INTRODUCTION
Recent reporting of human footprints in the muds of a New Mexican playa apparently dating to 23,000–21,000 cal yr BP at White Sands National Park (WHSA; Bennett et al., Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) has generated keen interest among both the scientific community and the public at large. Age estimates based on radiocarbon dating of “seeds” (technically, drupelets; see Kantrud Reference Kantrud1991, p. 6) of a submerged aquatic plant (Ruppia cirrhosa) stratigraphically bracketing the footprint trackways seem to show that people entered the Americas much earlier than is commonly accepted, thus opening a wide range of new possibilities as to how, where, and when this initial emigration occurred. However, as pointed out by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb), Ruppia cirrhosa obtains its carbon for photosynthesis from the water in which it grows rather than directly from the atmosphere. The water used by the plant may contain dissolved inorganic carbon (DIC) that is depleted in radioactive carbon (14C) relative to the carbon in atmospheric CO2, resulting in radiocarbon-based age estimates of the plant tissues that can be much older than the plants themselves (e.g., Deevey et al., Reference Deevey, Gross, Hutchinson and Kraybill1954; Marty and Myrbo, Reference Marty and Myrbo2014). Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) recognized this potential problem but did not adequately account for it. We suggested (Madsen et al., Reference Madsen, Davis, Rhode and Oviatt2022) that these age estimates needed independent confirmation through other dating methods or sources. Unfortunately, space limitations imposed on Technical Comments for Science precluded an expanded discussion, so here we take the opportunity to provide a more detailed explanation as to why we think such confirmation is necessary.
If the ages of Ruppia drupelets are only several thousand years older than the true age of the footprints, what difference does it make if the footprints are only, for instance, 18,000–17,000 cal yr BP, rather than 23,000–21,000 cal yr BP? That would still mean the footprint site in the Tularosa Valley could well be one of the oldest, if not the oldest, well-dated archaeological site in the Americas. But a difference of only a few thousand years can be crucial for informing our understanding of how, and from where, ancient Native Americans first moved into the New World. Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, p. 1531) initially held that their footprint chronology “confirms that humans were present in North America before the glacial advances of the LGM closed the Ice-Free Corridor and the Pacific Coastal Route and prevented human migration from Asia,” and argued that any radiocarbon reservoir is expected to be “insignificant” and “less than a few hundred years” (Bennett et al., Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb). Now, however, in a response to our Technical Comment for Science, they indicated only that “we stand behind our conclusion that humans were present in southwestern North America during the Last Glacial Maximum” (Pigati et al., Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 1). We think there is a crucial difference between these two conclusions. If an adjustment of a few thousand years is applied to their measured Ruppia drupelet ages, the timing of human presence in the lake basin could shift into the age range of other pre-Clovis North American sites. While we agree that people were likely in the Americas during the latter part of the local last glacial maximum (LLGM; Clark et al., Reference Clark, Dyke, Shakun, Carlson, Clark, Wohlfarth, Mitrovica, Hostetler and McCabe2009), whether or not they had arrived before the ice free corridor (IFC) closed or the Pacific coastal route became available is a more controversial conclusion and needs to be confirmed by more accurate dating.
The age of opening of the IFC between the Laurentide and Cordilleran ice sheets after the last glacial maximum is now well known—a slew of cosmogenic 10Be exposure ages, combined with geomorphic mapping and high-quality minimum-limiting 14C ages indicate it opened sometime after ca. 13,500 cal yr BP, possibly ca. 13,200 cal yr BP (Heintzman et al., Reference Heintzman, Froese, Ives, Soares, Zazula, Letts and Andrews2016; Norris et al., Reference Norris, Tarasov, Monteath, Gosse, Hidy, Margold and Froese2021). The age of ca. 13,500 to 13,200 cal yr BP for the opening of the IFC is too late for even proponents of a “Clovis First” model to postulate an initial entry via a land route from Beringia through the IFC, because sites with Clovis technology appear to be widespread throughout southeastern North America by at least ca. 13,350 cal yr BP (Haynes et al., Reference Haynes, Donahue, Jull and Zabel1984; Haynes, Reference Haynes, Taylor, Lon and Kra1992; Waters and Stafford, Reference Waters and Stafford2007; Waters et al., Reference Waters, Stafford and Carlson2020). However, when the IFC first closed at the beginning of the LLGM is less well known due to data limitations, with mapping of the ice sheet margins suggesting the route was closed sometime between ca. 30,000–25,000 cal yr BP (Dalton et al., Reference Dalton, Stokes and Batchelor2022). But if the pre-LLGM closing of the IFC occurred prior to the oldest 14C ages from the Lake Otero footprint site, as Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) suggested, humans may well have entered the Americas via an interior IFC route before the LLGM. This is one major reason why the Ruppia dates from the Tularosa basin need to be confirmed.
Another major reason why confirmation of the footprint Ruppia ages is important is that those ages conflict with a variety of other data that suggest humans first entered the Americas during the LLGM rather than prior to its beginning. Many of these data are derived from age estimates provided by paleogenetics, with molecular clocks indicating that initial occupation of the Americas south of the ice sheets occurred sometime after ca. 20,000 yr BP (e.g., Raghavan et al., Reference Raghavan, Steinrücken, Harris, Schiffels, Rasmussen, DeGiorgio and Albrechtsen2015; Llamas et al., Reference Llamas, Fehren-Schmitz, Valverde, Soubrier, Mallick, Rohland and Nordenfelt2016; Moreno-Mayar et al., Reference Moreno-Mayar, Vinner, Damgaard, de, de la Fuente, Chan, Spence and Allentoft2018; Pinotti et al., Reference Pinotti, Bergstrom, Geppert, Bawn, Ohasi, Shi and Lacerda2019; Roca-Rada et al., Reference Roca-Rada, Politis, Messineo, Scheifler, Scabuzzo, González and Harkins2021; Willerslev and Meltzer, Reference Willerslev and Meltzer2021; Colombo et al., Reference Colombo, Traverso, Mazzocchi, Grugni, Migliore, Capodiferro and Lombardo2022). If the Lake Otero footprint ages are valid, then as Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 2) suggested, “researchers in archaeology and allied fields may need to reexamine existing hypotheses and the underlying data related to the peopling of the Americas in light of these discoveries.” It is certainly possible the Ruppia ages from a single site in a New Mexico lake basin are valid and that age estimates from a variety of other fields are invalid, but we think the Ruppia 14C ages should be confirmed with results from other dating methods.
