dissolution, with the rest being detrital. multaneously with the development of the bauxitic The residual theory of terra rossa origin has the horizons. (Their evidence was an isopach map positive aspect of explicitly linking terra rossa gen- showing that the greater thicknesses of bauxite oc- esis to karst formation, thus apparently explaining cur precisely above limestone sinks and the smaller their association, but it has two problems. First, thicknesses of it occur above the domes between limestone dissolution can yield only a fraction of sinks.) Second, it does not account for the common the observed thickness of terra rossa, because lime- occurrence of “floaters” in terra rossa—in situ stones contain little insoluble minerals to begin limestone blocks of any size completely sur- with and, in particular, little Si, Al, and Fe, which rounded by, and “floating” in, the red clay: how 64 E. MERINO AND A. BANERJEE could detrital mud or dust get into a limestone and meters thick that grows authigenically at its base isolate one such block? (Floaters are uncovered fre- by replacement of the underlying limestone. Of quently by developers in Bloomington. Many are course, the top portion of that claystone may later exhibited in front yards of private houses for their be—if it is not eroded first—pedologically altered interesting shapes, abounding in concavities; one into a soil. is shown in fig. 6C.) As noted, Ruhe et al.’s (1961) study (summarized Methodological Shortcomings. Aside from the by Ruhe [1975], p. 17–19) of the terra rossa of Ber- problems listed with both ideas of terra rossa for- muda is, to our knowledge, the only one where thin mation, there are also methodological flaws in the sections of terra rossa samples taken from small evidence adduced to support them in previous re- karst funnels were examined with a polarized-light search. Many researchers, trying to demonstrate microscope. Ruhe found what he called “sand-sized that a terra rossa is a detrital deposit of eolian dust grains of calcium carbonate” surrounded by red or volcanic ash, have compared chemical or iso- clay and, perhaps unaware that those “grains” topic analyses (of Al/Fe, Zr, and Sr isotope ratios, might be simply unreplaced bits of limestone—un- U isotope ratios, or rare earths) of terra rossa, of the replaced small floaters—went on to call them “sand underlying limestone, and of dust or ash and have grains” (no longer just sand-sized grains); on this demonstrated that a number of chemical elements sole basis, he concluded that the terra rossa “must in the terra rossa are indeed allochtonous (see Del- be an accretionary layer in the soil,” that is, a de- gado et al. 2003; Durn 2003; Mee et al. 2004; Muhs trital sediment or pedisediment. But replacement et al. 2007), but this finding does not necessarily may be very difficult to detect (Pettijohn 1957, p. imply that the mineral grains containing those el- 111–112), especially where it is complete and ements are themselves detrital. They could be au- where preservation of internal details is poor (or thigenic instead, but this possibility does not ap- where there were no details to preserve), even if the pear to have been considered by previous authors. petrographer is explicitly looking for it. It implies It seems to us that chemical analysis is by itself no criticism of Ruhe et al. (1961) to say that they blind to whether a rock’s mineral grains are detrital could have missed it. or authigenic, a crucial distinction ascertainable Another problem for previous workers has been petrographically. seeing the terra rossa/limestone boundary as This brings us to a second methodological prob- “sharp” (Comer 1974; Hall 1976; R. V. Ruhe, pers. lem. To our knowledge, and with the single excep- comm., 1978; Olson et al. 1980) and taking that tion of Ruhe et al. (1961), discussed in the next sharpness as evidence that the terra rossa is detrital. paragraph, soil scientists appear to have neglected As our figure 1 demonstrates, however, the bound- the study of terra rossa origin by polarized-light ary is actually a reaction zone several centimeters petrography, the only technique that could—with thick, which we describe next. An unintended con- luck—enable an observer to establish whether min- sequence of seeing the contact with the underlying eral grains in rocks are detrital or authigenic (Wil- limestone as sharp, combined with routinely re- liams et al. 1954; Pettijohn 1957, p. 111–112) and, ferring to terra rossa as a “soil,” has been to deflect if authigenic, whether they are cements, replace- the attention of subsequent students of terra rossa ments, or displacive crystals (Folk 1965). It may be origin away from its bottom, which turns out to be relevant to note in this regard that benchmark text- the best place to demonstrate petrographically that books of soil micromorphology, such as Brewer’s the red clay replaces the calcite. (1964) and FitzPatrick’s (1993), do not stress con- cepts and textures fundamental in understanding New Evidence: Reaction Front the genesis of rocks, such as “authigenic,” “cement,” and “replacement.” The excellent micro- In 2005, we discovered a metasomatic reaction morphological atlas by Delvigne (1998) goes a long front, shown in figure 1, between terra rossa and way toward bridging the gap between the aims of the underlying Salem Limestone, on a wall of the soil micromorphology and the aims of petrography excavation for the foundation of a new science of rocks, although the effort is marred by an opaque building on the campus of Indiana University, terminology. An additional subtle problem may be Bloomington. The front is subvertical, cuts across that terra rossa is often referred to as “terra rossa bedding, and is 1–2 m below the surface. It consists soils,” a term that, in effect, takes for granted that of a bleached zone A, 3 cm wide, and a replacement the genesis of terra rossa clays is pedological by zone B, 6 cm wide. Zone B is the region where the definition. As we show in this article, however, it replacement of limestone by clay was last in pro- is more accurate to view terra rossa as a claystone gress: white, still-unreplaced bits of calcite are vis- Journal of Geology TERRA ROSSA GENESIS 65

