Fingerprinting the metal endowment of early continental crust to test for secular changes in global mineralization
Published:January 01, 2006
- PDF LinkChapter PDF
Christien Thiart, Maarten J. de Wit, 2006. "Fingerprinting the metal endowment of early continental crust to test for secular changes in global mineralization", Evolution of Early Earth's Atmosphere, Hydrosphere, and Biosphere - Constraints from Ore Deposits, Stephen E. Kesler, Hiroshi Ohmoto
Download citation file:
Archean cratons are fragments of old continents that are believed to be more richly endowed with mineral deposits than younger terrains. The mineral deposits of different cratons are also diversely enriched with useful (to humankind) chemical elements. Cratons are therefore mineral diversity hotspots that represent regional geochemical heterogeneities of early Earth, the evidence for which remains encoded on each craton as unique metallogenic “fingerprints.” Using six selected elements groups from our extensive in-house GIS database of Gondwana mineral deposits, we derive the metallogenic fingerprints of 11 Archean cratons of the Southern Hemisphere, and compare these against metallogenic fingerprints of the same selected elements in younger crust of three of their host continents (Africa, Australia, and South America). After adjusting the mineral inventory of each craton to account for underexploration of regions lacking infrastructure and other political and economic conditions for mineral investment, we show that mineral deposit density and diversity of Earth's continental lithosphere has decreased with time. We conclude that metallogenic elements were transferred more efficiently from the mantle to the continental lithosphere in the Archean and/or that subsequent (younger than 2.5 Ga) recycling of these elements (mineral deposits) back into the mantle has become more effective. How most of these fragments of old continents inherited their rich and diverse metallogenic characteristics is unresolved, because different cratons are likely to represent only small remnants of once much larger and possibly more varied Archean continents, and part of the total metal inventory of Archean continents must have been recycled back into in the mantle. The latter has implications for understanding the secular change in the redox state of the Archean mantle and fluid envelope.
The lithosphere of Archean cratons (older than 2.5 Ga) is distinct from that of younger continental lithosphere in that they are underlain by relatively thin crust (∼30–40 km) and thick mantle lithosphere (up to ∼250–300 km; de Wit, 1998; James et al., 2001; Stankiewicz et al., 2002; Fouch et al., 2004). The origin of these cratons is still a matter of intense debate. Part of that debate centers on the origin of the cratons' mineral deposits and, in turn, how knowledge of these deposits may shed light on the formation of Earth's early continents. However, surprisingly few quantitative data are available in a collective format with which to embark on a comparative study of the riches found in continental crust of different ages. Near surface, the crust of most cratons is well endowed in concentrations of metallic elements useful and economic to humankind (hereafter referred to as mineral deposits), but it is not known with any degree of certainty if the crust of the old continents was mineralized to a greater degree than that of younger continents, as is often assumed intuitively. As a test of such a temporal change, here we compare the mineral endowment of old cratonic crust with that of younger crust, using our in-house mineral deposit database.
Cratons are metallogenically distinct in that their mineral deposits contain different mixtures of metallic elements from craton to craton (Wilsher et al., 1993; Wilsher, 1995; de Wit et al., 1999). For example, the Kaapvaal Craton of South Africa is known to be relatively enriched in gold and platinum group elements (PGE), the Zimbabwe Craton and the Yilgarn Craton of Australia in gold and tungsten, the São Francisco Craton in gold and base metals (Cu/Pb/Zn), and the Amazonian Craton in gold and tin (Groves et al., 1987; Wilsher, 1995; Barley et al., 1998; de Wit et al., 1999; de Wit and Thiart, 2005). We assume that mineral deposits of Archean cratons are contemporaneous with their host rocks and reflect processes of geochemical concentration that operated during the formation of the cratons. In most cases this can be verified if the age of the mineral deposits is constrained by geological relations or isotopic ages. In some cases it is not clear whether these mineral enrichments were inherited from their host Archean craton or were added to the craton at a later stage. For example, the platinum deposits of the Kaapvaal Craton are associated mostly with the igneous rocks of the Bushveld Complex that intruded the center of this craton in the Paleoproterozoic (2.05–2.06 Ga; Cawthorn and Walraven, 1998; Eglington and Armstrong, 2004) and could have been added from the asthenospheric mantle at that time. However, the PGE geochemistry (and that of other siderophile elements like Ni, Cr, Au) of lithospheric mantle xenoliths found in kimberlites across the Kaapvaal Craton (McDonald et al., 1995), as well as that of Archean mid to lower crust (Hart et al., 2004) and of Archean mineral deposits in the greenstone belts of this craton (Tredoux et al., 1989), suggests that both the lithospheric mantle and crust of this craton were already rich in PGE in Archean times. Indeed it has been suggested that the PGE in the parental magmas of the Bushveld Complex may have been inherited from contamination with its underlying PGE-enriched Archean mantle lithosphere (McDonald et al., 1995). Similarly, most of the extraordinary enrichment of the Kaapvaal Craton in gold is hosted by Archean sedimentary rocks of the Witwatersrand Basin and is thought to have been derived largely by erosion of earlier crust, although some gold may have been introduced and/or remobilized at a later time (Frimmel and Minter, 2002; Phillips and Evans, 2004). Gold deposits are also common in older Archean greenstone belts (Herrington et al., 1997). Although not all of these greenstone belt deposits are firmly dated, we assume they reflect inheritance of Archean gold. Thus, unless there is clear evidence to the contrary, some geodynamic process(es) served to concentrate precious elements together into the continental lithosphere of the Kaapvaal Craton during its formation in the Archean (Groves et al., 1987; Tredoux et al., 1989; McDonald et al., 1995).
