The rise and fall of stromatolites in shallow marine environments Shanan E. Peters, Jon M. Husson, and Julia Wilcots Department of Geoscience, University of Wisconsin–Madison, Madison, Wisconsin 53706, USA ABSTRACT Stromatolites are abundant in shallow marine sediments deposited before the evolution of animals, but in the modern ocean they are restricted to locations where the activity of animals is limited. Overall decline in the abundance of stromatolites has, therefore, been attributed to the evolution of substrate-modifying metazoans, with Phanerozoic stromatolite resurgences attributed to the aftermaths of mass extinctions. Here we use a comprehensive stratigraphic database, the published literature, and a machine reading system to show that the rock record–normalized occurrence of stromatolites in marine environments in North America exhibits three phases: an initial Paleoproterozoic (ca. 2500 Ma) increase, a sustained interval of dominance during the Proterozoic (2500–800 Ma), and a late Neoproterozoic (700–541 Ma) decline to lower mean prevalence during the Phanerozoic (541–0 Ma). Stro- matolites continued to exhibit large changes in prevalence after the evolution of metazoans, and they transiently achieved Proterozoic-like prevalence during the Paleozoic. The after- maths of major mass extinctions are not well correlated with stromatolite resurgence. Instead, stromatolite occurrence is well predicted by the prevalence of dolomite, a shift in carbonate mineralogy that is sensitive to changes in water-column and pore-water chemistry occurring during continent-scale marine transgressive-regressive cycles. Figure 1. Spatial distribution and number of stromatolite-bearing sedimentary units in INTRODUCTION abundance (Awramik, 1971, 1991; Awramik and Macrostrat (https://macrostrat.org) for North Stromatolites, attached accretionary sedi- Sprinkle, 1999; Walter and Heys, 1985; Semikha- America–Caribbean region. Polygons show approximate spatial boundaries of 1013 geo- mentary structures that are formed either inor- tov and Raaben, 1996; Riding, 2000). However, logic columns used here. Shading indicates the ganically (Lowe, 1994; Grotzinger and Rothman, most previous compilations have focused on total number of stratigraphic units identified 1996) or by microbial interactions between car- specifc time intervals, and none has taken into as containing stromatolites in each column. bonate sediment and the overlying water (Grotz- consideration temporal changes in stromatolite inger and Knoll, 1999; Bosak et al., 2013), are numbers that might be attributable to variation one of the most distinctive of all sedimentary in the quantity of sedimentary rock. defned units, 83% of which are inferred based structures. The stratigraphic distribution and Here we use a comprehensive stratigraphic on their contained biota and/or geographic and abundance of stromatolites has been viewed database covering North America and the cir- stratigraphic position to have been deposited in as a proxy record that can integrate informa- cum-Caribbean region (Fig. 1) to measure the marine and/or marginal marine environments. It tion about the physical, chemical, and biological total quantity and age of shallow marine sedi- is often diffcult to determine whether Precam- environment (Hofmann, 1973; Grotzinger, 1990; ment and the rock record–normalized frequency brian sedimentary units are marine or lacustrine Grotzinger and Knoll, 1999). For example, stro- of occurrence of stromatolites. Our approach is in origin. However, most Phanerozoic sedimen- matolite growth and morphology are responsive agnostic with respect to the genetic interpreta- tary rock is shallow marine (Peters and Husson, to the presence and abundance of grazing meta- tion of stromatolitic structures, their morphol- 2017), so our default assumption is that Precam- zoans (Garrett, 1970; Walter and Heys, 1985) ogy, or their local abundance within individual brian sedimentary rocks are also shallow marine, and possibly to the input of skeletal debris (Pratt, stratigraphic units. Instead, our goal is to mea- unless explicitly identifed otherwise. Each 1982). The evolution and diversifcation of meta- sure the relative frequency of occurrence of sedi- Macrostrat unit is assigned at least one lithol- zoans has, therefore, been identifed as a cause mentary features described as stromatolites in ogy, and most units are associated with multiple for the decline of stromatolites and other micro- the literature in a way that explicitly accounts lithologies. More than 97% of all sedimentary bially formed structures, with disruptions of for variation in the total amount of sedimen- units older than the Quaternary are assigned a animal communities in the aftermaths of major tary rock. We then use the same stratigraphic lithostratigraphic