Figure 1. Reaction front between terra rossa and Salem Limestone at the construction site for Simon Hall, Indiana University campus, Bloomington. The Salem Limestone is a coarse-grained calcarenite, here massive and horizontal. The front thus cuts across bedding. It consists of a zone of bleached carbonate (zone A) 3 cm wide, containing a thin subzone of opaques, and a zone of clay-for-calcite replacement (zone B) 6 cm wide. The labels “2a,” “2b,” “3a,” etc., correspond to the approximate spots of photomicrographs in figures 2a,2b,3a, etc. ible in zone B (fig. 1b). The bleached zone A con- ure 3a shows another partial replacement of the tains a thin subzone of opaque grains of a Mn oxide. center of a calcite cement plate by orange clay; the The Mississippian Salem Limestone is a massive two unreplaced ends are still in optical continuity calcarenite consisting mostly of coarse crinoid and (fig. 3b), showing that they have not been displaced bryozoan fragments cemented by clear calcite; or rotated and that the replacement is therefore iso- chemical analyses provided by the Indiana Geolog- volumetric. Figure 3c shows an incipient replace- ical Survey indicate that the unaltered Salem con- ment of another crinoid fragment by orange clay; tains small amounts of Al, Mn, Si, and Fe, generally the incipient replacement is also clearly volume for less than 1 wt% of their oxides. volume and exhibits a striking serrated microtex- Petrography, Zone B. The field evidence of a re- ture also shown in figure 4a,4b. We return to that placement-and-bleaching front is confirmed by pet- serrated texture in “Replacement Physics.” rographic analysis of six thin sections from samples The crossed-polars photomicrographs (fig. 3b,3d) from the front. (Another 60 terra rossa samples not show that the orange clay aggregates replacing the from the front have been examined petrographi- calcite cement and fossils have crystalline texture; cally, and about 25 of them have been studied mag- that is, they consist of interlocking crystals and netically.) Figure 2c shows a circular crinoid col- thus indicate authigenesis. The orange crystals are umnal from the replacement zone that is partly Fe3ϩ-bearing kaolinite; see “Mineralogy.” The fact replaced by an orange clay aggregate that preserves that, in the replacement zone, the calcium carbon- the shape and volume of the replaced portion of the ate fossils and cement are only partly replaced and crinoid, a feature characteristic of replacement. Fig- occur in all stages of replacement from incipient 66 E. MERINO AND A. BANERJEE

Figure 2. Petrography of the leaching and replacement zones. The site of each photomicrograph is marked in figure 1. a, Unaltered Salem Limestone consists of bryozoans and equinoderm fragments (black arrows) well cemented by clear calcite cement. Plane-polarized light. b, Dissolution voids (see also those in the unreplaced portion of the crinoid in c) are produced in the bleaching zone by leaching of calcite driven by the Hϩ released by the replacement taking place in zone B, reaction (1). Crossed polars. c, Circular crinoid columnal at center is partly replaced by orange kaolinite (black arrow) in the replacement zone (B) of figure 1. The circular shape of the replaced portion is roughly preserved, indicating isovolumetric replacement, a fact used in adjusting reaction (1). The crinoid was slightly leached when it was in the bleaching zone (A), before it started to be replaced. The leaching can be seen in the unreplaced portion (white arrow) of the calcitic columnal. Plane-polarized light. d, A fairly mature iron oxide pisolite from far behind and above the front, surrounded by terra rossa clay. Both pisolite and clay contain many quartz silt grains, probably left undissolved from dissolution of eolian dust and worked into the terra rossa pedologically; see “Elements Making Up the Red Clay Probably Come from Dissolved Dust.” The pisolite grew slowly (from an incipient precursor like the one in fig. 3a), replacing the surrounding clay and incorporating its quartz silt grains. Plane-polarized light. to nearly complete makes the identification of the happen in a soil, but it seems doubtful that it could authigenic replacement unmistakable. By contrast, reach the bottom of a terra rossa claystone several in most thin sections of terra rossa samples taken meters thick. Also, detrital illuviation cutans dis- far behind the front, the carbonate is completely play sweeping (not sharp) extinction under crossed replaced, so that fossil shapes are no longer une- polars. (Interestingly, Brewer [1964, p. 224] says that quivocally preserved or recognizable. illuviation cutans are “formed by movement of the The authigenic kaolinite crystals may reach 1 cutanic material in solution or suspension [our ital- mm or more in size. Some display faint growth ics] and subsequent deposition.” Note that Brewer rings. We do not interpret the rings as evidence of did not see it as relevant whether a cutan is au- cutans formed by detrital illuviation because they thigenic, detrital, or recrystallized-detrital [as is the are contained within large, authigenic, single crys- matrix of graywackes; Williams et al. 1954, p. 297] tals, as seen by their sharp extinction under crossed and included all possibilities in the definition.) polars. Illuviation, if understood as detrital, can Bleaching Zone, Zone A. The bleaching zone of Journal of Geology TERRA ROSSA GENESIS 67

Figure 3. Clay-for-calcite replacements in zone B of the terra rossa front. Sites of the photomicrographs are marked in figure 1. Zone B is the zone of current replacement. a, Large orange clay aggregate (between arrows) partly replaces the center of a plate of calcite cement in zone B, leaving two still-unreplaced portions (crosses), one on each side. The crinoid at center is also still unreplaced. The clay aggregate contains an incipient, fuzzy iron oxide pisolite (pi) that has crystallized recently, like the clay itself. b, Same as a, under crossed polars, showing that the two portions of unreplaced calcite cement are in optical continuity and that the clay aggregate consists of interlocking crystals, indicating that the red kaolinite is authigenic. c, Brown clay starts (at arrow) to replace a crinoid fragment, producing a striking serrated texture (Pettijohn 1957, p. 674) also evident in figure 4a and throughout the replacement zone B; see “Petrography, Zone B.” Plane-polarized light. d, Same as c, under crossed polars, showing the large size of some of the authigenic clay crystals by the size of areas with one birefringence (arrow). the front, zone A in figure 1, is white, contains a incipient concentric layering that can be seen in thin belt of black opaque particles of a Mn oxide, thin sections of terra rossa far behind the front (see and under a microscope is seen to contain many fig. 2d), may reach 0.5 cm in diameter, and are large dissolution pores, visible in figure 2b as the eroded out of the terra rossa at many spots in areas under extinction. This high-porosity/perme- Bloomington. Most pisolites consist of goethite, ability region, occurring primarily in zone A (and to but a few consist of maghemite, a strongly mag- a lesser extent in zone B) but not in the fresh lime- netic polymorph of hematite (Pruett 1959; J. G. stone ahead of the front or in the terra rossa left Meert, F. Pruett, and E. Merino, unpub. manu- behind it, is crucial for the dynamics of the front. script). Pisolites. The authigenic clay in the replace- Mineralogy. X-ray diffractograms of Blooming- ment zone contains, floating in it, a few small, ton terra rossa show phyllosilicate peaks at ap- fuzzy aggregates (such as the one marked “pi” in proximately 7, 10, and (expandable) 12–13 A˚ ,aswas fig. 3a) of very dark brown Fe oxides that are also reported by Olson et al. (1980). The 7-A˚ peaks must authigenic (Nahon 1991, p. 158 and fig. 4.14). They correspond to kaolinite, identified optically by its are the precursors of more mature pisolites, with deep orange color, weak pleochroism, single cleav- 68 E. MERINO AND A. BANERJEE