RUPPIA PLANT GROWTH AND FRUIT PRODUCTION
An assessment of the validity of 14C age estimates based on R. cirrhosa must take into account how and where these plants grow, as well as the nature of seed production and dispersal. The genus Ruppia, commonly called widgeongrass (aka wigeongrass) or spiral ditchweed, consists of several species of submerged rooted aquatic vascular plants that inhabit shallow saline to alkaline lagoons, estuaries, lakes, and ponds worldwide, mainly along coastlines, but with scattered inland populations as well. Ruppia often forms large monospecific beds that serve as productive substrates for many aquatic organisms and ecosystems under conditions unfavorable to other aquatic plants, and it is widely regarded as a key ecological engineer (Verhoeven, Reference Verhoeven1980). A wide range of waterfowl and shorebirds favor the plants and their fruits as a considerable part of their diet (Martin et al., Reference Martin, Zim and Nelson1961).
Ruppia predominates in well-lit saline or alkaline waters from ~0.5–2 m depth, although it can tolerate even deeper waters (to ~4.5 m) and wide variation in water levels (Kantrud, Reference Kantrud1991). Ruppia tolerates a wider range of water salinity than any other aquatic vascular plant (Brock, Reference Brock1981; Murphy et al., Reference Murphy, Kinsey and Durako2003; Triest et al., Reference Triest, Beirinckx and Sierens2018); it does poorly in fresh, soft, or slightly acidic waters. Water transparency and availability of sunlight are also critical growth factors; it performs well in clear, still, or slowly moving waters, but not turbid, shaded, or turbulent conditions (possibly caused by wind and waves). Ruppia can survive periodic exposure to air in wetlands subject to tidal fluctuations, but not persistent drought or desiccation, and Ruppia does not grow in permanently subaerial terrestrial habitats.
Ruppia plants are simple, often nearly threadlike, with reduced vegetative and reproductive structures. Growth forms and life histories vary greatly depending on environmental conditions and phenotypic plasticity (Kantrud, Reference Kantrud1991; Mannino et al., Reference Mannino, Menéndez, Obrador, Sfriso and Triest2015; Triest et al., Reference Triest, Beirinckx and Sierens2018). In North America, two forms are recognized as distinct end members of one or more broad species complexes, with R. maritima considered mainly coastal and R. cirrhosa mainly inland (Setchell, Reference Setchell1946; Mason, Reference Mason1957; Kantrud, Reference Kantrud1991; Thorne, Reference Thorne and Hickman1993; Haynes, Reference Haynes2000). Ruppia cirrhosa is distinguished primarily by having long flowering tendrils (peduncles) with spiral coils up to 10 dm long and small non-fleshy fruits (drupelets, sometimes called “seeds”), while the R. maritima form has shorter peduncles, few to no coils, and smaller fruits.
The flexuous spiraling peduncle serves an important function in R. cirrhosa's reproductive ecology, which depends on hydrophilous pollination on the water surface of its submerged aquatic habit (Les, Reference Les1988; Cox and Knox, Reference Cox and Knox1989). Bearing two tiny naked flowers at its tip, the peduncle elongates to place the flowers’ stigmas at the water surface to be fertilized by rafts of floating pollen; after pollination, the peduncle may recoil and draw the floral structures back under water during fruit development (Taylor et al., Reference Taylor, Giffei, Dang, Wilden, Altrichter, Baker, Nguyen and Oki2020). Underwater self-pollination of the smaller R. maritima occurs via pollen attached to tiny gas bubbles (Verhoeven, Reference Verhoeven1979; Richardson, Reference Richardson1983; Kaul, Reference Kaul1993; Lacroix and Kemp, Reference Lacroix and Kemp1997; Triest and Sierens, Reference Triest and Sierens2015; Triest et al., Reference Triest, Beirinckx and Sierens2018). Because pollination cannot happen in a terrestrial setting, such as an alluvial mudflat, no drupelets form (no drupelets = no radiocarbon dates at WHSA).
Drupelets are produced in large quantities within two weeks after first flowering (Kantrud, Reference Kantrud1991). These may drop to the bottom to lie dormant until the next growing season, either submerged or exposed to the air. The drupelets are very drought-resistant (and can persist as sub-fossils in sediments for millennia). They are often transported for considerable distances on floating mats, or via the guts of fish and waterfowl. Kantrud (Reference Kantrud1991, p. 7) noted that “wigeongrass drupelets mix with small amounts of other plant material, forming compact balls up to the size of small watermelons—these are often found along the beaches of saline lakes in windy locations. Such balls presumably form by wave action.” Germination requires at least 5–10 cm of water depth and is affected by water and sediment chemistry, salinity, oxygen content, temperature, water permanency, and depth.