The mantle lithosphere of most cratons is also invariably enriched in diamonds that range in age from Mesoarchean to Phanerozoic (Hart et al., 1997; Shirey et al., 2002; Jelsma et al., 2004). Because the preservation potential of these minerals in the mantle lithosphere relates mostly to the relatively low heat flow recorded across most cratons, we do not here focus further on diamonds. They do, however, serve to illustrate the resilience of cratons as Archean geologic archives.
We first set out to verify that different Archean cratons have distinct metallogenic patterns, which we refer to as their metallogenic “fingerprints.” These fingerprints may reveal something fundamental about the formation of cratons and Earth's earliest continents. Next we address the question of whether a unit of Archean continental crust is more enriched in mineral deposits than that of younger crust. We compare our results from Archean cratons with those from younger crust (younger than 2.5 Ga) at different scales: first at a continental scale (e.g., South America, Australia, Africa), and then at a supercontinental scale using all the known mineral deposits of Gondwana. This reveals important information about the evolution of continental crust.
We are limited in our analyses to the Southern Hemisphere, because our mineral database is confined only to continental fragments of the former supercontinent Gondwana (see below). A shortcoming of our previous work (de Wit and Thiart, 2005) relates to the extent that the distribution of known mineral deposits (as in our database) is skewed because some regions are explored better than others. Less-developed countries with poor infrastructure and political instability are less likely to have had their mineral inventory tested to the same degree as better-developed nations. Here, we attempt to adjust for underexploration by including a socioeconomic measure in our calculations.
Establishing metallogenic fingerprints of different cratons by using selected elements from their mineral deposits allows us to address several controversial tectonic models that contrast early Earth processes with those of the present. For example, modern plate tectonic processes yield specific mineral deposit types in distinct plate tectonic environments (Sawkins, 1990; Windley, 1995). Similar associations, if found in Archean cratons, would argue for plate tectonic processes in the Archean. Conversely, on an early Earth dominated by vertical tectonics (e.g., driven by plume and diapir dynamics, as postulated by some workers [Hamilton, 1998; Zegers and van Keken, 2001; van Kranendonk et al., 2004]), metallogenic provinces would be expected to display distinct metal associations not found in the younger crust of the present continents (e.g., Hutchinson, 1981, 1992). In addition, variations in the total concentration of these elements in continental crust of different ages may be used to test for changes in the rates of related tectonic processes over time. This is one aim of this paper, but there are others: First, can metallogenic elements serve as chemical tracers to establish that the earliest continents were assembled by tectonic processes as diverse as those at present? Second, can the mineral diversity patterns of cratons help us to decipher the recycling history of Earth's continental materials? Third, can mineral diversity patterns (or mineral hotspots) of cratons and younger lithosphere be used to test reconstructions of past supercontinents like Gondwana and Rodinia? Finally, because mineralization in the crust requires concentration processes that often involve large fluid fluxes between the mantle and/or crust and the hydrosphere, it is of interest to ask whether secular change in the total mass of mineral deposits in the crust represents a potential proxy for concomitant changes in chemical fluxes and redox states of Earth's major reservoirs such as its mantle and fluid envelope.