age, parallel extinction, and low birefringence (though higher than pure kaolinite’s). By comparison with the kaolinites typical of laterites (Nahon 1986, p. 171; Muller et al. 1995), the orange color and pleochroism undoubtedly reﬂect some substi- tution of Fe3ϩ for Al in the kaolinite.

Replacement Physics To interpret the clay-for-calcite replacement geo- chemically and to model terra rossa formation by metasomatic replacement at a moving reaction front (A. Banerjee and E. Merino, unpub. manuscript), it is essential to understand how a replacement takes place. In this section, we summarize the new theory of replacement that emerges from Maliva and Siever (1988), Merino et al. (1993), Na- hon and Merino (1997), Merino and Dewers (1998), and Fletcher and Merino (2001). The essence of the new theory is that a replacement takes place because the new mineral pressure-dissolves the host by means of the induced stress, or crystallization stress, that it exerts as it grows. The term “replacement,” as long used by petrographers (e.g., Bastin et al. 1931, p. 586 ff; Wil- liams et al. 1954; Pettijohn 1957, p. 111), refers to the occurrence of a guest mineral A just where the host B used to be but preserving both the volume and (often) some internal morphological details of B. Replacement is thus detectable only visually; see ﬁgure 4c. Because morphological details are preserved, A growth and B dissolution must be simultaneous. Because the volume is preserved, A growth and B dissolution must be coupled so as to make their volumetric rates mutually equal, as these could not be equal otherwise. But what is the coupling factor?

Figure 4. Mineral replacement via pressure solution. a, Serrated microtexture produced by the growing clay (up- by dissolution-precipitation, the sphalerite layering per half) as it replaces the carbonate (lower half) in zone would have been destroyed by sphalerite dissolution be- B of the terra rossa front of ﬁgure 1. This microstylolitic fore the dolomite grew and could not have been pre- texture is excellent evidence that the replacement of the served. Also, sphalerite dissolution could not have trig- host calcite took place not by dissolution-precipitation, gered the growth of dolomite because the two minerals as is generally thought, but via pressure solution driven have no elements in common and sphalerite could not by the crystallization pressure exerted by the growing have “known” how to dissolve leaving a void with the clay; see “Replacement Physics.” Crossed polars. b, An- shape of the future dolomite idiomorph. But with re- other microstylolitic texture, produced here by the replacement viewed as resulting from pressure solution of placement of pentlandite by actinolite in an igneous ore the sphalerite by the growing dolomite, the petrographic from South Africa; see text. (Backscattered-electron pho- features of the replacement—preservation of host mor- tomicrograph by C. Li; from Li et al. 2004). c, Replace- phological details and of volume (and equality of volu- ment of dolomite for sphalerite in Pb-Zn ores at Galmoy, metric rates)—are easily accounted for; see “Replace- Ireland, beautifully preserving the sphalerite layering ment Physics.” (Photomicrograph by A` . Canals, (Merino et al. 2006). If the replacement had taken place University of Barcelona.) Journal of Geology TERRA ROSSA GENESIS 69

