spatial variation in the temporal resolution of subtropical ... - BioOne

PALAIOS, 2017, v. 32, 572–583 Research Article DOI: http://dx.doi.org/10.2110/palo.2017.003

SPATIAL VARIATION IN THE TEMPORAL RESOLUTION OF SUBTROPICAL SHALLOW-WATER MOLLUSCAN DEATH ASSEMBLAGES ˜ CARLOS COIMBRA,2 MATIAS DO NASCIMENTO RITTER,1 FERNANDO ERTHAL,2 MATTHEW A. KOSNIK,3 JOAO 4 AND DARRELL S. KAUFMAN 1

Programa de P´os-Gradua¸ca˜o em Geociˆencias, Instituto de Geociˆencias, Universidade Federal do Rio Grande do Sul, Porto Alegre, CEP 91501-970, CP. 15001, Brazil Departamento de Paleontologia e Estratigrafia, Instituto de Geociˆencias, Universidade Federal do Rio Grande do Sul, Porto Alegre, CEP 91501-970, CP. 15001, Brazil 3 Department of Biological Sciences, Macquarie University, New South Wales 2109, Australia 4 Northern Arizona University, School of Earth Sciences and Environmental Sustainability, Flagstaff, Arizona, 86011-4099, USA e-mail: [email protected]

2

ABSTRACT: Fossil assemblages are expected to be time-averaged as a result of biological and physical processes that mix skeletal remains. Our quantitative understanding of time-averaging derives primarily from actualistic studies, in which direct numerical dating of individual specimens is used to assess the scale and structure of age mixing in death assemblages (incipient fossil assemblages). Here we examine the age, and the time-averaging of Mactra shells (Bivalvia: Mollusca) gathered from surface mixed siliciclastic-bioclastic sands at three sites on a passive-margin subtropical shelf (the Southern Brazilian Shelf; ~ 338S). Sixty Mactra specimens were individually dated using amino acid racemization (AAR) calibrated using radiocarbon ages (n ¼ 15). The time-averaging and the total age variability was based on a Bayesian approach that integrates the estimation errors and uncertainties derived from the posterior distribution associated with the AAR calibration average model. The 14C-calibrated AAR ages, pooled across all three sites, are strongly right-skewed with 97% of the individual mollusk shell age estimates ranging from 0 to 6 cal kyr BP. The magnitude of time-averaging varied inversely with the water depth, from , 15 yr at the deepest site (21 m) up to 1020–1250 yr at the shallowest site (7 m). The substantial variation in the temporal resolution across nearby sites, which are located in a seemingly homogenous depositional setting, indicates the presence of notable (if cryptic) spatial heterogeneities in local sedimentation, production, and exhumation, all increasing with water depth.

INTRODUCTION

The temporal resolution of a fossil assemblage is a fundamental parameter for studies that aim to understand the fossil record (Walker and Bambach 1971). Consequently, it is essential to quantify time-averaging at different spatial scales, across multiple sedimentary environments, and across a range of latitudes. Furthermore, expanding these approaches to multiple taxa will be critical for better understanding the fossil record formation, as time-averaging determines the scale and precision of paleoecological and evolutionary studies (Kowalewski 1996; Hunt 2004; Tomaˇsov´ych and Kidwell 2010; Kidwell 2013). Increased availability of cost-effective dating techniques has led to an increasing number of studies quantifying time-averaging in order to understand the temporal resolution of the fossil record. Most of them are based on surficial shelly death assemblages (e.g., Kidwell et al. 2005; Kosnik et al. 2007, 2009, 2013, 2015; Krause et al. 2010; Dexter et al. 2014; Tomaˇsov´ych et al. 2014, 2016a, 2016b; Olszewski and Kaufman 2015; Dominguez et al. 2016). Additionally, Scarponi et al. (2013) provide an example of a Holocene transgressive-regressive stratigraphic succession based on shell ages. The magnitude of empirical surveys of time-averaging fluctuate from a few years (Olszewski and Kaufman 2015) to some decades (Scarponi et al. 2013) or even at scales of hundreds of years to a few thousand or tens of thousands of years, though the age-distributions are generally right-skewed (see review in Kidwell 2013). Many time-averaging studies have focused on marine continental shelves with low sedimentation rates during the Quaternary, as the Sa˜o Paulo Bight (Krause et al. 2010; Dexter et al. 2014) and the southern California continental shelf

(Tomaˇsov´ych et al. 2014, 2016a). A multitude of studies were also conducted in carbonate domains (Kidwell et al. 2005; Kosnik et al. 2009, 2013, 2015; Albano et al. 2016). In parallel with the empirical studies, models of fossil deposit formation have become increasingly sophisticated (e.g., Olszewski 2004, 2012; Tomaˇsov´ych et al. 2014; Terry and Novak 2015). The development of a fossil assemblage is related to the balance between biological production, rates of fossil destruction, sedimentation, and the degree and depth of sediment mixing. In the case of surface samples, for example, the net difference between production and destruction in the taphonomically active zone (TAZ) determines what survives into the burial zone (Davies et al. 1989; Tomaˇsov´ych et al. 2006; Berkeley et al. 2014). Moreover, the presence of old shells does not necessarily indicate long residence times near or at the sediment surface solely, but can also imply long residence time in the mixed zone or a negative sedimentation regime when deposits buried for a long time are exposed by net erosion. Therefore, quantifying the age distribution of fossil assemblages is essential to understanding the rate of destruction, recycling, and preservation of biological remains over time for marine invertebrates (Tomaˇsov´ych et al. 2014, 2016a; Kosnik et al. 2015; Olszewski and Kaufman 2015) as well as terrestrial vertebrates (Terry and Novak 2015). This study provides the first quantitative estimate of time-averaging on the southernmost Brazilian shelf (~ 338S) based on 60 Mactra shells gathered from surface samples collected between seven and 21 m water depth. The southernmost Brazilian shelf is presently a sediment-starved shelf containing many shell-rich patches displaying a depth-related taphonomic profile (Erthal et al. 2014). This investigation contributes:

Published Online: September 2017 Copyright Ó 2017, SEPM (Society for Sedimentary Geology) 0883-1351/17/032-572/$03.00

PA L A I O S

SPATIAL VARIATION IN TIME-AVERAGING

573

(1) a window into time-averaging in a low-sedimentation-rate shelf environment; (2) a time-averaging estimate from the subtropical Atlantic; and (3) a chronological context supporting upcoming studies on paleontological and ecological approaches at subtropical shelves (Ritter et al. 2016). Similarly, an estimate of the time-averaging of molluscan assemblages on shelves with relict sedimentation and high biological productivity can bring new insights into this bias in the fossil record from those environments (Halfar et al. 2004; Rivers et al. 2007; James and Bone 2011). Furthermore, the use of amino acid racemization in warm-temperate areas in the Southern Hemisphere may add to our understanding of timeaveraging across various environments and on a global scale. MATERIAL AND METHODS