Depositional environments
This basic Ruppia biology is important in the Lake Otero context because the dated Ruppia drupelets are said to occur in “distinct seed layers” in alluvial deposits above lake level (Pigati et al., Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 1). Ruppia would not germinate, grow, or persist in a subaerial environment, and the drupelets must have developed in water, not on land. How, then, did layers of them become deposited in subaerial alluvium? Three possibilities come to mind. (1) The Ruppia drupelet layers could represent windrow strandlines of Ruppia detritus floated by wind and wave action to the lake's shoreline from nearshore Ruppia growth beds. Such detrital strandlines are common around lagoons, estuaries, bays and lakes where Ruppia is plentiful (Verhoeven, Reference Verhoeven1979). (2) The layers could represent old-growth Ruppia beds from a time when the lake had been higher, that subsequently were left dry and exposed as the lake dropped. The presence of Ruppia drupelet layers in several bedding planes in the stratigraphic profile may suggest an alternating rising and falling lake, with repeated re-establishment of Ruppia beds followed by subaerial exposure. (3) Ruppia drupelets and other detritus might have been washed down from upslope by fluvial processes if upstream water sources were deep enough to support Ruppia growth and reproduction. One of the streams that flowed toward Lake Otero (Lost River, east of the WHSA footprint locality) could have been a source, and Ruppia has been reported growing elsewhere in the Tularosa basin (Turner, Reference Turner1987). How such a stream source could have deposited the dense Ruppia drupelet beds illustrated by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) without containing a mixture of other terrestrial plant remains is difficult to imagine, though (in addition, see the exchange between Haynes, Reference Haynes2022, and Pigati et al., Reference Pigati, Springer, Holliday, Bennett, Bustos, Urban, Reynolds and Odess2022b).
Assimilation of old carbon by Ruppia
Ruppia routinely assimilates carbon via epidermal photosynthesis using dissolved inorganic carbon (DIC) in the water, including aqueous carbon dioxide (CO2), bicarbonate (HCO3−) (Lucas, Reference Lucas1983; Sand-Jensen and Gordon, Reference Sand-Jensen and Gordon1984; Hellblom and Axelsson, Reference Hellblom and Axelsson2003), and/or carbonate (CO32−). CO2 dissolves in water to form carbonic acid, and HCO3− and CO32− are produced during the dissolution of CaCO3 by carbonic acid (Drever, Reference Drever1988; Falkowski and Raven, Reference Falkowski and Raven1997). Ancient 14C-depleted CaCO3 is abundant in sediments and bedrock of the Tularosa Valley and surrounding mountains (e.g., Herrick, Reference Herrick1900, Reference Herrick1904; Meinzer and Hare, Reference Meinzer and Hare1915; Leuth et al., Reference Leuth, Giles, Lucas, Kues, Myers and Ulmer-Scholle2002; Lucas and Hawley, Reference Lucas, Hawley, Leuth, Giles, Lucas, Kues, Myers and Ulmer-Scholle2002).
Ruppia readily uses carbon in the form dissolved CO2 in the water, but the concentration of carbon from that source is typically low even in air-saturated water, and Ruppia plant growth is strongly limited using that source alone (Sand-Jensen, Reference Sand-Jensen1983; Fontana, Reference Fontana2007). As with many submerged aquatic plants, Ruppia uses various carbon-concentrating mechanisms to extract bicarbonate from the water to overcome the dissolved CO2 carbon limitation (Sand-Jensen and Gordon, Reference Sand-Jensen and Gordon1984; Madsen and Sand-Jensen, Reference Madsen and Sand-Jensen1991; Marty and Myrbo, Reference Marty and Myrbo2014; Larkum et al., Reference Larkum, Davey Peter, Kuo, Ralph and Raven2017). The use of dissolved carbonates by Ruppia growing in Lake Otero is likely to have been substantial even if its shallow waters were well-mixed and in CO2 equilibrium with the atmosphere.
Dissolved carbonate in the water column is the primary source of the hard-water or reservoir effect on radiocarbon-based age estimates (Olsson, Reference Olsson, Seuss and Berger1980). Dissolved carbonate has a substantially different δ13C signature than aqueous CO2, so the δ13C of the dated Ruppia drupelets may provide a useful indicator of the relative proportion of the atmospheric versus carbonate sources used by Ruppia in the Lake Otero sediments (e.g., Smith and Walker, Reference Smith and Walker1980; Marcenko et al., Reference Marcenko, Srdoc, Golubic, Pedzic and Head1989; Larkum et al., Reference Larkum, Davey Peter, Kuo, Ralph and Raven2017). Unfortunately, Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) did not give δ13C values, which would help determine whether, and to what extent, dissolved carbonate was a source of carbon in the dated Ruppia drupelets (we hope that those δ13C values will be made available).
Instead, one of the arguments used by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, p. 1530) was that “the geologic and hydrologic settings along the shallow lake margin that hosts the human trackways make it unlikely that stands of emergent aquatic plants would harbor prominent hard-water effects.” It is true that emergent aquatic plants (such as bulrush or cattail) are not expected to display prominent hard-water effects because their aerial leaves obtain their carbon from the atmosphere above the waterline, not the water; only very limited carbon assimilation takes place via submerged roots. For that reason, the fruits or other remains from those emergent aquatic plants would be an ideal dating alternative to the Ruppia drupelets (such remains unfortunately do not appear to be present in the WHSA locality sediments). In contrast to Bennett et al.'s (2021a) assertion, submerged aquatic plants such as Ruppia that obtain their carbon from the water column and not directly from the atmosphere should be expected to exhibit hard-water effects.