Perhaps since the influential article by Weyl cally forces the rates of A growth and B dissolution (1959), who—without mentioning replacement and to become mutually equal, thus preserving mineral apparently unaware of its uniqueness as grasped by volume. Merino and Dewers (1998) showed that Bastin et al. (1931), Williams et al. (1954), Pettijohn morphological details of the host can be preserved (1957), and other petrographers—implicitly re- if the growth increments of the guest are smaller garded mineral replacement as dissolution plus pre- than the details themselves (but are erased other- cipitation, mineral replacement has been identified wise, much as the details of a photograph would with its assumed mechanism, “dissolution-precip- be erased if we tried to replace them with too-large itation” (e.g., Walker 1962; Carmichael 1969; tiles or pixels). Fletcher and Merino (2001) calcu- Plummer 1975; Knauth 1979; Milliken 1989; Put- lated the growth-induced stress and the replace- nis 2002; Putnis and Mezger 2004; Rendo´ n-Angeles ment rate from interacting rheology and stress- et al. 2006), a mechanism whereby the dissolution driven kinetics. of the host is assumed to precede and chemically Another hint that replacement does involve pres- trigger the precipitation of the guest. However, dis- sure solution comes now from the clay-for-calcite solution-precipitation fails to account for the replacements in the Bloomington terra rossa. As dolomite-for-sphalerite replacement shown in fig- the clay replaces the calcite, a striking serrated mi- ure 4c. (1) If the sphalerite had dissolved first, how crotexture develops, shown in figure 4a and seen could the later dolomite have preserved its botrioi- throughout the zone of current replacement, zone dal layering at all? (2) If the sphalerite had dissolved B. Pettijohn (1957, p. 216, 674) discussed such ser- first, how could it have “known” to dissolve leav- rated texture but for minerals different from clay ing a hollow with the euhedral shape of the future and calcite. If the serrated texture in the terra rossa dolomite? (3) How could sphalerite dissolution trig- means what it means in stylolitization—namely, ger instant dolomite growth if the two minerals that it is produced by pressure solution—then it have no chemical elements in common? (4) Even implies that the replacement of calcite by clay does if there were elements in common between host happen via pressure solution as well, lending un- and guest (as there are indeed between calcite and expected support to Maliva and Siever’s (1988) con- dolomite in dolomitization), how could that chem- jecture. The replacement in figure 4b of pentlandite ical feedback ensure—except by chance or by ex- (a nickel sulfide) by actinolite (a Ca-Mg amphibole) perimental design—that the guest mineral would in an igneous ore from South Africa (Li et al. 2004) precipitate at the exact place the host had dissolved displays the same microstylolitic texture and is and at the same volumetric rate (so as to preserve also excellent independent evidence that pressure volume)? Other cases where the chemical coupling solution was involved in the replacement. (Note implicitly taken for granted in dissolution-precip- that the absence of serrated texture in a particular itation also does not exist are the common replace- replacement does not mean that pressure solution ment of limestone by chert (Maliva and Siever was not involved in it.) For the following section 1988) and the replacement of pentlandite by actin- and for the second article of this series (A. Banerjee olite shown in figure 4b (Li et al. 2004). and E. Merino, unpub. manuscript), we retain the Thus, Maliva and Siever (1988), after discuss- idea that the replacement of calcite by clay is iso- ing the weaknesses of dissolution-precipitation volumetric and takes place by clay-growth–driven pointed out above and others, all of which invali- pressure solution of the calcite. date it as accounting for a replacement texture in the sense of Bastin et al. (1931), had the brilliant Dynamics of the Reaction Front new idea that a replacement happens when the guest mineral A starts to grow at a point in a rigid The dynamics of the replacement-and-leaching front rock and, via the crystallization stress that it exerts is determined by the reactions happening in it and on host B, pressure-dissolves B as it grows. Maliva their feedbacks, and it has two interrelated aspects. and Siever invoked not a chemical coupling but a The chemical dynamics discussed in this section physical coupling between A growth and B disso- gives rise to, and interacts with, a morphological lution; that is why it works in all cases, regardless dynamics of the front, discussed in “The Reactive- of whether A and B have components in common. Infiltration Instability and the Origin of Karst.” Dewers and Ortoleva (1989) demonstrated Maliva The chemical dynamics is schematically shown and Siever’s conjecture, using the Navier-Stokes in figure 5: aqueous Al, Si, and Fe reach the front equation for momentum conservation. Nahon and from the terra rossa side and combine at the front Merino (1997) showed how replacement by in- to produce clay crystals that replace calcite grains duced-stress–driven pressure solution automati- and fossils by reaction (1) below, releasing acid that 70 E. MERINO AND A. BANERJEE

Local Mass Balance upon Replacement. The constant-volume replacement of calcite by clay (here assumed to be pure kaolinite for simplicity) can be written as

ϩ 3ϩ ϩ 2.7CaCO3(calc)2Al 2SiO 2 ϩ p 5H22254(kaol) O Al Si O (OH) ϩ ϩϩϩ Ϫϩ ϩ 2.7Ca 2.7HCO3 3.3H , (1)