Study Area The Southern Brazilian shelf (SBS, Fig. 1) is, presently, a sedimentstarved continental shelf. Deposition of a significant amount of post-rift, primarily clastic sediment produced a wide (100–300 km range, 125 km average), shallow (100–140 m) continental shelf, characterized by a smooth morphology (Corrˆea 1996). Sandy and bioclastic (mainly formed by molluscan) sediments dominate the shelf in shallow areas (up to 60 m), whereas muddy sediments prevail in deeper water (Corrˆea 1996). Southward, the SBS shifts from a relatively narrow and homogeneous (100–170 km) to a wider and heterogeneous topography (100–200 km) (Zembruski 1979). In the southern area, the usually flat shelf morphology is interrupted by sand bodies, sand waves and elongated bioclastic deposits, interpreted as ancient shorelines (Corrˆea 1996). The bioclastic deposits are abundant between isobaths of 10 and 50 m and developed over rippled bottom topographies, where mega-ripples are present (Corrˆea and Ponzi 1978). The geological evolution of the southern Brazilian coastal plain and continental shelf during the Quaternary was actively controlled by the sealevel variations (Villwock et al. 1986; Villwock and Tomazelli 1995; Corrˆea 1996; Martins et al. 1996; Tomazelli and Villwock 2000; Dillenburg et al. 2004). The present physiography of the coastal plain and shelf is mainly the result of Quaternary high-frequency (4th to 5th order), glacioeustatic sea-level changes, which affected the sedimentary systems along the coast. The post-glacial sea-level history of this area extends back circa 15.5 kyr, when the sea level was 120–130 m below present (Corrˆea 1996). After that time, the sea level rose at an average rate of 1.2 cm/yr, beginning immediately after the Last Glacial Maximum (LGM). Sea level stabilized at 9 ka (between depths of 32 and 45 m) and 8 ka (between depths of 20 and 25 m) (Corrˆea 1996). Reliable data on the sea-level fluctuations during middle to late Holocene are unavailable for the studied area, but sea-level curves for areas further north indicate that during the culmination of the post-glacial marine transgression (which occurred between 4.1 and 2.7 kyr ago), the sea level was 1–3 m (perhaps up to 4 m) above its present level (Tomazelli et al. 1996; Angulo et al. 2006). Correlation between the sea-level highstands and major peaks of the oxygen isotope curve was established by Villwock and Tomazelli (1995), which should confirm the correlation between sea-level oscillations and climate changes. However, there is no agreement on the accurate ages of the current sea-level rise and stabilization in the SBS (Dillenburg and Barboza 2014; Nagai et al. 2014). The SBS is a subtropical margin, where oceanographic features are dominated by the confluence of two main current systems (Nagai et al. 2014). The outer shelf is dominated by the southward flow of the Brazil Current (BC) (Mahiques et al. 2004) which brings the relatively warm (. 22.78C) and salty (salinity . 36.76) tropical water. Two fresher water masses (salinity ranging to 26 up to 36) may be distinguished on the inner shelf: the Subtropical Shelf Water and the Plata Plume Water. These water masses present a clear seasonal temperature pattern, with values ranging

FIG. 1.—Study area and sample sites on the inner Southernmost Brazilian Shelf. Sedimentary features were based on CPRM (2009).

from 10–218C during winter and 15–268C during summer (M¨oller et al. 2008). Also, during wintertime, these water masses are transported northwards along the inner shelf by the Brazilian Coastal Current (BCC) (Souza and Robinson 2004). The annual mean sea surface temperature in the sampled area is roughly 178C (Locarnini et al. 2013). Sample Collection The samples of surface sediment were gathered from three sites along the southernmost area on the SBS (Fig. 1, Table 1). Sites 2 (19 m) and 3 (21 m) were sampled by box-corer (topmost ~ 10 cm of the stratigraphic succession) in 1974 by the Brazilian Navy in cooperation with several universities (GEOMAR VII expedition; Martins 1987), while shells from site 1 (7 m) were recently (2013) collected by dredge, and then recovered from the topmost 10 cm of the sample resulting. These sites cover the southern bioclastic province, a shell-rich domain on inner shelf at the SBS (Erthal et al. 2014), where the relictual sands (Pleistocene), as well as biological remains, are dominant (Figueiredo 1975; Corrˆea and Ponzi 1978; Corrˆea 1983). Dated Specimens Mactra Linnaeus, 1767 is a widespread marine bivalve genus, a relatively thin-shelled infaunal filter/suspension feeder with a varied size range (Fig. 2). Mactra was selected for this investigation because it is abundant throughout the study area, where it has a wide bathymetric range of 11 to 61 m (Cap´ıtoli and Benvenuti 2004). Three species of Mactra are known to occur on the SBS, Mactra isabelleana d’Orbigny, 1846, Mactra petitii d’Orbigny, 1846, and Mactra janeiroensis E.A. Smith, 1915 (Absala˜o 1991; Cap´ıtoli and Benvenuti 2004). As the identification to species level can be problematic from empty shells, we limit our identification to the genus level. However, all the shells dated in this study are morphologically identical, and most likely belong to the species Mactra isabelleana. The length of dated specimens ranged

574

PA L A I O S

M.N. RITTER ET AL.

TABLE 1.—14C ages used to calibrate D/L values in the Bayesian models. Note that four valves at 0 cal yr BP were removed from the calibration, leaving a total of 15 independently dated valves. UAL ¼ Northern Arizona University. UAL number 10986 10990 10992 10993 10994 10998 11002 11004 11005 11006 11007 11011 11013 13471 13472 13480 13485 13486 13487

Dating method Carbonate Carbonate Graphite Carbonate Graphite Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate Carbonate

target AMS target AMS target AMS target target target target target target target target target target target target target target

AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS AMS

Depth (m)

Site

Fraction modern

7 7 7 7 7 7 7 19 19 19 19 19 19 21 21 21 21 21 21

1 1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3

0.93167 0.01699 0.00724 1.05666 0.78480 0.50046 0.50046 0.81185 0.86787 0.73124 0.60205 0.76036 0.83549 0.94770 0.95040 0.95370 0.95200 0.93890 0.93930

from 7 to 25 mm with a mean and median size of 11 mm. The average length of shells from site 1 was 13.35 mm, while it was 9.88 mm and 12.45 mm in the sites 2 and 3, respectively.

14

C age

Calibrated age (2 sigma)

6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

177 36521 43007 1306 1475 3087 5934 1215 681 2161 4090 1776 980 0 0 0 0 138 124

570 32700 39590 1170 1945 3280 5560 1675 1140 2515 4075 2200 1445 430 410 380 400 510 500

30 2000 590 40 15 35 50 30 30 45 40 30 30 60 70 60 80 80 60

408 6 18 yr (DR 8 6 17 yr) was assumed, as established by Angulo et al. (2005). All ages were calibrated relative to AD 1950, and four valves with ages ranging from AD 1950 and younger were considered to be modern in this study (AD 1950 ¼ 0 cal yr BP) (Table 1).