A search of the literature for reported radiocarbon ages of Ruppia samples shows that Ruppia very commonly yields radiocarbon ages that exhibit prominent radiocarbon reservoir effects (Table 1). Thus it seems that radiocarbon dates of Ruppia samples should be assumed to show a radiocarbon reservoir effect, and substantial efforts are required to demonstrate that the Lake Otero Ruppia did not utilize DIC derived from dissolution of limestone in photosynthesis, which resulted in a significant reservoir offset of its radiocarbon content. δ13C values may provide a useful test of that hypothesis.
Examples of reservoir ages (14C yr) from lacustrine deposits based on Ruppia propagules.
DATING OF MODERN RUPPIA
To assess the magnitude of a potential hard-water reservoir effect on radiocarbon dates from the WHSA Ruppia samples, we sought to date plants of known age from a natural system within the Tularosa basin. On June 3, 1947, Edward F. Castetter and W.J. Koster visited Malpais Spring, a large spring/salt marsh/shallow lake ecosystem on the southwest end of the Carrizozo Malpais lava flow (Myers and Naus, Reference Myers and Naus2004), in the watershed of Pleistocene Lake Otero and ~45 km north of the WHSA footprint site. There they collected Ruppia cirrhosa specimens as well as a specimen of chairmaker's bulrush (Schoenoplectus americanus), a common emergent aquatic plant in the wetland. We obtained samples from these collections from the University of New Mexico Herbarium for radiocarbon dating and 13C isotopic analysis. We expected that the Ruppia should utilize dissolved carbonate in the water and would exhibit a significant hard-water reservoir effect by having a radiocarbon date significantly older than modern and high δ13C consistent with its carbonate assimilation physiology. Water quality analysis of Malpais Spring by Cruz (Reference Cruz1983) indicated that it had a measured pH = 7.8, specific conductance = 6050, dissolved CO2 = 0.2 mg/l, dissolved organic C = 0.6 mg/l, dissolved inorganic C = 7.0 mg/l, with a 13C/12C ratio = −8.6‰. Measured alkalinity (CaCO3) = 44–49 mg/l, and bicarbonate (HCO3−) = 54–60 mg/l. Cruz (Reference Cruz1983) reported an age of the water from Malpais Spring of >27,900 14C BP with <3.1% modern 14C, but Myers and Naus (Reference Myers and Naus2004, p. 1) cautioned that “this age may not be accurate because the aquifer system is not a closed system.” In contrast, we expected that the Schoenoplectus would have assimilated its carbon from the atmosphere and exhibit little to no hard-water offset, having a modern or near-modern radiocarbon age and a low δ13C value typical of C3 plants.
The Ruppia specimen (UNM 0109200) yielded an age estimate of 6420 ± 30 14C BP, with δ13C = −13.6. The age estimate results in a median Calib8.20 (Intcal20) calibration of ca. 7353 cal yr BP (7274–7422 cal yr BP, 2σ range). The Schoenoplectus specimen (UNM 0066467) yielded an age estimate of 310 ± 30 14C BP (median 389 cal yr BP, 301–460 cal yr BP, 2σ range), with δ13C = −23.9. These values confirmed our expectations. Ruppia, growing in spring-fed wetlands within the Lake Otero watershed, appears to utilize the abundant dissolved carbonate and exhibits a large hard-water effect (on the order of ca. 7350 years in this instance), whereas emergent aquatic vegetation inhabiting the same water does not exhibit such a large effect (although it may exhibit some smaller effect).
Due to unmeasured variables, such as possible differences in sources and amounts of waters supplying paleolake Otero, as Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a) noted, the strong hard-water reservoir value exhibited by this date on modern Ruppia cannot be directly applied to the WHSA specimens. However, if this value were consistent with the postulated hard-water effect of Lake Otero waters on the Ruppia at the WHSA footprint site, then the dates reported by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) would be on the order of ca. 15,500–13,500 cal yr BP, which is similar in age to those for a number of other dated sites in North America (e.g., Jenkins et al., Reference Jenkins, Davis, Stafford, Campos, Hockett, Jones and Cummings2012; Halligan et al., Reference Halligan, Waters, Perrotti, Owens, Feinberg, Bourne and Fenerty2016; Waters et al. Reference Waters, Keene, Forman, Prewitt, Carlson and Wiederhold2018; Williams et al., Reference Williams, Collins, Rodrigues, Rink, Velchoff, Keen-Zebert, Gilmer, Frederick, Ayala and Prewitt2018; Davis et al., Reference Davis, Madsen, Becerra-Valdivia, Higham, Sisson, Skinner and Stueber2019). Because Lake Otero is known to have similar carbonate-rich waters, it would make sense that Ruppia growing in its waters would exhibit a similar reservoir effect.
PROBLEMS WITH THE VALIDITY OF RUPPIA CIRRHOSA RADIOCARBON DATES IN THE TULAROSA VALLEY
Despite the well-known problems with dating Ruppia drupelets due to their incorporation of DIC in its photosynthesis, Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) provided several reasons why they think the Ruppia age estimates from WSHA Locality 2 are “robust.” Madsen et al. (Reference Madsen, Davis, Rhode and Oviatt2022) suggested this reasoning is more problematic than Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) implied. Here, we expand on this discussion and examine other problematic issues raised by Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a).
Alluvial/Aeolian sedimentation
Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021b, p. 9), in arguing that the stratigraphic context of the radiocarbon-dated Ruppia drupelets gives reason to suspect a radiocarbon reservoir did not affect the plants, stated that “…the hydrologic conditions at WHSA Locality 2 are not conducive for perpetuating significant age offsets through hard-water effects because water levels must have been quite shallow if humans and megafauna were walking on the plants during or shortly after the plants were alive. This interpretation is based on our observations that well-preserved R. cirrhosa [drupelets] were often found embedded in human footprints at WHSA Locality 2. Therefore, long-term stratification and isolation from the atmosphere could not have taken place in this setting.” This statement contrasts with the statement of Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 1) that “Although aquatic plants such as Ruppia living in lakes that are this deep may indeed be subject to hard-water effects, the [drupelets] chosen for dating at WHSA Locality 2, as well as the footprints they stratigraphically constrain, were not present in lake deposits.”