where the 2.7 (ratio of formula volumes of kaolinite andcalcite p 99/37 ) ensures constant volume. As implied in the previous section, the mass balance (eq. [1]) is really the sum of two stress-coupled reactions: clay growth from its aqueous ions and crystallization-stress–driven pressure solution of an equal volume of calcite. Choice of Species in Equation (1). Before we discuss the consequences of the release of acid by reaction (1), a word is in order about the choice of ϪϪ aqueous and mineral species in it. AllCO3 released by calcite is assumed to pair with Hϩ to be- Ϫ come the releasedHCO3 . The aqueous aluminum in reaction (1) is in the form of Al3ϩ because it would probably be the predominant Al species in Figure 5. Dynamics of the replacement-and-leaching the pore fluid (because meteoric and soil waters in reaction front of figure 1. Under advective supply of aque- authigenic clays are bound to be weakly acid) and ous Fe, Al, and Si from the right, clay crystals grow and because, as an already octahedrally coordinated partly replace calcite fossils and cement in the current replacement zone, releasing Hϩ ions according to reac- species, it is the ion that actually enters into the tion (1). This Hϩ leaches more limestone, creating the octahedral sheet of the kaolinite structure (Merino adjacent bleached zone. In the near future, when replace- et al. 1989). ment in the current replacement zone is complete, the If, instead of pure kaolinite, we had chosen Fe3ϩ- locus of limestone replacement will shift to today’s bearing kaolinite or illite in reaction (1), which bleached zone, and the acid generated there will leach would be more realistic than pure kaolinite, the left now still-fresh limestone. The front travels toward the side of equation (1) would contain species such as limestone, leaving behind a trail of terra rossa clay, the 3ϩ 2ϩ ϩ 3ϩ 3ϩ age of which is increasingly older the farther it is behind Fe ,Mg , and K , in addition to Al . Again, Fe the front (or increasingly older the higher it is above the would be the appropriate and perhaps predominant front, if the front is moving downward). aqueous iron species in the acid environment of authigenic terra rossa clays. In all cases, the replacement mass balance would still release Hϩ. immediately bleaches and dissolves voids in an ad- Consequences of Replacement (Eq. [1]). Reaction ϩ ϩ ditional slice of limestone, zone A. When the un- (1) has an important by-product, H . The H should replaced calcite in today’s replacement zone B is leach an adjacent slice of limestone, giving rise to completely replaced in the near future, replacement a bleached zone, which is indeed what happens; see will start in what is today the bleached zone, A, and zone A in figure 1. It should also liberate the trace the acid generated there will start leaching and of Mnϩ2 contained in the limestone (the Salem con- bleaching what is today still fresh limestone just to tains up to 0.06 wt% Mn; Indiana Geological Sur- the left of zone A in figure 1b. This is how the meta- vey, pers. comm.), which would immediately oxi- somatic front moves across the limestone, leaving dize to Mnϩ3 and/or Mnϩ4 and reprecipitate as tiny behind (and/or above) it a trail of terra rossa that is black Mn oxide particles. This is also what hap- increasingly older the farther it is from the front. pens, in fact; see the thin black subzone of zone A Journal of Geology TERRA ROSSA GENESIS 71 in figure 1b. But most important, the released Hϩ interfere with the operation of the reactive-infil- should dissolve out voids in zone A by the reaction tration instability (see “The Reactive Infiltration Instability and the Origin of Karst”), by plugging ϩ ϩϩϩϪp ϩ CaCO3(calc)H Ca HCO 3 , (2) some of the porosity/permeability created by the leaching. and this is also confirmed by the dissolution voids shown in figure 2b. The voids increase the porosity/ Elements Making Up the Red Clay Probably permeability of zones A and (to some extent) B and Come from Dissolved Dust appear to be mostly plugged later by clay growth that replaces the bleached calcite. In the language of dy- Many investigators, cited in “Previous Work on namics, the moving bleached zone becomes a po- Terra Rossa,” have shown that specific terra rossa rosity soliton, or a single porosity wave. The porosity formations contain chemical elements that can be wave traveling across the limestone should trigger traced to elements in eolian dust, confirming the a reactive flow instability that makes the replace- fact that eolian dust must settle on the earth’s sur- ment self-accelerating and the front spontaneously face. But since the red clay making up terra rossa fingered, as discussed in detail in “The Reactive- is authigenic, as shown in figures 1, 3b, and 3d, the Infiltration Instability and the Origin of Karst.” Al, Fe, and Si needed to make the clay crystals must The occurrence of the predicted bleached zone be supplied to zone B as aqueous ions. We therefore adjacent to the replacement zone, with its thin suggest here that the aqueous Si, Fe, and Al sup- black belt and its dissolution pores, is excellent plied to the front come from the dust fraction that evidence that the observation of replacement, the is dissolved at the surface of the existing terra rossa conjecture that it occurs via pressure solution, and and then delivered, by infiltration, to the front a its representation by reaction (1), are correct. This few meters below. Acid rain and soil water would is an important point. Note that the conventional quickly dissolve the finest fraction of the dust, dissolution-precipitation process—namely, the cal- which presumably is clay rich. In support of this cite host dissolves first, followed by growth of guest idea is the following petrographic evidence. kaolinite; see “Replacement Physics”—could not If the finest dust is dissolved, the undissolved possibly preserve mineral volume, a basic property fraction, in general mainly quartz (and feldspar) silt, of replacement and one we observe petrographi- should be left on the terra rossa and would be cally (fig. 2c; fig. 3a,3b), since there is only at best worked into it pedologically. (The quartz silt would a weak coupling (via Hϩ) between the the calcite be insoluble in the soil pore water, enriched in sil- and the kaolinite and their rate constants differ by ica by dissolution of the finest dust fraction.) In- five or more orders of magnitude. In addition, if the deed, we have seen (fig. 2d) scattered quartz silt in calcite host had dissolved first, there would be no many thin sections of Bloomington terra rossa. way for the growth of the guest kaolinite to preserve morphological features of the calcite, such as The Reactive-Infiltration Instability the circular shape of the crinoid in figure 2c, the and the Origin of Karst other basic feature of replacement. (These problems of the dissolution-precipitation mechanism be- Where a reactive fluid both flows through a porous come evident in our quantitative reaction-transport rock and partly dissolves it, a porosity-making front model of terra rossa genesis [A. Banerjeee and E. is established that was quantitatively predicted in Merino, unpub. manuscript].) the 1980s and 1990s to become fingered (Chadam Cementation. As the unleached calcite of zone et al. 1986; Ortoleva et al. 1987; Aharonov et al. A starts to be replaced by kaolinite, not only does 1997); see figure 6A. The dissolution increases po- reaction (1) take place but also the previously rosity/permeability, which accelerates advection of dissolved-out pores may become cemented or half- reactive water. In turn, the faster advection accel- cemented by kaolinite, according to erates further dissolution. This is the reactive- infiltration instability. It works as follows. Any 3ϩ ϩ ϩ p higher-than-average porosity/permeability in a vol- 2Al 2SiO225H O ume element of the front, such as at point a in ϩ ϩ Al225 Si O (OH) 4(kaol) 6H , (3) figure 6A, captures flux from the neighbor elements. Dissolution thus accelerates at a, increasing which also produces acid that would help that pro- its porosity/permeability even more, drawing still duced by reaction (1) to generate the next leaching more flux from the neighbor elements and produc- zone. As noted in “Discussion,” this cement may ing a high-permeability finger. Simultaneously, the 72 E. MERINO AND A. BANERJEE