Amino Acid Racemization Amino acid racemization (AAR) dating is based on the interconversion between the two arrangements of amino acids: left (L) and right (D) handed (see Demarchi and Collins (2015), and Kaufman (2015) for a recent review of the method and its applications). Living bivalves exclusively use left-handed amino acids to construct proteins. After death, the organism’s protein degrades into their constituent amino acids, which spontaneously interconvert to their right-handed configurations (racemization) at a rate that is mainly a function of taxa and temperature. Therefore, the extent of racemization (D/L value) relates primarily to the length of time since death and the post-mortem thermal environment. The time since death can, therefore, be determined by relating the extent of AAR (D/L value) to calendar years using a subset of known-age specimens to obtain numerical ages for the remaining specimens. Single samples from 60 Mactra right valves were prepared for amino acid analysis at Northern Arizona University according to the conventional procedure (Wehmiller and Miller 2000) (Fig. 3). The chromatographic instrumentation and procedure used to separate amino acid enantiomers were described by Kaufman and Manley (1998). We focused on eight amino acids: aspartic (Asp), glutamic (Glu), serine (Ser), alanine (Ala), valine (Val), phenylalanine (Phe), isoleucine (Ile), and leucine (Leu) that are abundant in molluscan-shell protein, and well resolvable by reverse-phase HPLC. D/L values for all eight amino acids are reported in the online Supplementary Data File 1. Radiocarbon Dating Of the 60 Mactra shells analyzed by AAR, 19 were sent for 14C dating (Table 1). All valves were dated by the Keck Carbon Cycle Accelerator Mass Spectrometry Facility at the University of California-Irvine (UCIKCCAMS) using powdered carbonate (n ¼ 17) (Bush et al. 2013) or graphite (n ¼ 2) targets (Table 1). Radiocarbon ages were calibrated to calendar years with CALIB version 7.1 (Stuiver et al. 2005) using the database marine13 (Reimer et al. 2013). A mean marine reservoir age of

Taphonomic Analyses All shells dated herein (n ¼ 60) were assigned damage states using a categorical scoring system developed and adapted from previous work (Zuschin et al. 2003; Best 2008; Ritter et al. 2013 and references therein) (Table 2). For taphonomic analysis, specimens were examined under low (10–203) magnification using a stereoscopic microscope, and coded for variables describing encrustation, bioerosion, fragmentation, fine-scale surface alteration, and color alteration (summarized in Table 2). Many terms for fine-scale surface alteration have been applied, but most are descriptive and do not allude to a particular process (Best 2008 and references therein). In our protocol, therefore, fine-scale alteration refers to various degrees of degradation of original luster; it may be due to any combination of microboring and other microbioerosion, partial dissolution of mineral crystallites, maceration of shell organic matrix, and physical abrasion, which require a scanning electron microscope (SEM) to distinguish (Best 2008). Prior to analysis, the taphonomic profile (scoring damage for each signature on each shell; online Supplementary Data File 2) was standardized, dividing the weighted mean of the taphonomic damage by the maximum value scored for each shell. Thus, the taphonomic profile for each shell will be ranged between 0 (pristine) and 1 (maximum taphonomic alteration). The shells are also classified hierarchically according to their taphonomic damage: pristine (shells those have a taphonomic damage  0.3), moderate alteration (0.31 to 0.69), and high (7.0 to 1.0). In order to test for differences in the standardized taphonomic damage among sites, we performed the Kruskal-Wallis Test. Statistical Analyses AAR age models were fit using the Bayesian model fitting procedures described by Allen et al. (2013). This procedure entailed first fitting a series of models to the paired age-D/L data for 15 Mactra with 14C age estimates. Then, an overall calibration model was constructed by weighting

PA L A I O S

SPATIAL VARIATION IN TIME-AVERAGING

575

FIG. 3.—Aspartic acid and glutamic acid D/L ratios for each shell analyzed in this study binned by collection site (n ¼ 60).

represented as Bayesian posterior distributions, which represent probability distributions for each model parameter conditional on the available evidence (in this case paired D/L values and 14C age estimates for a subset of shells). For this analysis, the posterior distribution for each model parameter is approximated by a set of values (10,000 values, corresponding to 10,000 Markov-Chain Monte-Carlo (MCMC) trials: see Gelman et al. 2014 for a discussion of MCMC). These posterior distributions for the age TABLE 2.—Taphonomic protocol utilized in this study. Taphonomic attribute Fragmentation FIG. 2.—A right valve of Mactra. The photographed specimen matches the dated shell UAL 10986 (see online Supplementary Data File 1). Scale bar ¼ 10 mm.

the relative contributions of the models of best fit, as characterized using the Bayesian Information Criterion (BIC). In total, 84 different age models were fit using the statistical computing program ‘‘R’’ (R Core Team 2016). Together, these 84 models encompass all eight potentially age-informative amino acids (Asp, Glu, Ser, Ala, Val, Phe, Ile, Leu), three different mathematical functions used to describe average age-DL relationship [time-dependent rate kinetics (TDK), constrained power-law kinetics (CPK) and simple power-law kinetics (SPK)] and two uncertainty distributions (gamma and lognormal) used to describe variability about these relationships (Allen et al. 2013). For each unique combination of amino acid, mathematical function, and uncertainty distribution, two models were fit, one with the initial D/L value (the D/L ratio at the time of death) set at R0 ” 0, and one with R0 treated as a fitted parameter (0 , R0 , min(D/L)). All models are plotted and summarized in the electronic supplementary material (online Supplementary Data File 3). The Bayesian fitting procedure of Allen et al. (2013) yields not only an overall best-fit age model, but it also yields associated error estimates for each of the fitted model parameters, and it propagates that model error to the D/L-inferred estimates of age, thereby yielding error estimates for each age estimate. These error estimates for the model, and its predictions are

Fine-scale surface alteration (FSA)

FSA extent

Bioerosion and predation

Encrustation

Secondary color (or color alteration)**

Scores 0 ¼ whole valves articulated 1 ¼ 1 whole valve 2 ¼ large fragment (80% of shell) 3 ¼ small fragment (,20% of shell) 0 ¼ pristine 1 ¼ present 1.1 ¼ small pits 1.2 ¼ large pits 1.3 ¼ small and large pits 1.4 ¼ holes 1.5 ¼ small pits and holes 1.6 ¼ large pits and holes 1.7 ¼ small and large pits and holes 2 ¼ Corrasion extent 2.1 ¼ ,30% of corrasion 2.2 ¼ 30 to 60% of corrasion 2.3 ¼ .60% of the shell surface with corrasion 0 ¼ absent 1 ¼ present 1.1 ¼ drill 1.2 ¼ sponge 1.3 ¼ worn 1.4 ¼ bryozoan 1.5 ¼ unidentified 0 ¼ absent 1 ¼ present 1.1 ¼ serpulid 1.2 ¼ bivalve 1.3 ¼ barnacle 1.4 ¼ forams 1.5 ¼ Sponge 1.6 ¼ Algae 0 ¼ none 1 ¼ oxidized color 2 ¼ reduce color

More information Zuschin et al. (2003)

Ritter et al. (2013)

Best (2008)

Best (2008)

Best (2008)

Best (2008)

** oxidized color (cream, yellow, ocher and red); reduce color (white, gray and black)

576

PA L A I O S

M.N. RITTER ET AL.