DERIVING METALLOGENIC FINGERPRINTS OF ARCHEAN CRATONS AND YOUNGER CRUST USING SELECTED ELEMENTS FROM THEIR MINERAL DEPOSITS
The geological and mineral deposit data used for this study are incorporated in a GIS relational database, called GO-GEOID (GOndwana – GEOscientific Indexing Database), housed at our center, AEON. This database is restricted to the major continental fragments of Gondwana; its geological component is based on the geological map of Gondwana (de Wit et al., 1988), whereas the mineral layer of the database was constructed from open-access literature sources (Wilsher, 1995). Originally the mineral layer comprised roughly 10,000 deposits across Gondwana, covering 15 metallogenic elements (commodities). The mineral layer consists of shape files tied to 28 attribute tables, with attributes ranging from commodities, size, type, and age of the mineral deposits, as well as information on the host rocks of the deposits and the references from which the data were obtained. In 1998, in collaboration with the BRGM (France), we embarked on an “added-value” project to create a metallogenic-potential GIS of Gondwana. During this process the original database was updated and revalidated. Roughly 6,000 new deposits were added to the database. In the final stage of the project, this “updated” mineral database was integrated with geostatistical software and a browser (GEORAMA). The end product yielded a metallogenic-potential GIS of Gondwana (available on CD-ROM; for details see: http://gondwana.brgm.fr/index_eng.htm). Thus, currently there are roughly 16,000 deposits in the mineral layer, ranging from active mines to undeveloped occurrences. Figure 1 shows the typical density of these mineral deposits across Gondwana, and from which our data are extracted. The origin and evolution of the database has been described in detail elsewhere (Wilsher et al., 1993; de Wit et al., 1999; Thiart and de Wit, 2000; de Wit et al., 2004).
Figure 2 shows the global distribution of cratons that we selected for this study. There are 11 Gondwana cratons (seven in Africa, two in South America, and two in Australia) for which we have sufficient mineral deposit data in our Gondwana database. For general comparison, we also selected one Canadian craton, the Superior Province. The mineral deposit data for the latter are from the Geological Survey of Canada (Kirkham et al., 1994, 2002; Jenkins et al., 1997; Eckstrand and Good, 2000; Kirkham and Dunne, 2000, 2002; Eckstrand et al., 2002). We emphasize that this Canadian-based data structure is significantly different from our Gondwana database, and because the subset of their data provided to us may not include all known mineral deposits in the Superior Province, the results in the subsequent sections should be evaluated in this light.
Eleven metallogenic elements were selected for our analyses, and these were divided into six element groups according to their geochemical affinities (e.g., lithophile, chalcophile, siderophile), as well as their relative abundances in the database. The six element groups, and the total number of deposits in which these groups occur on each craton, are tabulated in Table 1A. In total there are just over 6000 deposits spread over 12 identified cratons (Table 1A).
Previously we defined metallogenic fingerprints of fragments of continental crust through their spatial association with a combination of six element groups, and applied this at three scales—cratons, continents, and supercontinents (de Wit and Thiart, 2005). The measure of spatial association (normalized to area) we termed the spatial coefficient, rij , where
The spatial coefficient (equation 1) represents the proportion of deposits (say gold, j) of all the j th deposits (in cratons) that occur in the specified craton (i) per unit area of all 12 cratons. Our spatial coefficient is similar to the measure of spatial association used by Mihalasky and Bonham-Carter (2001) to measure the spatial association between lithodiversity and minerals deposits in Nevada. The value of the spatial coefficient ranges from 0 to infinity; it is equal to 1 if there is no spatial association between a craton and an element group (e.g., if the proportion of j th mineral is the same as the proportion of area occupied by the i th craton). For values of rij > 1 (e.g., where the expected number of deposits is greater than by chance), there is a positive association between mineral j and craton i; rij < 1 (e.g., where there are fewer deposits than expected by chance) indicates a negative association. Because all negative associations are compressed in the range from 0 to 1, and all positive associations fall in the range of 1 to infinity, we use the natural log of rij to eliminate this skewness. Thus, ln(rij ) is a symmetric value around 0: positive associations are greater than 0, and negative associations are less than 0.