rosity/permeability fingers. (This competitive morphological dynamics also takes place in the completely different context of quartz growth within agates in basalts; see Wang and Merino 1995. In that case, the fingers of the instability are the quartz fibers themselves, which make agates in- variably fibrous.) After the terra rossa front has become fingered, the competition for reactive flux continues among the fingers themselves, leading to successive jumps in the spacing and size of the fingers (Szymczak and Ladd 2006). The predicted “cascade” of scales is shown schematically in figure 6B. From being a set of advancing fingers, the front passes to being a set of funnels one order of magnitude larger than the fingers. In turn, the funnels later jump to sinks an order of magnitude greater in size and spacing than the funnels. The reaction front we have discovered between terra rossa and the underlying Salem Limestone ap- pears to be a natural case of the kind of moving porosity-making front whose dynamics is described above and in figure 6. The instability-triggering dissolution is the leaching (eq. [2]) carried out in the bleached zone by the acid released by reaction (1) in the replacement zone. The replacement-plus- bleaching front should therefore become fingered as it advances into limestone, and the fingers should jump to funnels and these to sinkholes. But these predicted forms coincide with the morphological features that are characteristic of karst carbonates (e.g., Thornbury 1954; White 1988; Don- ovan 2002; Twidale 2004). We thus arrive at the surprising realization that the terra rossa’s authigenic clay indirectly makes the very karst lime- Figure 6. The reactive infiltration instability; see text. stone morphology that contains it, which explains A, Schematic representation of the spontaneous fingering why the two are associated. of a dissolution front moving through a porous rock, pro- The singularity and separation between karst duced by the reactive infiltration instability. Any slight funnels or sinks are traditionally attributed to the excess porosity/permeability at a point of the front, point focusing of descending acid water by intersecting a, captures reactive water (arrows) from its neighbor fractures, which are conveniently assumed to have regions, b, increasing the dissolution rate at its tip, a , the spacing required to produce the observed sink- and simultaneously starving its neighbor zones, b, which hole spacing (e.g., White 1988; Twidale 2004). Ad- are left behind (b ) as the front advances, creating new fingers at c, which themselves grow faster than b and mittedly, limestone dissolution would be faster at become new fingers. See the predicted fingering in Ahar- the intersection of two subvertical fractures, giving onov et al. (1997; their fig. 3). B, Schematically, jumps rise to a funnel or sink, but intersecting fractures in the scale of finger spacing predicted to take place spon- are an accidental feature, one that cannot be taneously (Szymczak and Ladd 2006). C, Actual fingers counted on systematically to determine the loca- or wormholes in a “floater” from Bloomington. tion of every sink of every karst, and furthermore, the assumed intersecting fractures have to have the neighbor volume elements b, deprived of reactive required spacing to produce the observed spacing flux, get left behind by a and also by volume ele- between karst features. (Most concavities in many ments farther out, c, which now start to become floaters in Bloomington’s front yards do not exhibit new fingers or funnels of higher permeability them- intersecting fractures; see fig. 6C.) On the other selves and later produce still others farther away. hand, the reactive-infiltration instability is a mech- Soon, the front becomes an advancing set of po- anism triggered internally, and it automatically Journal of Geology TERRA ROSSA GENESIS 73 would go through a cascade of scales, which seems 4. We suggest that the major elements needed to best able to account for the spectrum of scales ob- make authigenic clay at the front must come in served in karst limestones. aqueous form from the dissolution of the finest dust fraction at the surface. This suggestion finds tentative confirmation in the widespread occur- Discussion rence of undissolved quartz silt in many thin sec- Perhaps the most striking aspect of the new geo- tions of older terra rossa far behind and above the dynamic theory presented here is the prediction front (see fig. 2d). that terra rossa genesis causes the karst morphol- Implications for Limestone Weathering. The geo- ogy, a prediction confirmed by the worldwide as- dynamic model suggests that limestone weathering sociation of terra rossa with karst limestones. The takes place—see figures 1, 5, and 6—partly by pres- new picture of terra rossa genesis starts with the sure solution in the replacement zone and partly discovery of a 10-cm metasomatic front at the base by chemical dissolution in the leaching zone by of the Bloomington terra rossa. The petrographic reaction (2), driven by the acid released by the re- evidence of replacement of limestone by authigenic placement (eq. [1]). Note that this acid comes from red clay is conclusive. The field evidence, figure 1, clay formation, not from meteoric carbonic acid. It of a bizonal replacement-plus-leaching front also is is because the two dissolutions—pressure solution conclusive. The solutes needed to make the clay and chemical dissolution—are mutually acceler- must come from dissolution, by rain and soil water, ating that carbonate weathering probably proceeds of the finest fraction of the eolian dust supplied to (at the wormholes, funnels, and sinks) at much fas- the top of the terra rossa. The top of the terra rossa ter rates than hitherto thought, if the dust supply is commonly altered pedologically. The terra rossa permits. (In contrast, according to the conventional can be easily eroded, especially in advanced karst. view in geomorphology, limestone weathers by A particular terra rossa formation, if still uneroded, chemical dissolution alone, with the dissolution may now be viewed as the authigenic claystone driven at roughly constant kinetics by meteoric car- recording the finest dust fraction fallen at the site bonic acid.) in question. Future Work. We hope that other investigators The new picture generates predictions that are will undertake petrographic analysis of the narrow confirmed by observations: reaction zone between other terra rossas and their 1. The prediction that the replacement reaction underlying carbonates to confirm and refine our (eq. [1]), via the acid it releases, should leach ad- petrography and to establish (by petrophysical ditional limestone is confirmed by the occurrence study) the porosity and permeability ranges of red of the leaching zone A just ahead of the replace- claystones. Also, the new model must be checked ment zone B (fig. 1). (1) by quantitative reaction-transport modeling of 2. Maliva and Siever’s (1988) conjecture of the the formation of terra rossa clays (A. Banerjee and mechanism of replacement, which calls for pres- E. Merino, unpub. manuscript) and comparison of sure solution of host driven by the induced stress its predictions to observations; (2) by isotopic and generated by the guest, is confirmed by our dis- chemical analyses aiming to establish that (or covery of microstylolitic, serrated replacement whether) the major elements do reach the front as contacts between clay and calcite (figs. 3, 4), to our aqueous species from dust dissolution; and espe- knowledge not reported before. cially (3) by quantitative modeling of the reactive- 3. The new porosity created in the front’s leaching zone qualitatively should trigger the reactive- infiltration instability in the specific case of the infiltration instability (but see “Future Work” bizonal front of figure 1. Aharonov et al.’s 1997 about the need to check this quantitatively). This impressive three-dimensional modeling of the re- instability theoretically must deform the moving active-infiltration instability referred to a moving ↔ front into a set of wormholes (and then funnels and front at which only adissolution advection feed- then sinks). This predicted morphology is con- back takes place. But for the terra rossa–making firmed by the occurrence of karst limestones as- front of figure 1, the feedback that must be studied sociated with terra rossa everywhere. (There are by instability analysis is that between advection many karst limestones, especially mature karst and several coupled reactions, namely, growth of towers, without terra rossa on them, but it must clay with simultaneous pressure solution of calcite be remembered that the red clays can be eroded by reaction (1), cementation of previous voids by and/or washed off easily from the underlying reaction (3), and simultaneous leaching of voids by limestone.) reaction (2). 74 E. MERINO AND A. BANERJEE

ACKNOWLEDGMENTS Li, J. Schieber, and J. R. Dodd of Indiana University Ideas and petrography were discussed with R. L. for comments on a draft, the photomicrograph in Hay, professor emeritus of the Universities of Illi- ﬁgure 4b, the use of a Zeiss polarizing microscope, nois and California at Berkeley, in the fall of 2005. and advice on Bloomington’s Mississippian carbon- He died on February 10, 2006, in Tucson. This ar- ates, respectively. We appreciate critical reviews by ticle is dedicated to his memory. Thanks to our three reviewers for the journal. We thank Indiana friends A` . Canals of the University of Barcelona for University for a grant-in-aid to cover thin section- the photomicrograph in ﬁgure 4c and A. Basu, C. ing and travel.