This clarification, including the phrase “the largely gypsiferous and calcareous sediments contain both alluvial and aeolian components” (Pigati et al., Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 2), changes possible interpretations of the R. cirrhosa 14C ages. By definition, alluvial and aeolian sediments are redeposited, the former by water and the latter by wind. If the drupelets contained in the sediments studied by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) were redeposited in terrestrial beds, then they are older, by some unknown amount, than the footprints found in these beds. Exactly how much older is unknown. They may be relatively contemporaneous, as Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) and Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a) assumed, or they could easily be hundreds to thousands of years older. Unless the origin of the dated R. cirrhosa drupelets is known, it isn't possible to know which of these interpretations is correct.
Atmospheric mixing in shallow lakes
Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) thought that the dates on R. cirrhosa drupelets were robust partly based on the reasoning that the radiocarbon reservoir effect in a shallow lake would be minimal because of atmospheric mixing. But, as we have noted, dating of modern Ruppia from shallow lakes in different locations around the world in many cases produce age estimates that are thousands of years older than their correct age (Table 1). However, an extended discussion of these alternate interpretations is complicated because of the possibility that the dated R. cirrhosa drupelets at the WHSA site were redeposited. Since it is impossible to know the depth of the water in which the Ruppia plants originally grew, debate about whether or not atmospheric mixing in shallow lakes eliminates problems with a radiocarbon reservoir is pointless, at least in the case of the WSHA Locality 2 age estimates.
Age and effects of groundwater in the Tularosa Valley
Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 1) contended that “…the consistency of our calibrated ages argues against contamination by groundwater with hardwater effects on the order of many thousands of years.” One of the reasons we question this conclusion is based on the geology of Tularosa Valley, which is well understood (e.g., Herrick, Reference Herrick1900, Reference Herrick1904; Meinzer and Hare, Reference Meinzer and Hare1915; Leuth et al., Reference Leuth, Giles, Lucas, Kues, Myers and Ulmer-Scholle2002; Lucas and Hawley, Reference Lucas, Hawley, Leuth, Giles, Lucas, Kues, Myers and Ulmer-Scholle2002). The hydrology of the basin is similar to that of other hydrographically closed basins in western North America and elsewhere in the world (e.g., Langbein, Reference Langbein1961; Rosen, Reference Rosen and Rosen1994). The Tularosa Valley is surrounded by mountain ranges that include a glaciated mountain (Sierra Blanca) and other high mountains (the Sacramento Mountains) on the east, and the San Andres Mountains on the west, which are lower in elevation and were not glaciated. In both bounding mountain ranges, and in other high-elevation terrain within the drainage basin of Tularosa Valley, groundwater is recharged. In the closed Tularosa basin (closed to surface water), some groundwater seeps southward into another closed basin (Love et al., Reference Love, Allen, Morgan, Myers, Rawling, McLemore, Timmons and Dunbar2014, p. 135). Closed-basin lakes are known for their frequent and rapid water-level changes (Bohacs et al., Reference Bohacs, Carroll, Neal, Mankiewicz, Gierlowski-Kordesch and Kelts2000), which in some cases might lead to shallow lakes appearing and disappearing on seasonal, decadal, or centennial time scales. It seems likely that the behavior of shallow Lake Otero would have been similar to that of other closed-basin lakes.
The water table on the floor of the basin is relatively high (typically within a few meters of the ground surface), and there is an upward component of groundwater flow (Meinzer and Hare, Reference Meinzer and Hare1915), so that the basin floor is a large groundwater discharge area (capillary rise above the water table and evaporation at the surface). Several studies in Tularosa Valley have documented ages for groundwater in wells and springs of hundreds to thousands of years (Cruz, Reference Cruz1983; Huff, Reference Huff, Leuth, Giles, Lucas, Kues, Myers and Ulmer-Scholle2002; Mamer et al., Reference Mamer, Newton, Koning, Timmons and Kelley2014). Therefore, the discharge of old groundwater into Pleistocene lakes on the basin floor is more likely than not to have occurred. If it were postulated that groundwater did not discharge into Pleistocene lakes in this basin, that groundwater would have to be disposed of in some way.
Newton and Allen (Reference Newton and Allen2014, p. 3) noted that the “…regional groundwater component [in the Tularosa basin] has a distinct geochemical signature and is greater than 10,000 years old”; they also said (p. 17) “This water likely represents regional groundwater that originated as precipitation in the Sacramento Mountains thousands of years ago.” These statements only partially align with the comment by Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 1), that “The age of shallow groundwater in the White Sands area ranges from a few hundred to a few thousand years [Newton and Allen (Reference Newton and Allen2014)], which is in stark contrast to groundwater emerging from deep-seated, regional-scale aquifers that are known to impart exceptionally large hard-water effects [in other valleys, specifically in southern Nevada].” Considering that old groundwater is making its way to the surface in Tularosa Valley, it seems reasonable to assume, although it does not prove, that the radiocarbon ages of the remains of aquatic organisms that lived in Tularosa Valley are likely to be older than carbon-bearing materials (organic and inorganic compounds) derived from organisms that received their carbon directly from the contemporaneous atmosphere.
Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a) pointed out correctly that groundwater ages could be spatially and temporally variable. WSHA Locality 2 is a single location, so spatial variability is not in question in this case. Taken at face value, the radiocarbon dates of Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) span a range of about 2000 years. However, considering the uncertainty in those ages, as discussed here, the apparent 2000-year period may not be correct—the length of the period is unknown, but easily could be larger or smaller than 2000 years. Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a) cited a study in a lake basin in China (Zhou et al., Reference Zhou, Chen, Wang, Yang, Qiang and Zhang2009) where the reservoir effect during the late Holocene varied from ca. 4300 yr to ca. 2700 yr (a range of ca. 1600 yr) during a period of ca. 2100 yr. Each groundwater-discharge setting is different, however, and the details of the situation in Tularosa Valley are not currently known. We regard temporal variability in the radiocarbon reservoir at the WSHA site as a possible, but undocumented, complicating factor in determining the true age of the footprints.
Stratigraphic order and validity of radiocarbon ages
Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, p. 1530) stated “…our calibrated 14C ages maintained stratigraphic order even when samples were separated by a few centimeters, which would not be the case if hard-water effects were large and variable.” We agree that correct stratigraphic order is necessary if the sequence of ages is to be interpreted as valid, but correct stratigraphic order is not sufficient by itself to indicate validity because the ages could be in stratigraphic order but not representative of the age of sediment deposition.
An example of correct stratigraphic order of radiocarbon ages, but incorrect individual ages, is provided by Thompson et al. (Reference Thompson, Oviatt, Honke, McGeehin, Oviatt and Shroder2016), where >50 radiocarbon ages (not of Ruppia, but other organic materials) in a core of sediments from Great Salt Lake, Utah, are in stratigraphic order (with few exceptions). However, all of the ages had to be adjusted by some amount, as shown by comparison with independently determined ages at five stratigraphic levels. Thompson et al. (Reference Thompson, Oviatt, Honke, McGeehin, Oviatt and Shroder2016) chose 1800 yr as an appropriate average for adjustment based on the available evidence (i.e., the unadjusted radiocarbon ages of bulk organic matter in small samples from the Great Salt Lake core are 1800 radiocarbon years too old), but the exact amount required at each stratigraphic level is not known. The important observation in the context of the ages of the White Sands footprints is that it is possible for ages to be in proper stratigraphic order, but individually incorrect (i.e., “reliable,” but not “valid”).
If, in contrast to the assumption of Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a), there were no variability in the reservoir effect during the ca. 2000-yr dated interval in this case, a sequence of radiocarbon dates from that time interval might be in correct stratigraphic order, even if samples were separated by only a few centimeters, but the entire sequence could be shifted toward an older age by an amount proportional to the radiocarbon reservoir. Therefore, it would be important to determine independently whether a radiocarbon reservoir existed at the time the dated samples were forming. As discussed below, the age of redeposited charcoal fragments in combination with ages of aquatic organisms in Tularosa Valley, as reported by Allen et al. (Reference Allen, Love and Myers2009), does not adequately provide an independent test of the absence of a radiocarbon reservoir at the time humans created the footprints at WHSA locality 2.
The scattered pattern in the chronostratigraphic record reported from the western edge of Alkali Flat by Bustos et al. (Reference Bustos, Jakeway, Urban, Holliday, Fenerty, Raichlen and Budka2018, fig. S1; in a plot of those ages versus depth, ~48% of the variance is explained by a linear regression) suggests that the radiocarbon samples of “palustrine mud” they dated from a nearby section may have had unresolved problems, such as variable incorporation of carbon-bearing clastic debris in the mud samples, reworking and mixing as the muds were deposited, or a variable radiocarbon reservoir effect in that area. If reworking and/or radiocarbon reservoir effects are a problem on the west side of Alkali Flat, they may also be problematic on the east side of the flats where the Ruppia drupelets were found in association with the human footprints (Bennett et al., Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a).
Confirmation of the footprint chronology using other Tularosa Basin sections
Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021b) created a Bayesian model for radiocarbon ages from a stratigraphic section in the northern Tularosa basin (locality 8 of Allen et al., Reference Allen, Love and Myers2009) to support their notion that a possible long-term reservoir effect amounted to a few hundred years at most. However, we find their use of radiocarbon ages of charcoal fragments, “macrophytes,” and ostracodes, to be problematic. Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021b, fig. S14) referred to the charcoal fragments as “terrestrial material.” However, the charcoal fragments presumably were transported to the site of deposition and are not necessarily the same age as the deposition at that site. Allen et al. (Reference Allen, Love and Myers2009, p. 17–18) stated “…charcoal fragments were probably transported to depositional sites by alluvial processes, and uncertainty in the duration of transport and the antiquity of the plants when they died also suggests a maximum-age interpretation for those samples.”
The charcoal that ended up as fragments that were dated by Allen et al. (Reference Allen, Love and Myers2009) formed from stems of some plant that was receiving its carbon from the atmosphere and therefore would yield a radiocarbon age not influenced by a radiocarbon reservoir, but the chances that the plant that ultimately yielded the charcoal fragments was living at the site of deposition, and at the same time as sediment deposition, are slim. The transport mechanism(s), the history of potential temporary storage and re-entrainment, the location where the plant was growing, and the species of plant are not known. This means that the charcoal age should be regarded as a limiting age (that is, the age of deposition is equal to or younger than the charcoal age), not as a direct age of deposition of the sediment (as presumably the “macrophyte” and ostracode ages are, if they were not affected by radiocarbon reservoirs). The statement “…that terrestrial and aquatic material yielded concordant 14C ages between ~44 and 25 ka [44,000 and 25,000 cal yr BP]” (Bennett et al., Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, p. 1531) is based on assumptions that are unlikely to be correct.