The above approach ignores the fact that each study region (e.g., defined cratons) has a different exploration history, a bias that might influence the analyses. Intuitively, we expect, for example, that greater accessibility (e.g., infrastructure) and political stability increases exploration and discovery rates and mining activity. Therefore mineralization in some cratonic areas of less-developed nations may not be adequately represented in our database and analyses. We address this problem by weighting the spatial coefficient (equation 1), with an “exploration index,” using socioeconomic data from the World Bank development indicators database (World Bank, 2004). This “weighted” spatial coefficient (defined below in equation 4) can be used with increased confidence to evaluate the metal endowments of different continental fragments.
Rich countries are better explored than poor countries because of their better-developed infrastructure and investment regimes. We tested various world development indicators published by the World Bank (2004), designed to quantify this differential development. Here we use the measure of gross national income per capita (GNI-C, for 2002; Table 2) because it contains the most obtainable data for all the countries of our interest. To derive the GNI-C for a specific craton, we calculate the percentage of area that each country contributes to the total area of that craton. The craton GNI-C is then calculated as a weighted average (by area percentage) of all the GNI-C values of each of the countries involved. For example the Kaapvaal Craton constitutes 14% of Botswana, 4% of Lesotho, 80% of South Africa, and 2% of Swaziland. The GNI-C for the entire Kaapvaal Craton then is as follows: 14%*GNI-C for Botswana + 4%*GNI-C for Lesotho + 80%*GNI-C for South Africa + 0.2%*GNI-C for Swaziland.
To calculate an exploration index, we use the United States (USA) as a benchmark (e.g., the United States has an exploration index of 1). The exploration index for Craton i is the ratio of the GNI-C for USA to the GNI-C for Craton i(equation 2), with k = 1) of the GNI-C ratios. These exploration indices are given in Figure 3A. For this investigation, we chose to work with the power function k = 1/3, as it closely follows the natural log function and has the advantage that it ranges from 1 (United States) to roughly 5 (least explored cratons).
For our analyses at a continental scale we derive the GNI-C ratio for each continent in a similar manner, and use the combined GNI-C of the United States and Canada (North America, top of the last column of Table 2) as our benchmark. The GNI-C for Africa is from the African Development Indicators (2004). The GNI-C for North America and Australia are from the World Bank development indicators database (World Bank, 2004, ). The GNI-C for South America is from the country data of the World Bank development indicators database (World Bank, 2004, ). The derived exploration indices for the continents are given in Figure 3B.
From the original mineral deposit data (Table 1A) we can now calculate a proportion (pij) as the number of deposits in Cratoni and Element Groupj divided by the total number of deposits in the database (6097; Table 1A). This calculated proportion is then multiplied by the exploration index, wi, where wi is the ratio defined in equation 2 (with k = 1/3) of the ith craton, and divided by the sum of the product of wi, and pij. Thus(equation 3) is then used in the calculation of the “weighted” spatial coefficient rijw:
Metallogenic Fingerprints of Selected Archean Cratons
The number of mineral deposits of the six selected element groups within each of the 12 cratons is given in Table 1A, and the natural log of the spatial coefficients between the element group and each craton (ln(rij ), rij defined in equation 1) is given in Table 1B. The natural log of the “weighted” spatial coefficient (ln (rij w ), equation 4) for the 12 selected cratons and the six selected element groups is given in Table 1C. Figures 4 and 5 summarize these data in graphical format to illustrate the mineral diversity of each craton. For comparison both the unweighted (ln(rij ), solid bars) and the weighted spatial coefficient (ln(rij w ), striped bars) are given in the same figure. Figure 4 expresses the diversity among cratons for each set of elements of the six element groups, and provides 12 cratonic (spatial) coefficients for each set of elements. Figure 5 represents the total spatial coefficient combining all sets of elements (e.g., all six element groups) for each craton; these are the “metallogenic fingerprints” of the 12 cratons. Each individual craton has a unique fingerprint even though some may appear very similar (e.g., Congo and Tanzania Cratons). Where values are high (e.g., Zimbabwe) the imprint of total mineralization is high relative to the other cratons: metaphorically a “strong fingerprint.” Values between −1 and 1 represent a random association between the total element set and the specific craton; values below 0 (Fig. 4) indicate a negative association for the given element group and the specific craton. The values for the Tanzania and Congo Cratons represent the lowest values, and these values may be unreliable because of the low number of discovered mineral deposits on these cratons (and thus in our database). In turn this corroborates the need to incorporate into the analyses an exploration-index value that reflects the past history of (under)exploration of areas under consideration. Of additional interest here is that the relative differences between the weighted and unweighted results express a degree of underexploration, and therefore provide a signal for potential near-surface mineral deposits still to be found in areas where infrastructure development and exploration investment has lagged behind that of North America. The greatest positive values are those of the Zimbabwe Craton. This reflects a high count of mineral deposits covering all six element groups in Zimbabwe and the relatively small size of this craton. It is the only craton that has no negative spatial coefficient in any of the element groups. The similarity between its weighted and unweighted spatial coefficients indicates a mature degree of exploration throughout this craton, in concert with a long history of intense mineral exploration.