TABLE 3.—Bayesian model averaging summary of the gamma distribution, whereas ln(a), ln(b), c, R0, ln(d) are parameters of the various models; k is the number of model parameters; Bayesian Information Criterion (BIC) is a measure of goodness of fit; BIC weight (wBIC) is the contribution of the model to the best fit model; and model refers to the model numbers in Figure 4 and online Supplementary Data File 3. Model 2 3 5 6 4

Amino acid

Distribution

Function

ln(a)

ln(b)

c

ln(d)

k

BIC

P value

R2

R0

ln(R0)

wBIC

Asp Asp Asp Asp Asp

gamma gamma gamma gamma gamma

TDK0 SPK0 TDK1 SPK1 CPK1

11.503 12.200 11.537 12.195 6.928

1.103 1.246 1.033 1.243 0.923

NA NA -0.871 -4.461 1.262

5.199 5.122 5.186 5.142 5.181

3 3 4 4 4

238.3 238.5 240.8 241.3 241.5

,0.001 ,0.001 ,0.001 ,0.001 ,0.001

0.943 0.943 0.943 0.943 0.943

0 0 1 1 1

NA NA -3.925 -5.467 -16.227

0.38 0.34 0.08 0.11 0.09

estimates can be combined for subsequent analyses to evaluate overall differences in age for collections of shells taken from different collection sites, and to estimate the extent of time-averaging while accounting for the error associated with the age estimates. In this study, we used the Bayesian posterior distributions of the age estimates, calculated using standard MCMC procedures as described in Allen et al. (2013), to test for age differences between collections of shells taken from different sites. For each pair of collections, we determined what fraction of the age estimates representing the first posterior distribution had values less than the median age estimate for the second posterior distribution (see Dominguez et al. 2016). Comparisons where . 95% or , 5% of the age estimates from the first posterior distribution were less than the median value of the second posterior distribution were judged to be distinct with respect to overall age, whereas collections with between 25 and 75% overlap of the first posterior distribution with the median of the second posterior distribution were judged to contain shells created during the same time period. R scripts (vers. 1371, updated on July 2016) implementing these tests are provided in the supplementary material (online Supplementary Data File 4). In this study, we combined the posterior distributions of the age estimates of all specimens within a collection to calculate the total ageestimate variability (the same terminology and procedure as used by Dominguez et al. 2016). Importantly, this total age-estimate variability incorporates both model-based uncertainties for individual age estimates (age-estimation error) and the mixing of shells of different ages (timeaveraging) effects on the collection’s age distribution. As a first step towards disentangling these two distinct sources of age-estimate variability, we subtracted each of three summary statistics summarizing the interquartile ranges of the individual age estimates (mean, median, and 95% confidence interval as measures of age-estimation error), from the interquartile range of the total age-estimate variability to determine the age variation in the total age-estimate variability that is due to timeaveraging (R scripts provided in online Supplementary Data File 4). We use the term ‘residual time-averaging’ to distinguish these time-averaging indices from previous time-averaging values which do not explicitly account for age-estimation error. While we cannot yet assess the formal

statistical significance of these ‘residual time-averaging’ indices, they provide useful metrics for evaluating time-averaging effects because, while the posterior distributions of the individual age estimates solely reflect errors in age determination, the combined posterior distributions for collections of specimens incorporate this source of error as well as time-averaging effects. We view these metrics as a useful but preliminary step in statistically quantifying the extent of time-averaging in fossil assemblages. RESULTS

Age Determination We fitted 84 age models using three mathematical functions (TDK, CPK, and SPK) and two uncertainty distributions (gamma and lognormal). While both uncertainty distributions produced consistent results, no lognormal model has a BIC value within four units of the best model. In addition, assuming lognormally distributed uncertainty led to confidence intervals for young shells that were smaller than the radiocarbon ages used to create the calibration model. Assuming gamma distributed uncertainty resulted in confidence intervals that were more in keeping with the uncertainty of the radiocarbon ages used to fit the calibration models. The gamma distribution was considered to be the superior uncertainty distribution for these data, and no lognormal model has a BIC value within four units of the best model (online Supplementary Data File 5, Fig. 4). We evaluated two methods of estimating the initial D/L ratio (R0). Both approaches yielded similar inferred ages and inferred uncertainty. The average fit R0 value was 0.007 (range was 0.00 to 0.019) (Table 3), which was comparable to previous studies (Kosnik and Kaufman 2008). As we found no reason to prefer one method over the other, so both fit R0 and R0 ” 0 models are included in the final averaged model. Models with R0 ” 0 contributed 72% of the final averaged model whereas models with a R0 fit from the data contributed 28% of the final averaged model (Table 3).

TABLE 4.—Summary of MCMC replicates for all the age estimates by site. Quantiles of the ‘‘total age-estimate variability’’, the interquartile range of the total age-estimate variability (oIQR). The total age-estimate variability includes both time-averaging and age-estimation error. Mean, median, and 95% ‘‘age estimation error’’ determined using the interquartile ranges of the individual shell’s age distribution in each site. ‘‘Residual time-averaging’’ for each site using the median, mean, and 95% age-estimation error. Distributions outside of 5–95% are considered to sample distinct time periods whereas collections within 25–75% are deemed to sample overlapping time periods. Age-estimate variability (yrs.) Site Site 1 Site 2 Site 3

Age-estimation error (yrs.)

Time-averaging (yrs.)

Depth (m)

0%

25%

50%

75%

100%

oIQR

mean

median

95%

mean

median

95%

7 19 21

70 93 2

1892 912 146

2658 1318 230

3806 1833 346

56231 7085 3095

1914 921 200

896 472 189

663 431 186

2999 590 216

1018 449 11

1251 490 14

0 331 0

PA L A I O S

SPATIAL VARIATION IN TIME-AVERAGING

577

FIG. 4.—The four best Asp-models and their respective weights on the average gamma model. The dark shading corresponds to the confidence interval for age, and light shading corresponds to the prediction interval for age (numerical details in Table 3). The squares indicate the age determination by 14C (see also Table 1).

While eight amino acid D/L values (Asp, Glu, Ser, Ala, Val, Phe, Ile, Leu) were fitted, only five models using Asp contributed to the final averaged model (Table 3). All three of the mathematical functions fit [timedependent rate kinetics (TDK), constrained power-law kinetics (CPK), simple power-law kinetics (SPK)] contributed more than 7% to the final averaged model (online Supplementary Data File 5). TDK contributed the most (46%) to the final averaged model followed by SPK (45%) and CPK (9%).

Age Distributions While the total age-estimate variability of Mactra shell-rich assemblages was strongly right skewed, with the full range of age estimates spanning the modern to 56,000 cal yr BP, 97% of the age estimates ranged between modern and 6000 yr. The three sites have distinct age distributions (Table 4, Fig. 5, online Supplementary Data File 6). The total age-estimate variability from site 1 is the broadest with a median age of ~ 2660 yr (IQR:

578

M.N. RITTER ET AL.