REFERENCES CITED

Aharonov, E.; Spiegelman, M.; and Kelemen, P. 1997. Fletcher, R., and Merino, E. 2001. Mineral growth in Three-dimensional flow and reaction in porous media. rocks: kinetic-rheological models of replacement, vein J. Geophys. Res. 102:14,821–14,834. formation, and syntectonic crystallization. Geochim. Bardossy, G. 1982. Karst bauxites. Amsterdam, Elsevier, Cosmochim. Acta 65:3733–3748. 441 p. Folk, R. L. 1965. Some aspects of recrystallization in an- Bastin, E. S.; Graton, L. C.; Lindgren, W.; Newhouse, W. cient limestones. In Pray, L. C., and Murray, R. D., H.; Schwartz, G. M.; and Short, M. N. 1931. Criteria eds. Dolomitization and limestone diagenesis. Soc. of age relations of minerals, with especial reference Econ. Paleontol. Mineral. Spec. Publ. 13:14–48. to polished sections of ores. Econ. Geol. 26:561–610. Foster, J., and Chittleborough, D. 2003. Soil development Brewer, R. 1964. Fabric and mineral analysis of soils. on dolomites of the Cambrian Normanville Group at New York, Wiley, 470 p. Delamere, South Australia. In Roach, I. C., ed. Ad- Brimhall, G. H.; Lewis, C. J.; Ague, J. J.; Dietrich, W. E.; vances in regolith. Bentley, Cooperative Research Hampel, J.; Teague, T.; and Rix, P. 1988. Metal en- Centre for Landscape Environments and Mineral Ex- richment in bauxites by deposition of chemically ma- ploration (CRC LEME), p. 131–132. ture aeolian dust. Nature 333:819–824. Frumkin, A., and Stein, M. 2004. The Sahara-east Med- Carmichael, D. M. 1969. On the mechanism of prograde iterranean dust and climate connection revealed by metamorphic reactions in quartz-bearing pelitic strontium and uranium isotopes in a Jerusalem spe- rocks. Contrib. Mineral. Petrol. 20:244–267. leothem. Earth Planet. Sci. Lett. 217:451–464. Chadam, J.; Hoff, D.; Merino, E.; Ortoleva, P.; and Sen, Guendon, J. L., and Parron, C. 1985. Les phenome`nes A. 1986. Reactive infiltration instability. IMA J. Appl. karstiques dans les processus de la bauxitization sur Math. 36:207–221. substrat carbonate´: examples de gisements du sud-est Comer, J. B. 1974. Genesis of Jamaican bauxite. Econ. de la France. Ann. Soc. Geól. Belg. 108:85–92. Geol. 69:1251–1264. Hall, R. D. 1976. Stratigraphy and origin of surficial de- ———. 1976. Genesis of Jamaican bauxite; a reply. Econ. posits in sinkholes in south-central Indiana. Geology Geol. 71:822–823. 4:507–509. de Lapparent, J. 1930. Les bauxites de la France me´ri- Herwitz, S. R.; Muhs, D. R.; Prospero, J. M.; Mahan, S.; dionale. Paris, Imprimerie Nationale, 187 p. and Vaughn, B. 1996. Origin of Bermuda’s clay-rich Delgado, R.; Martıń-Garcıá, J. M.; Oyonarte, C.; and Del- Quaternary paleosols and their paleoclimatic signifi- gado, G. 2003. Genesis of the terra rossa of Sierra de cance. J. Geophys. Res. 101:23,389–23,400. Ga´dor, Almerıá, Spain. Eur. J. Soil Sci. 54:1–16. Knauth, L. P. 1979. A model for the origin of chert in Delvigne, J. E. 1998. Atlas of micromorphology of min- limestone. Geology 7:274–277. eral alteration and weathering. Que´bec, Min. Assoc. Kurtz, A. C.; Derry, L. A.; and Chadwick, O. A. 2001. Can.; and Paris, ORSTOM, 494 p. Accretion of Asian dust to Hawaiian soils: isotopic, Dewers, T., and Ortoleva, P. 1989. Mechano-chemical elemental, and mineral mass balances. Geochim. Cos- coupling in stressed rocks. Geochim. Cosmochim. mochim. Acta 65:1971–1983. Acta 53:1243–1258. Li, C.; Ripley, E. M.; Merino, E.; and Maier, W. D. 2004. Donovan, S. K. 2002. A karst of thousands: Jamaica’s Replacement of base metal sulfides by actinolite, ep- limestone scenery. Geol. Today 18:143–151. idote, calcite, and magnetite in the UG2 and Meren- Durn, G. 2003. Terra rossa in the Mediterranean region: sky Reef of the Bushveld Complex, South Africa. Econ. parent materials, composition and origin. Geol. Croat. Geol. 99:173–184. 56:83–100. Maliva, R. G., and Siever, R. 1988. Diagenetic replace- Durn, G.; Ottner, F.; and Slovenec, D. 1999. Mineralog- ment controlled by force of crystallization. Geology ical and geochemical indicators of the polygenetic na- 16:688–691. ture of terra rossa in Istria, Croatia. Geoderma 91:125– Mee, A. C.; Bestland, E. A.; and Spooner, N. A. 2004. Age 150. and origin of terra rossa soils in the Coonawarra area FitzPatrick, E. A. 1993. Soil microscopy and micromor- of South Australia. Geomorphology 58:1–25. phology. Chichester, Wiley, 304 p. Merino, E.; Canals, A` .; and Fletcher, R. 2006. Genesis of Journal of Geology TERRA ROSSA GENESIS 75