name (e.g., formation) that is mass extinctions promoting their transient rees- framework to evaluate stromatolite occurrence linked to a nomenclatural hierarchy, if applicable. tablishment (Schubert and Bottjer, 1992; Shee- in relation to carbonate mineralogy and marine Here, we also employ a simple age model that han and Harris, 2004; Baud et al., 2007; Mata animal diversity. linearly distributes time between superposed and Bottjer, 2012; but see Riding, 2000, 2005, stratigraphic units using basic geological prin- 2006). Increases in the carbonate saturation state DATA SETS AND METHODS ciples. Constraining the age model is diffcult in of the ocean have also been invoked to explain Stratigraphic data derive from 1013 region- the Precambrian, but average rates of Precam- increases in microbialite abundance (Riding, ally composited Macrostrat database (https:// brian sedimentation derived from it are compa- 2005) and inorganic seafoor carbonate precipi- macrostrat.org) geological columns (Fig. 1) rable to Phanerozoic rates (Husson and Peters, tates (Grotzinger and Knoll, 1995). covering 26 × 106 km2 in North America and 2017). The effects of correlation errors are also Several attempts have been made to com- the Caribbean (Peters, 2006, 2008; Husson and minimized in this analysis because we focus on pile stromatolite occurrences and quantify large- Peters, 2017). Within these columns there are count-based metrics and relative proportions in scale temporal trends in their morphology and 22,282 lithologically and chronostratigraphically aggregate data (Adrain and Westrop, 2000). GEOLOGY, June 2017; v. 45; no. 6; p. 1–4 | Data Repository item 2017155 | doi:10.1130/G38931.1 | Published online XX Month 2017 ©GEOLOGY 2017 Geological | Volume Society 45 | ofNumber America. 6 For| www.gsapubs.org permission to copy, contact [email protected]. 1 Figure 2. Occurrence of stromatolites over geo- logical time normalized by total number of sedimen- tary rock units in study area (Fig. 1). A: Number of stromatolite-bearing marine units normalized by all marine sedimentary rock units. B: Number of stromatolite-bearing units normalized by number of marine carbonate-bear- ing rock units. Estimated proportion ± one stan- dard error of estimate in each 1 m.y. increment is shown. Abbreviations cor- respond to inter national chrono s t ra t i g ra p h i c time intervals: Neoar.— Neoarchean; Neopro- teroz.—Neo protero zoic; Cm—Cambrian; O— Ordovician; S—Silurian; D—Devonian; C—Car- boniferous; P—Permian; Tr—Triassic; J—Jurassic; K—Cretaceous; Pg—Paleogene; Ng—Neogene. Vertical light gray bars identify Phanerozoic marine transgressive-regressive cycles (Meyers and Peters, 2011). Stromatolite prevalence is defned here as Stanford natural language processing (NLP), Whether some of these Archean structures the proportion of all North American–Caribbean which decomposes sentences into parts of formed because of biological activity remains an named marine sedimentary rock units (20,484 speech and linguistic dependencies (Manning open question (Lowe, 1994), but by the Paleo- total) or carbonate-bearing marine rock units et al., 2014). Manual assessment of a random proterozoic, stromatolites occur in ~75% of (9,643 total) that are described as preserv- sample of 3.8% of the machine-extracted stro- all named sedimentary units (Fig. 2A). There ing stromatolites. The GeoDeepDive (https:// matolite–rock unit pairs indicates an accuracy of is a decline in stromatolite occurrence dur- geodeepdive.org) digital library and machine 86%–89% at the individual level (Table DR1). ing the middle Mesoproterozoic to ~40%, but reading system (sensu Peters et al., 2014) was Incorrect pairs identifed during assessment were it increases again to ~80% by the end of the used to locate mentions of the term “stromato- removed from the analysis and the sources of Mesoproterozoic. After achieving a peak at lite” (and variations thereof) within the full text error assessed. In most cases, errors are limited ca. 1000 Ma, stromatolites decline until reach- of published documents. At the time of this anal- to situations involving multiple sentences that ing a Proterozoic minimum at end of the eon. ysis, GeoDeepDive contained most geoscience- express comparative or complex relationships Stromatolite occurrence remains low during relevant titles from Elsevier, Wiley, Canada Sci- between superposed stratigraphic units. Approx- the early Cambrian, but by the Early Ordovi- ence Publishing, and PLoS One, and content imately 30% of the incorrect pairs found during cian stromatolites achieve Neoproterozoic- to from the U.S. Geological Survey, the Society for manual assessment were identifed as correct in Mesoproterozoic-like prevalence
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages89 Page
-
File Size-