PA L A I O S

FIG. 5.—The total age-estimate variability by site. A) All ages. B) The spindle is truncated at 8 ka. Spindles are histograms of the total age-estimate variability; each spindle has the same area, and spindle height is proportional to the frequency of that age. The thick vertical line expresses the median age estimate. The dark gray encompasses 50% of the age estimates; the medium gray includes the 95% of the age estimates, and the lightest gray encompasses 100% of the age estimates. These spindles include both uncertainties in the age estimates for each shell and timeaveraging.

1890–3810 yr). Mactra from site 2 were intermediate with a median age of ~ 1320 yr (IQR: 912–1830 yr), whereas Mactra from site 3 were the youngest with a median age of ~ 230 yr (IQR: 146–346 yr). Residual time-averaging as defined as the variability in ages beyond that expected due to uncertainty in the age model also varies between sites: site 1 has the highest residual time-averaging (1020–1250 yr), followed by site 2 (470– 430 yr), and site 3 (, 15 yr). While not formally testable, we consider these age distributions to be distinct. Taphonomic Damage The taphonomic damage of shells was similar among sites, with no significant differences (Kruskal-Wallis Test, X2 ¼ 3.43, p ¼ 0.179) (Fig. 6A). Site 1 exhibited less taphonomic damage (0.32) compared to sites 2 and 3, where shells had presented the same variation (0.78) (Fig. 6A). Pristine shells were found only at site 3, while the most altered shells were found at the site 2. No clear taphonomic pattern arises when comparing individual taphonomic scores. For example, fragmentation was relatively higher at the sites 2 and 3 than site 1, where one-third of the shells are unfragmented. However, the color alteration and the FSA extension were more frequent in the shells from the shallowest site. The relative taphonomic damage (pristine, moderated, and high alteration) is not a linear predictor of shell ages. Younger shells include both pristine or highly individuals (Fig. 6B). However, shells with moderate taphonomic alteration have a mean best-fit age estimated ~ 10-fold higher than the mean-age of pristine shells. Also, the mean-age of shells displaying the highest taphonomic alteration was roughly four-fold younger than those shells characterized by moderate damage. DISCUSSION

Spatial Variation of Time-Averaging Bioclastic accumulations vary consistently in thickness and shelly content throughout Cenozoic marine deposits (Kidwell and Brenchley

FIG. 6.—Taphonomic state (damage index) of dated shells from SBS and its relation to estimated ages. A) Box plots (showing interquartile range, 95% confidence intervals, and outliers) comparing taphonomic state from sites 1, 2 and 3. Dashed lines indicate categories of taphonomic alteration: pristine (lower interval), moderate (intermediate interval) and high damage (upper interval). Both pristine and highly altered shells exhibit the youngest ages. Also, shells of the oldest ages are most abundant in the moderate alteration category. B) Box plots comparing categories of taphonomic alteration against age, showing no clear pattern. The box plot (B) y-axis was truncated at 10 ka to highlight the differences among taphonomic categories, shells older than 40 ka also exhibit moderate damage.

1994). This pattern plays a major role, but unknown in extent, in the distribution of directly dated organic remains. For example, a high environmental contrast in sedimentation rate appears to influence spatial variation in the time-averaging of Fulvia shells (Bivalvia: Mollusca) from Sydney Harbour (Dominguez et al. 2016). Sydney Harbour is dominantly siliciclastic, similar to the sites herein studied, but the shell density is much lower, likely due to relatively higher (~ 2 m) sediment accumulation over the last 6 kyr. In the SBS, the thickness of the shell layers may reach 2.5 m

PA L A I O S

SPATIAL VARIATION IN TIME-AVERAGING

579

FIG. 7.—Histograms based on MCMC replicates of all ages estimates by site and its relation to the global sea-level curve since before the Last Glacial Maximum (LGM) (data compiled from Lambeck et al. 2014). Included in the plot are a few relictual shells older than 40 kyrs, when the sea level position was ~ 60 m below modern position. However, most of the shells belong to the Holocene period, associated with the post-LGM transgression.

(Calliari et al. 1999). Human disturbance on sedimentary processes (bottom dredging, jetties, and seawalls) may be responsible for two-time averaging estimates . than 1,000 yr (Dominguez et al. 2016). Dominguez et al. 2016 suggested that the age-distributions relate primarily to sediment loss (omission) and condensation of the fossil assemblage, a situation very similar to a sediment-starved shelf. On the other hand, the median assemblage age and residual time-averaging at the deepest SBS site (~ 230 yr and , 15 yr, respectively) show similar values to those found at four of the Sydney Harbour sites where sedimentary processes were considered to be least influenced by human activities (~ 150 yr and , 40 yr, respectively). However, the median assemblage age and residual timeaveraging at the shallowest site of SBS (~ 2.6 kyr and ~ 1.1 kyr, respectively) are very similar to the Sydney Harbour site close to the dredged channel (~ 2.4 kyr and ~ 1.9 kyr). Also, the samples from Sydney Harbour were all collected from ~ 10 m water depth, with time-averaging ranging from ~ 40 yr at four sites to . 1900 yr at another two sites (Dominguez et al. 2016). Remarkably, the extent of time-averaging in subtropical areas such as the SBS appears to be more widespread in age range (modern to 56,000 yr) when compared to studies in northward portions of the southeastern Brazilian shelf. Dexter et al. (2014) estimated time-averaging of bivalverich assemblages also using AAR calibrated against radiocarbon ages. They found ages ranging from modern to 10307 cal yr BP in surficial bulk samples from three depths in a shallow marine embayment of the Sa˜o Paulo Bight (latitude around 238 to 258S). In the same study area, Krause et al. (2010) found ages ranging from modern to 8118 cal yr BP for brachiopods, and modern to 4437 cal yr BP for bivalves. The magnitude of time-averaging (and median age) varied inversely with the water depth in the SBS (Figs. 5–7, Table 4). Dexter et al. (2014) have also analyzed the time-averaging variation across three depth ranges and found that age distributions are similar along the bathymetric gradient, despite site-to-site stochastic variation. As occurs in the SBS, the Sa˜o Paulo Bight has little