Some negative spatial coefficients might be intuitively surprising (e.g., those for gold in the Superior Province). The spatial coefficient (equation 1) is defined as a ratio between two proportions, that is, the number of deposits (say Au) in a specified craton relative to the number of deposits in all the cratons, divided by the proportion of area that the specific craton occupies relative to that of the area of all the cratons. If the ratio is greater than 1, there is a positive association (more deposits expected than by chance), whereas a value between 0 and 1 results in a negative association (fewer deposits expected than by chance). The natural log of the spatial coefficient for the exceptionally large Superior Province is probably negative because it expresses its mineralization relative to other cratons: in the Superior Province there are only 162 Au deposits in the Canadian database compared to a total of 2762 in all cratons combined; and the Superior province is the largest of all the cratons (2.73 × 106 km2). We suspect that there are probably a lot more small deposits in the Superior Province that are not incorporated into the Canadian database, because they are in the GO-GEOID database. The results would also change if one could factor in the actual area of outcrop (most of the Superior Province is covered by glacial drift (till, transported overburden). Geophysical and geochemical exploration methods cannot “see” gold deposits through this cover as easily as they can “see” base metal (VMS) deposits. Other cratons with residual overburden would not be affected by this complication. It is probably unwise therefore to compare the two data sets directly, because they were constructed differently and with different specifications. This supports a call to standardize global databases. For the further analysis below we therefore eliminate the Superior Province from our input.
Mineral Inventory on a Continental Scale
The analyses described above were repeated for three large continental fragments (Africa, South America, and Australia) of the former supercontinent Gondwana, to enable direct comparison of the metallogenic inventory of cratons (older than 2.5 Ga; old crust) with mineralization in younger crust (younger than 2.5 Ga) on each individual continent. The mineral deposit data for the old cratons and younger crust are summarized in Table 3A. The natural log of the spatial coefficients is given in Table 3B, and the natural log of the “weighted” spatial coefficients is given in Table 3C. Figure 6 graphically displays the unweighted natural log of the spatial coefficients (ln(rij ), solid bars) against that of the natural log of the weighted coefficient (ln(rij w ), striped bars). The total mineral inventory for the continental crust older and younger than 2.5 Ga (O and Y, respectively) can now be directly compared also (Fig. 7). These results clearly depict the mineral deposit diversity between cratons and younger crust; they also depict a real difference (per unit area) in enrichment of specific elements between old and young crust on all three continents. In general the older fragments are more enriched in all element groups, except for tungsten in South America, and for the Sn, Sb, and UThREE groups in Australia. This result is probably independent of the “exposure” problem mentioned above and therefore a strong conclusion.
The strongest spatial coefficient observed is between tin and antimony (Sn and Sb) and the old crust (cratons) of South America. A strong spatial coefficient for Sn and Sb is also observed for Africa, and may be controlled mostly by the West African cratons that were part of the Amazonian Craton until Gondwana breakup. This supports a long (3.0–0.5 Ga) conterminous history for this old crust (Trompette, 1994). We have previously explored this Gondwanatin association between West Africa and South America to refute a frequently advocated fit between them and eastern North America (where there are no Archean–Mesoproterozoic tin deposits or occurrences), a relation that is important to models of the proposed Mesoproterozoic supercontinent Rodinia (de Wit et al., 1999).
In summary, it seems that the younger crust of the continents in the Southern Hemisphere is relatively enriched in the lithophile element group (UThREE), whereas the older crust of all three continents is affiliated to a greater degree with concentrations of siderophile element group (Cr Ni PGE Ti). In contrast, there is no discernable degree of difference in concentrations of chalcophile element group (base metals) between old and young crust. Our integrated results also support general statements that old crust is more richly endowed with mineral deposits than young crust (Fig. 7).