Pluvial Lake Otero and climatic implications of the footprint ages
Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, p. 1531) suggested that pluvial Lake Otero reached levels higher than the footprint site before ca. 23,300 cal yr BP: “The depositional sequence reported here exhibits a potential paleoclimate signal in which lacustrine conditions were succeeded by alluvial sedimentation that occurred in temporal correspondence with abrupt warming during Dansgaard-Oeschger event 2, beginning at ~23.3 ka (Svensson et al., Reference Svensson, Andersen, Bigler, Clausen, Dahl-Jensen, Davies and Johnsen2008). This drying event resembles sequences observed in other paleohydrological records in the southwestern US (Menking et al., Reference Menking, Polyak, Anderson and Asmerom2018; Springer et al., Reference Springer, Mankera and Pigati2015).”
Despite this claim, the three records (from Otero, Estancia, and southern Nevada) resemble one another only in that they all exhibit a number of wet and dry episodes. The geologic record of fluctuations of Lake Estancia (Menking et al., Reference Menking, Polyak, Anderson and Asmerom2018), ~200 km north of Lake Otero, shows evidence of increased moisture at the same time that, according to the dating of the WHSA footprint record by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a), Lake Otero was dry enough for humans to leave footprints on the exposed lake floor (Fig. 1). The geologic record of wet periods in groundwater-discharge deposits near Las Vegas, Nevada (Springer et al., Reference Springer, Mankera and Pigati2015) shows both wet and dry episodes at the time human footprints were being formed in Tularosa Valley (using the ages reported by Bennett et al., Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a). While it is possible these other records are incorrect due to dating issues, it is also quite possible that the Lake Otero record, as reported by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a), is incorrect.
Diagram showing the relationship between Ruppia ages for the human footprints in the Lake Otero basin (red bar labeled “R” in the figure; the elevation scale on the left is for the red bar), as dated by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) compared with the history of Pleistocene Lake Estancia. The Lake Estancia record shown in the blue line (using the scale on the right side of the diagram, relatively wetter [high] or dryer [low]) is simplified from Menking et al. (Reference Menking, Polyak, Anderson and Asmerom2018, fig. 4; see also Allen and Anderson, Reference Allen and Anderson2000, fig. 12). According to Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a), Lake Otero dropped from an indetermined elevation to below the elevation of the footprint site after ca. 23,000 cal yr BP. The approximate duration of the part of the LLGM shown on the figure is marked with the light blue rectangle (the full duration of the LLGM is 26,500–19,000 cal yr BP; Clark et al., Reference Clark, Dyke, Shakun, Carlson, Clark, Wohlfarth, Mitrovica, Hostetler and McCabe2009).
Ages reported for the LLGM highstands of nearby Pleistocene lakes provide useful comparisons with Lake Otero. Lake Estancia highstands were between ca. 24,000 and 18,300 cal yr BP according to Allen and Anderson (Reference Allen and Anderson2000, fig. 12) and Allen (Reference Allen, Lucas, Morgan and Zeigler2005; note the use of radiocarbon years rather than calibrated years in older publications—we have calibrated those ages for this paper), or between ca. 22,000 and 20,000 cal yr BP (Menking et al., Reference Menking, Polyak, Anderson and Asmerom2018; see also Fig. 1). In Pleistocene Lake King, ~180 km southeast of Tularosa Valley, four highstands occurred between ca. 27,000 and 19,000 cal yr BP (Wilkins and Currey, Reference Wilkins and Currey1997). On a larger scale, the Devils Hole speleothem climate proxy record from southern Nevada shows that groundwater was at a relatively high level at ca. 19,850 cal yr BP, and that it was ~1.5 m lower between ca. 30,000 and 20,000 cal yr BP (Wendt et al., Reference Wendt, Dublyansky, Moseley, Edwards, Cheng and Spötl2018, fig. 4). At Lake Otero, the Late Pleistocene highstand was between ca. 18,900 and 18,700 cal yr BP (Allen et al., Reference Allen, Love and Myers2009).
POTENTIAL RESOLUTION
For the reasons outlined above, we think the footprint 14C ages on Ruppia drupelets need to be confirmed using other chronometric methods. As we noted previously, propagules or other suitable remains from terrestrial or emergent aquatic plants would not be expected to have the same hard-water issues. If available (and keeping in mind possible re-deposition of older materials), they could constitute a useful validity test of the Ruppia dates. To our knowledge, though, such remains are not available.
Madsen et al. (Reference Madsen, Davis, Rhode and Oviatt2022) suggested that optically stimulated luminescence (OSL) methods might be a useful way to validate the 14C ages. Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 2) were less hopeful, saying that OSL methods “…may not yield robust ages at WHSA Locality 2 because the largely gypsiferous and calcareous sediments contain both alluvial and aeolian components. Multigenetic deposits like these often violate the underlying assumption of total bleaching prior to deposition [Rhodes, Reference Rhodes2011].” We agree that when single/multiple aliquot methods are used, which involve averaging the luminescence of thousands to millions of individual grains, alluvial sediments may produce inaccurate OSL ages due to poor resetting or partial bleaching of some of the sediments as they are transported and redeposited. In newer single-grain methods, on the other hand, the luminescence signals of a series of separate grains (typically 100) are read individually. Poorly reset outliers producing aberrant older ages, as well as aberrant younger ages for grains introduced through bioturbation, can then be removed using statistical methods (e.g., Gilbraith and Green, Reference Galbraith and Green1990).