sediment input, and reworking of late Quaternary sediments also prevails (Mahiques et al. 2011). This inverse relation between age and water depth among sites contrasts with the sea-level history on SBS (Fig. 7) (but see Angulo et al. 2006 for a critical review of mid- to late-Holocene sea-level fluctuations). The shell-enriched deposits are interpreted as having been formed during periods of stabilization in the sea level during the Holocene transgressive event (Corrˆea 1996; Martins et al. 1996). However, combined spatial gradients in production, sedimentation, and exhumation, all increasing with depth up to 21 m, have biased the temporal accuracy of these shell-rich assemblages. The shallowest site on SBS showed the highest time-averaging, displaying the oldest shell ages estimate, back to 56 ka, when the relative sea level was roughly 50–60 m below modern position (Fig. 7). Considering that Mactra has a depth range down to roughly 60 m, we infer that the Mactra’s death assemblages were formed in its preferential water depth prior to the LGM. As the present-day base-wave level is approximately 40 m on SBS (Cecilio 2015), some within-habitat reworking could be expected in the modern SBS shoreface. Additionally, the shoreface width and offshore mid-shelf declivities generate variability in the offshore wave power and along shelf currents (Cecilio 2015). Likewise, the shallow sandy bioclastic deposits on SBS that interrupt the smooth morphology of the shoreface provide the source for some withinhabitat transport across shallow depths. Because living Mactra is more abundant between 20 and 30 m on SBS (Cap´ıtoli and Bemvenuti 2004), younger shells in shallower water were probably transported shoreward (timeaveraged within-habitat assemblages), while younger shells at the deepest site (site 3, 21 m) are more consistent with the observed Mactra habitat. The patches of shell-rich accumulations on SBS are several kilometers long and a few meters wide (Figueiredo 1975). Our samples are from these shell-rich accumulations, which mostly formed during the Holocene transgressive event. However, there are also unexpected shells, although rare, older than 40 ka. Some previous workers have also reported rare

580

M.N. RITTER ET AL.

PA L A I O S

shells older than the LGM on SBS sea floor (e.g., Figueiredo 1975; Kowsmann et al. 1977), but could not explain their presence. Pleistocene sediments are expected to crop out on the shoreface, and inner continental shelf because (1) the present-day clastic sedimentation rate is negligible when compared to Pleistocene rates (Vicalvi 1977), and (2) post-glacial sea-level rise probably eroded at least 10 m of the sedimentary sequence (Dillenburg 1994). The presence of beach rocks and coquinas with Pleistocene mammalian fossils (in 0 to 20 m of water depth mainly, Figueiredo 1975; Lopes and Buchmann 2011) reinforces this hypothesis. The old shells may have been eroded from these coquinas, and then mixed with numerically dominant younger shells forming the shell-rich patches during the Holocene transgression. Taphonomic Bias on Time-Averaged Assemblages The spatial variation in time-averaging among sites is not reflected in the spatial variability in the taphonomic patterns (Fig. 6). Both pristine and highly altered shells are younger than those with moderate alteration, which are the oldest. The lack of relationship between age and preservation implies that shells did not spend all their time in TAZ. Shell destruction and reworking tend to decrease with depth in the sediment column (Olszewski 2004). Some shells can move deeper into the sediment column during physical or biological reworking events, which would decrease their rate of destruction and/or promote their diagenetic stabilization. Such shells can be later reworked back to the TAZ where they are mixed with the younger shells. Thus, the presence of old shells can be explained by their protracted presence at sediment depths where they are relatively safe from destructive processes. Under active production, surface-age distributions are characterized by a mixture of very young shells that experience rapid loss at the sediment surface but also by shells exhumed from the sequestration zone and older than expected under harsh destructive regimes occurring in TAZ, forming a strongly right-skewed age frequency distribution, with a tail that is fatter than expected under a simple exponential loss of shells (Olszewski 1999; Olszewski and Kaufman 2015). However, as highlighted by Olszewski and Kaufman (2015), the time expected to reach the sequestration zone is also a function of biological and physical reworking (sequestration rate in Tomaˇsov´ych et al. 2014) and is thus not fully comparable to the sedimentation rate. The presence of very old shells implies the reworking of previously sequestered shells on the SBS. Temporal Completeness of the Death Assemblages

FIG. 8.—Age distribution of the 60 dated specimens of Mactra. Specimens were grouped using two different bin sizes. A) 250-year intervals. B) 500-year intervals.

The temporal completeness of a death assemblage measures the extent to which a collection of dated shells in aggregate provides a continuous time series (Kowalewski et al. 1998). To estimate the completeness including the age uncertainty, we took random draws from the MCMC (10,000 replicates) of each shell age (n ¼ 60), then binned the ages into 250 and 500 year intervals, repeating 1000 times and calculating interquartile ranges of the proportion of bins represented by at least one shell (i.e., median and 95% confidence interval of completeness). Based on this approach, our dated assemblage from SBS is 43% (38–47%) complete using 250 yr bins and 54% (50–59%) using 500 yr bins (Fig. 8). Because completeness is sensitive to the number of shells samples, we recalculated the completeness of shells in the larger dataset of Dexter et al. (2014) by randomly resampling that dataset to n ¼ 60 and including the age uncertainty derived from the reanalysis of their data and applying the same methods applied here (i.e., Bayesian calibration derived from Allen et al. 2013). Using 250 yr bins, the median completeness for the Holocene shells in the Dexter et al. collection is 34% (30–38%) and 45% (40–54%) for 500 yr bins. These results show that completeness is similar at sites separated by ~ 1000 km in the shallow water of the southern Brazilian shelf (24– 348S).

PA L A I O S

SPATIAL VARIATION IN TIME-AVERAGING CONCLUSIONS

Sixty individually dated Mactra shells from the inner shelf of the SBS yielded ages ranging from modern up to ~ 56,000 years, displaying a strongly right-skewed age-frequency distribution. Although 97% shell age estimates were younger than 6 kyr, a few age estimates were older than 40 kyr and are probably palimpsests. These rare old shells were probably reworked from older sediment during the Holocene transgression. The temporal resolution of death assemblages on SBS is spatially variable. The magnitude of time-averaging varied inversely with the water depth: , 15 yr at the deepest site (21 m) compared to 1020–1250 yr at the shallowest site (7 m). Combined spatial gradients in production, sedimentation, and exhumation, all increasing with depth up to 21 m certainly are important parts of the explanation for that pattern. Shoreward transport would likely include some young shells to the shallowest sites. These findings point to the importance of expanding time-averaging data to a wide range of sedimentary environments, and they contribute to a growing number of datasets that will eventually lead to a broader understanding to time-averaging in modern sedimentary environments. Also, this substantial temporal span of dated specimens points to their potential value as archives of the paleoenvironmental history of the Southern Brazilian shelf over the most recent millennia. Nevertheless, more integrative studies are still needed to provide more geological and sedimentological background regarding this matter. ACKNOWLEDGMENTS

This study was supported mainly by the FAPERGS (grant 1982-2551/13-7). Additional funds were covered by the CNPq (140568/2014-0 to MNR, and 140927/2008-5 to FE), and by the International Ocean Discovery Program (CAPES 0195/2016-02-BEX to MNR). We also thankful Felipe Caron, who shared the samples from site 1, as well as Iran C.S. Corrˆea, Maribel S. Nunes, and Jair Weschenfelder who allowed access to the samples from the sites 2 and 3. We are grateful to Katherine Whitacre for valuable help with AAR analyses at NAU, and John Southon for the AMS analyses at UCI. Fabrizio Scarabino is acknowledged for his taxonomic assistance. We would also like to thank Claudio G. De Francesco, Mar´ıa A.G. Pivel, and Michał Kowalewski for their valuable comments on earlier drafts of this report. The authors also thank Jos´e Ortiz and an anonymous reviewer, as well as the Associate Editor Adam Tomaˇsov´ych, for their helpful comments and constructive criticism that considerably improved the manuscript. SUPPLEMENTAL MATERIAL