Metallogenesis on a Gondwana Scale
The above analyses were repeated using all the data of the cratons and the three continents combined. This allows us to compare and contrast the mineral inventory of all old (Archean) crust with that of younger crust of the three continents combined as one, as they would have been in Gondwana times (ca. 200–500 Ma). To some degree this corrects a bias in the analyses by concentrations of recent mineral deposits related to one specific present-day plate tectonic environment (such as subduction below South America) that may skew the results and interpretations when using a single continent only. On this subglobal continental scale we combine all old crust (cratons) across all combined continents, but we apply no “weighting” to allow for differential development and exploration; both the relatively “rich” and “poor” of the Southern Hemisphere are dealt with collectively.
The data are summarized in Tables 4A to 4B and plotted in Figure 8. From this it is clear that Archean crust is indeed mineralized to a significantly greater degree than younger crust (except for the strongly lithophile element group UThREE), thus hinting at the possibility that young crust may have inherited at least some metal enrichment during remobilization of its embedded cratons (e.g., tin in South America, gold and PGE in Africa).
DISCUSSION AND CONCLUSIONS
By applying an “exploration index” to our existing mineral deposit data, we have attempted to compare the mineral inventory of different cratons across continents with variable exploration histories. Although our exploration index needs more rigorous testing, incorporating this type of data allows a more informed and robust comparative analysis between mineral riches of different cratons. The results confirm our earlier work (de Wit and Thiart, 2005) that, per unit area of crust, there is a greater concentration of mineral deposits in Archean cratons relative to younger crust. Although the mineral inventory is greater in cratonic crust than younger crust, we also confirm that a significant mineral diversity exists among cratons, and that each craton has a unique metallogenic fingerprint. These differences resemble variations of Phanerozoic mineralization on continents at scales that clearly link mineral deposits to different plate tectonic environments (e.g., oceanic arcs, continental subduction zones; Sawkins, 1990; Windley, 1995). If we assume a similar origin for the processes of mineralization in Archean times, this might provide a strong basis of support for models that advocate that plate tectonics operated on the Archean Earth. For example, many of the Neo-archean cratons have strong Au and base-metal signatures that fit with mineralization of subduction-accretion models as proposed for some of these cratons on the basis of geologic and geophysical evidence (e.g., Herrington et al., 1997; Barley et al., 1998).
Cumulative evidence indicates that cratonic crust is more enriched in mineral deposits than younger crust (younger than 2.5 Ga). The greater concentration of this Archean mineralization may represent more efficient mineralization processes, perhaps related to higher heat and/or volatile loss from the early Earth compared to today (Abbott et al., 1994; de Wit and Hart, 1993; de Wit and Hynes, 1995; Pollack, 1997; de Wit, 1998), in which case our results may be interpreted to reflect greater “partition coefficients” of selected elements between cratonic crust and (now depleted) mantle during the formation of Archean lithosphere (Fig. 9).
However, we cannot rule out the possibility that the higher concentration of Archean mineralization represents a greater preservation potential of cratons (and their mineral deposits) relative to younger continents, as the presence of (Archean) diamonds in cratons might imply. In this case, the greater mineral wealth of cratons may be merely a consequence of greater rates of recycling of young continental crust relative to that of old Archean crust preserved in cratons (Fig. 9). Because the majority of the cratonic crust is preserved at low grades of metamorphism (and is thus representative of the upper crust) this seems an equally valid interpretation.
If this interpretation has merit, it implies significant changes in the efficiency of recycling of young continental crust since the Archean. Because subduction is the principal mechanism by which the hydrosphere recycles into the mantle to generate continental crust enriched in mineral deposits, secular change in crustal mineralization could be used to track secular changes in the chemical composition of Earth's fluid envelope and the redox state of its upper mantle.
Either way, Earth's crust appears to signal a decrease in new mineral deposit mass and diversity through time. But, as a word of caution, our results are based on only one mineral database tied to only one specific exploration index. In addition only two relatively long time spans are compared (e.g., Archean and younger). It would be wise to explore further, using a different databases and development indices, as well as with greater time resolution, before our general conclusions can be established beyond reasonable doubt.
We are grateful to Graham Bonham-Carter and Fritz Agterberg for their interest and encouragement in our work. Constructive discourse on an early draft by Keith Long, Stephen Kesler, and an anonymous reviewer helped to clarify some of our thoughts and improve this contribution significantly. This work is supported through funds of the South African National Research Foundation (NRF). This is AEON contribution number 008.