Gypsiferous and calcareous sediments are poor candidates for OSL dating, but the aeolian components of the WHSA Locality 2 sediments very likely contain quartz grains suitable for OSL dating methods. Bustos et al. (Reference Bustos, Jakeway, Urban, Holliday, Fenerty, Raichlen and Budka2018, Table S1) reported two OSL ages (26,130 ± 1920 cal yr BP and 41,350 ± 2650 cal yr BP) from Lake Otero basin deposits, which shows this method will work in these sediments. While OSL age estimates do not have the precision of 14C ages, they do have typical errors of only 5–10%. Therefore, single-grain age estimates could help provide a useful test of the robustness of the Ruppia 14C ages.
Allen et al. (Reference Allen, Love and Myers2009, p. 19) stated that “Kocurek et al. (Reference Kocurek, Carr, Ewing, Havholm, Nagar and Singhvi2007) report optically stimulated luminescence (OSL) ages on core samples obtained within the crescentic dunes of White Sands National Monument, with basal clay sediment at a depth of 9 m below the land surface yielding an age of 7.3 ± 0.5 ka and gypsum sand at a depth of 6 m yielding an age of 5.2 ± 0.4 ka.” This demonstrates that OSL does work on the gypsum-rich sediments of the basin and that the errors are likely to be small enough to test 14C age estimates of Ruppia drupelets.
CONCLUSIONS
We are open to the idea that humans could have entered the Americas during or prior to the LLGM. However, for the following reasons, we remain unconvinced that the evidence put forth by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynoldsb) and the response provided by Pigati et al. (Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a) represent secure evidence of such an event for many reasons. (1) The dated Ruppia drupelets are said to occur as discrete bedding layers in alluvial and aeolian deposits above lake level and “…were not present in lake deposits” (Pigati et al., Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a, p. 1). Ruppia does not grow in terrestrial environments, although it may survive in tidal mudflats for a limited time. Ruppia inhabits ponds, lakes, bays, and slow-moving streams with water depths ranging from ~0.5–2.5 m (it is more likely to be found in stream channels rather than in water spread thinly over flat terrain, such as is typical of Alkali Flat in Tularosa Valley). (2) Because the alluvial and aeolian deposits consist of redeposited sediments, the drupelets they contain must be redeposited as well and therefore older by some unknown amount than the human footprints they bracket. (3) The use of DIC by Ruppia growing in Pleistocene Lake Otero is likely to have been substantial, even if its shallow waters were well mixed with the atmosphere. It seems likely that Lake Otero's DIC contained a large component derived from dissolved carbonates that were not in equilibrium with the contemporaneous atmosphere, resulting in a hard-water reservoir effect that cannot be lightly dismissed. (4) Regional groundwater in the Tularosa Valley is >10,000 years old. The proportion of groundwater versus surface water in Lake Otero may have varied temporally and spatially, but possibly not. Either way, this has not been adequately investigated. (5) Dated modern R. cirrhosa plant material from Malpais Springs in the Lake Otero watershed shows a significant hard-water reservoir effect, with a date of ca. 7400 cal yr BP. Ruppia plants growing in the saline, carbonate-rich waters of Lake Otero also are expected to have had a significantly high reservoir effect. (6) The stratigraphic consistency of the Ruppia 14C ages does not provide definitive evidence of their accuracy because the persistent use of the same water source (in the absence of temporal variability) also would provide stratigraphic consistency. (7) Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a) used a Bayesian model from a stratigraphic section in the northern Tularosa basin to support their claim that their footprint chronology is “robust,” yet charcoal fragments from that section were redeposited and yield a maximum-limiting age. (8) The lake history for pluvial Lake Otero, as reconstructed from the Ruppia age estimates, is out of phase with that of other regional lakes, although the chronologies from other regional lakes are not in total agreement with each other. (9) A radiocarbon reservoir of less than a few hundred years was expected by Bennett et al. (Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a); however, the radiocarbon age of groundwater in the Tularosa basin was reported by the same authors (Pigati et al., Reference Pigati, Springer, Bennett, Bustos, Urban, Holliday, Reynolds and Odess2022a) to range from a few hundred to a few thousand years. If the Ruppia drupelet ages are a few thousand years too old (i.e., if old groundwater was influencing the Ruppia ages), then the ages of the human footprints in the Tularosa basin are within the age range of other pre-Clovis North American sites and do not confirm “…that humans were present in North America before the glacial advances of the LLGM closed the Ice-Free Corridor” (Bennett et al., Reference Bennett, Bustos, Pigati, Springer, Urban, Holliday and Reynolds2021a, p. 1531).
Given what we know about the hydrology of the Lake Otero (Tularosa) basin, and lacking an independent chronometric evaluation of the footprint-bearing deposits, we cannot know the ages of the Ruppia drupelets with certainty. If the Ruppia radiocarbon ages can be confirmed, the stratified Tularosa Valley footprints remain extremely important because that would overturn current conventional wisdom in a number of scientific fields and force us to rethink both our analytic approaches and the conclusions we have derived from them. However, until independent age controls are established for the footprint-bearing stratigraphic sequence, we can only say for sure that humans were present at Lake Otero sometime after the radiocarbon age of the Ruppia drupelets and before the extinction of mammoths in the Southwest (Faith and Surovell, Reference Faith and Surovell2009) (i.e., sometime between ca. 21,000–12,500 cal yr BP).
Acknowledgments
We wish to thank Bruce Allen for useful comments as we were preparing this paper. We are grateful to Dr. Hannah Marx, Curator of the University of New Mexico Herbarium, and the herbarium staff for providing samples from specimens collected at Malpais Springs. We also thank Donald K. Grayson, David J. Meltzer, and the editors of Quaternary Research for useful reviews of initial drafts.
Competing Interests
The authors of this paper have no competing interests.