Data are available from the PALAIOS Data Archive: http://www.sepm.org/pages.aspx?pageid=332. REFERENCES

ABSALA˜ O, R.S., 1991, Environmental discrimination among soft-bottom mollusc associations off Lagoa dos Patos, South Brazil: Estuarine, Coastal and Shelf Science, v. 32, p. 71–85, doi: 10.1016/0272-7714(91)90029-B. ˇ YCH ´ , A., STACHOWITSCH, ALBANO, P.G., FILIPPOVA, N., STEGER, J., KAUFMAN, D.S., TOMASOV M., AND ZUSCHIN, M., 2016, Oil platforms in the Persian (Arabian) Gulf: living and death assemblages reveal no effects: Continental Shelf Research, v. 121, p. 21–34, doi: 10.101 6/j.csr.2015.12.007. ALLEN, P.A., KOSNIK, M.A., AND KAUFMAN, D., 2013, Characterizing the dynamics of amino acid racemization using time-dependent reaction kinetics: a Bayesian approach to fitting age-calibration models: Quaternary Geochronology, v. 18, p. 63–77, doi: 10.1016/j. quageo.2013.06.003. ANGULO, R.J., LESSA, G.C., AND SOUZA, M.C, 2006, A critical review of mid- to lateHolocene sea-level fluctuations on the eastern Brazilian coastline: Quaternary Science Reviews, v. 25, p. 486–506, doi: 10.1016/j.quascirev.2005.03.008. ANGULO, R.J., SOUZA, M.C., REIMER, P.J., AND SASAOKA, S.K., 2005, Reservoir effect of the Southern and Southeastern Brazilian Coast: Radiocarbon, v. 47, p. 67–73. BERKELEY, A., PERRY, C.T., SMITHERS, S.G., AND HOON, S., 2014, Towards a formal description of foraminiferal assemblage formation in shallow-water environments:

581

qualitative and quantitative concepts: Marine Micropaleontology, v. 112, p. 27–38, doi: 10.1016/j.marmicro.2014.08.005. BEST, M.M.R., 2008, Contrast in preservation of bivalve death assemblages in siliciclastic and carbonate tropical shelf settings: PALAIOS, v. 23, p. 796–809, doi: 10.2110/ palo.2005.p05-076r. BUSH, S.L., SANTOS, G.M., XU, X., SOUTHON, J.R., THIAGARAJAN, N., HINES, S.K., AND ADKINS, J.F., 2013, Simple, rapid, and cost effective: a screening method for 14C analysis of small carbonate samples: Radiocarbon, v. 55, p. 631–40, doi: 10.2458/azu_js_rc.55.1 6192. ˆ , I.C.S., AND ASP, N.E., 1999, Inner shelf and beach seashell CALLIARI, L.J., CORREA resources in Southern Brazil, in L.R. Martins and C.I. Santana (eds.), Non Living Resources of the Southern Brazilian Coast Zone and Continental Margin: IOCUNESCO, OSNLR, SERG, p. 39–49. CAP´ITOLI, R.R. AND BEMVENUTI, C., 2004, Distribui¸ca˜o batim´etrica e varia¸co˜ es de diversidade dos macroinvertebrados bentˆonicos da plataforma continental e talude superior no extremo sul do Brasil: Atlˆantica, v. 26, p. 27–43. CECILIO, R.O., 2015, An ocean wind-wave climatology for the Southern Brazilian shelf: Programa de P´os-Gradua¸ca˜o em Geociˆencias, Universidade Federal do Rio Grande do Sul, Ph.D. Thesis, 116 p. http://www.renatocecilio.org/thesis/Thesis01_Cecilio_2015. pdf. Checked January 2017. ˆ , I.C.S., 1983, Dep´ CORREA ositos de material carbona´tico da plataforma continental interna do Rio Grande do Sul (Brasil): Proceedings International Conference on Marine Resources, Vi˜na del Mar, p. 155–163. ˆ , I.C.S., 1996, Les variations du niveau de la mer durant les derniers 17.500 ans BP: CORREA l’exemple de la plate-forme continentale du Rio Grande do Sul-Br´esil: Marine Geology, v. 130, p. 163–178, doi:10.1016/0025-3227(95)00126-3. ˆ , I.C.S. AND PONZI, V.R.A., 1978, Dep´ CORREA ositos de calca´rio biodetr´ıtico das regi˜oes do Albarda˜o e Mostardas na plataforma interna do Rio Grande do Sul: Anais do XXX Congresso Brasileiro de Geologia, Recife, v. 2, p. 851–866. CPRM, 2009, Geologia da plataforma continental jur´ıdica brasileira e a´reas oceˆanicas adjacentes: dados organizados em sistema de informa¸ca˜o geogra´fica: Rio de Janeiro, DVD. DAVIES, D.J., POWELL, E.N., AND STANTON JR., R.J., 1989, Taphonomic signature as a function of environmental processes: shells and shell beds in a hurricane-influenced inlet of the Texas coast: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 72, p. 317– 356, doi: 10.1016/0031–0182(89)90150–8. DEMARCHI, B. AND COLLINS, M., 2015, Amino Acid Racemization Dating, in W. Rink and W. Thompson (eds.), Encyclopedia of Scientific Dating Methods: Springer Netherlands, p. 1–22, doi: 10.1007/978-94-007-6326-5_73-1. ˜ , M.G., DEXTER, T.A., KAUFMAN, D.S., KRAUSE JR., R.A., BARBOUR WOOD, S.L., SIMOES HUNTLEY, J.W., YANES, Y., ROMANEK, C.S., AND KOWALEWSKI, M., 2014, A continuous multimillennial record of surficial bivalve mollusk shells from the Sa˜o Paulo Bight, Brazilian shelf: Quaternary Research, v. 81, p. 274–283, doi: 10.1016/j.yqres.2013.12.007. DILLENBURG, S.R., 1994, O potencial de preserva¸ca˜o dos registros sedimentares do Sistema Deposicional Laguna/Barreira IV na costa do Rio Grande do Sul: Notas T´ecnicas, v. 9, p. 1–11. DILLENBURG, S.R. AND BARBOZA, E.G., 2014, The strike-fed sandy coast of Southern Brazil, in I.P. Martini and H.R. Wanless (eds.), Sedimentary Coastal Zones from High to Low Latitudes: Geological Society, London, Special Publications, v. 388, p. 333–352, doi: 10.1144/SP388.16. DILLENBURG, S.R., ESTEVES, L.S., AND TOMAZELLI, L.J., 2004, A critical evaluation of coastal erosion in Rio Grande do Sul, Southern Brazil: Anais da Academia Brasileira de Ciˆencias, v. 76, p. 611–623, doi: 10.1590/S0001-37652004000300014. DOMINGUEZ, J.G., KOSNIK, M.A., ALLEN, A.P., HUA, Q., JACOB, D.E., KAUFMAN, D.S., AND WHITACRE, K., 2016, Time-averaging and stratigraphic resolution in death assemblages and Holocene deposits: Sydney Harbour’s molluscan record: PALAIOS, v. 31, p. 563– 574, doi: 10.2110/palo.2015.087. D’ORBIGNY, A.D, 1834–1847, Mollusques, in C.P. Bertrand (ed.), Voyage dans l’Amerique Meridionale (Le Bresil, La Republique Orientale de L’Uruguay, La Republique Argentine, La Patagonie, La Republique du Chili, La Republique de Bolivia, La Republique du Perou), execute pendant les annees 1826, 1827, 1828, 1829, 1830, 1831, 1832 et 1833: Chez Ve. Levrault, Paris, Livraison 82, p. 489–528. ERTHAL, F., KOTZIAN, C.B., AND COIMBRA, J.C., 2014, Quantitative environmental interpretation based on damage profile of recent molluscan assemblages from Brazil Continental Shelf: International Palaeontological Congress 4, Abstracts, Mendoza, IPA, p. 80. FIGUEIREDO, A.G., 1975, Geologia dos dep´ositos calca´rios biodetr´ıticos da Plataforma Continental do Rio Grande do Sul: Programa de P´os-Gradua¸ca˜o em Geociˆencias, Universidade Federal do Rio Grande do Sul, M.Sc. Thesis, 72 p. GELMAN, A., CARLIN, J.B., STERN, H.S., AND RUBIN, D.B., 2014, Bayesian Data Analysis: Chapman and Hall/CRC Press, v. 2, 621 p. HALFAR, J., GODINEZ-ORTA, L., MUTTI, M., VALDEZ-HOLGU´IN, J.E., AND BORGES, J.M., 2004, Nutrient and temperature controls on modern carbonate production: an example from the Gulf of California, Mexico: Geology, v. 32, p. 213–216, doi: 10.1130/G20298.1. HUNT, G., 2004, Phenotypic variation in fossil samples: modeling the consequences of time-averaging: Paleobiology, v. 30, p. 426–443, doi: 10.1666/0094-8373(2004)030,0 426:PVIFSM.2.0.CO;2.