self-organized zebra textures in burial dolomites: dis- Prospero, J. M.; Ginoux, P.; Torres, O.; Nicholson, S. E.; placive veins, induced stress, and dolomitization. and Gill, T. E. 2002. Environmental characterization Geol. Acta 4:383–393. of global sources of atmospheric soil dust identiﬁed Merino, E., and Dewers, T. 1998. Implications of replace- with the nimbus 7 total ozone mapping spectrometer ment for reaction-transport modeling. J. Hydrol. 209: (TOMS) absorbing aerosol product. Rev. Geophys. 40: 137–146. 1002, doi:10.1029/2000RG00009. Merino, E.; Harvey, C.; and Murray, H. 1989. Aqueous Prospero, J. M.; Olmez, I.; and Ames, M. 2001. Al and Fe chemical control of the tetrahedral aluminum content in PM2.5 and PM 10 suspended particles in south- of quartz, halloysite and other low-temperature alu- central Florida: impact of long-range transport of Af- minosilicates. Clays Clay Miner. 37:135–142. rican mineral dust. Water Air Soil Pollut. 125:291– Merino, E.; Nahon, D.; and Wang, Y. 1993. Kinetics and 317. mass transfer of replacement: application to replace- Pruett, F. D. 1959. A study of the magnetic properties of ment of parent minerals and kaolinite by Al, Fe and some residual soils. MA thesis, Indiana University, Mn oxides during weathering. Am. J. Sci. 293:135– Bloomington, 58 p. 155. Putnis, A. 2002. Mineral replacement reactions: from Milliken, K. L. 1989. Petrography and composition of macroscopic observations to microscopic mecha- authigenic feldspars, Oligocene Frio Formation, south nisms. Miner. Mag. 66:689–708. Texas. J. Sediment. Petrol. 59:361–374. Putnis, C. V., and Mezger, K. 2004. A mechanism of min- Moresi, M., and Mongelli, G. 1988. The relation between eral replacement: isotope tracing in the model system

the terra rossa and the carbonate-free residue of the KCl-KBr-H2O. Geochim. Cosmochim. Acta 68:2839– underlying limestones and dolostones of Apulia, Italy. 2848. Clay Miner. 23:439–446. Rendo´ n-Angeles, J. C.; Pech-Canul, M. I.; Lo´ pez-Cuevas, Muhs, D. R.; Budahn, J.; Prospero, J. M.; and Carey, S. J.; Matamoros-Veloza, Z.; and Yanagisawa, K. 2006. N. 2007. Geochemical evidence for African dust in- Differences on the conversion of celestite in solutions puts to soils of western Atlantic islands: Barbados, the bearing monovalent ions under hydrothermal condi- Bahamas and Florida. J. Geophys. Res. 112:F02009, tions. J. Solid State Chem. 179:3645–3652. doi:10.1029/2005JF000445. Ruhe, R. V. 1975. Geomorphology. Boston, Houghton- Muller, J. P.; Mainceau, A.; Calas, G.; Allard, T.; Ildefonse, Mifflin, 246 p. P.; and Hazemann, J. L. 1995. Crystal chemistry of Ruhe, R. V.; Cady, J. G.; and Gomez, R. S. 1961. Paleosols kaolinite and Fe-Mn oxides: relation with formation of Bermuda. Geol. Soc. Am. Bull. 72:1121–1142. conditions of low temperature systems. Am. J. Sci. Szymczak, P., and Ladd, A. J. C. 2006. A network model 295:1115–1155. of channel competition in fracture dissolution. Geo- Nahon, D. 1986. Evolution of iron crusts in tropical land- phys. Res. Lett. 33:L05401. scapes. In Colman, S. H., and Dethier, D. P., eds. Rates Thornbury, W. D. 1954. Principles of geomorphology. of chemical weathering of rocks and minerals. San New York, Wiley, 618 p. Diego, CA, Academic Press, p. 169–191. Twidale, C. R. 2004. Karst towers: some questions and ———. 1991. Introduction to the petrology of soils and answers. Stud. Geol. Salmanticensia 40:55–83. chemical weathering. New York, Wiley, 313 p. Walker, T. R. 1962. Reversible nature of chert-carbonate Nahon, D., and Merino, E. 1997. Pseudomorphic replace- replacement in sedimentary rocks. Geol. Soc. Am. ment in tropical weathering: evidence, geochemical Bull. 73:237–242. consequences, and kinetic-rheological origin. Am. J. Wang, Y., and Merino, E. 1995. Origin of fibrosity and Sci. 297:393–417. banding in agates from flood basalts. Am. J. Sci. 295: Olson, C. G.; Ruhe, R. V.; and Mausbach, M. J. 1980. The 49–77. terra rossa limestone contact phenomena in karst, Weyl, P. K. 1959. Pressure solution and the force of crys- southern Indiana. Soil Sci. Soc. Am. J. 44:1075–1079. tallization: a phenomenological theory. J. Geophys. Ortoleva, P.; Chadam, J.; Merino, E.; and Sen, A. 1987. Res. 64:2001–2025. Geochemical self-organization. II. The reactive-infil- White, W. B. 1988. Geomorphology and hydrology of tration instability in water-rock systems. Am. J. Sci. karst terrains. New York, Oxford University Press. 287:1008–1040. Williams, H.; Turner, F. J.; and Gilbert, C. M. 1954. Pe- Pettijohn, F. J. 1957. Sedimentary rocks (2nd ed.). New trography. San Francisco, W. H. Freeman, 406 p. York, Harper & Row, 718 p. Yaalon, D. H. 1997. Soils in the Mediterranean region: Plummer, L. N. 1975. Mixing of sea water with calcium what makes them different? Catena 28:157–169. carbonate ground waters. Geol. Soc. Am. Mem. 42: Yaalon, D. H., and Ganor, E. 1973. The influence of dust 219–236. on soils during the Quaternary. Soil Sci. 116:146–155.