582

M.N. RITTER ET AL.

JAMES, N.P. AND BONE, Y., 2011, Carbonate production and deposition in a warm-temperate macroalgal environment, Investigator Strait, South Australia: Sedimentary Geology, v. 240, p. 41–53, doi: 10.1016/j.sedgeo.2011.07.005. KAUFMAN, D.S., 2015, Amino Acid Racemization, Marine sediments, in W. Rink and W. Thompson (eds.), Encyclopedia of Scientific Dating Methods: Springer Netherlands, p. 1–4, doi: 10.1007/978-94-007-6326-5_16-4. KAUFMAN, D.S. AND MANLEY, W.F., 1998, A new procedure for determining DL amino acid ratios in fossils using reverse phase liquid chromatography: Quaternary Science Reviews, v. 17, p. 987–1000, doi: 10.1016/S0277-3791(97)00086-3. KIDWELL, S.M., 2013, Time-averaging and fidelity of modern death assemblages: building a taphonomic foundation for conservation paleobiology: Palaeontology, v. 56, p. 487–522, doi: 10.1111/pala.12042. KIDWELL, S.M., BEST, M.M.R., AND KAUFMAN, D.S., 2005, Taphonomic trade-offs in tropical marine death assemblages: differential time averaging, shell loss, and probably bias in siliciclastic vs. carbonate facies: Geology, v. 33, p. 729–732, doi: 10.1130/G21607.1. KIDWELL, S.M. AND BRENCHLEY, P.J., 1994, Patterns in bioclastic accumulations through the Phanerozoic: changes in input or in destruction?: Geology, v. 22, p. 1139–1143, doi: 10.1130/0091-7613(1994)022,1139:PIBATT.2.3.CO;2. KOSNIK, M.A., HUA, Q., JACOBSEN, G.E., KAUFMAN, D.S., AND WUST, R.A., 2007, Sediment mixing and stratigraphic disorder revealed by the age-structure of Tellina shells in Great Barrier Reef sediment: Geology, v. 35, p. 811–814, doi: 10.1130/G23722A.1. KOSNIK, M.A., HUA, Q., KAUFMAN, D.S., AND WUST, R.A., 2009, Taphonomic bias and timeaveraging in tropical molluscan death assemblages: differential shell half-lives in Great Barrier Reef sediment: Paleobiology, v. 35, p. 565–586, doi: 10.1666/0094-8373-35.4. 565. KOSNIK, M.A., HUA, Q., KAUFMAN, D.S., AND ZAWADZKI, A., 2015, Sediment accumulation, stratigraphic order, and the extent of time-averaging in lagoonal sediments: a comparison of 210Pb and 14C/amino acid racemization chronologies: Coral Reefs, v. 34, p. 215–229, doi: 10.1007/s00338-014-1234-2. KOSNIK, M.A. AND KAUFMAN, D.S., 2008, Identifying outliers and assessing the accuracy of amino acid racemization measurements for geochronology, II, data screening: Quaternary Geochronology, v. 3, p. 328–341, doi: 10.1016/j.quageo.2008.04.001. KOSNIK, M.A., KAUFMAN, D.S., AND HUA, Q., 2013, Radiocarbon-calibrated multiple amino acid geochronology of Holocene molluscs from Bramble and Rib Reefs (Great Barrier Reef, Australia): Quaternary Geochronology, v. 16, p. 73–86, doi: 10.1016/j.qua geo.2012.04.024. KOWALEWSKI, M., 1996, Time-averaging, overcompleteness, and the geological record: The Journal of Geology, v. 104, p. 317–326, doi: 10.1086/629827. KOWALEWSKI, M., GOODFRIEND, G.A., AND FLESSA. K.W., 1998, High-resolution estimates of temporal mixing within shell beds: the evils and virtues of time-averaging: Paleobiology, v. 24, p. 287–304. ˆ , L.A.P., KOWSMANN, R.O., COSTA, M.P.A., VICALVI, M.A., COUTINHO, M.G.M., AND GAMBOA 1977, Modelo da sedimenta¸ca˜o holocˆenica na plataforma continental sul brasileira. Rio de Janeiro: PETROBRAS, CENPES, DINTEP, S´erie Projeto REMAC, v. 2, p. 7–26. ˜ , KRAUSE JR., R.A., BARBOUR, S.L., KOWALEWSKI, M., KAUFMAN, D.S., ROMANEK, C.S., SIMOES M.G., AND WEHMILLER, J.F., 2010, Quantitative comparisons and models of timeaveraging in bivalve and brachiopod shell accumulations: Paleobiology, v. 36, p. 428– 452, doi: 10.1666/08072.1. LAMBECK, K., ROUBY, H., PURCELL, A., SUN, Y., AND SAMBRIDGE, M., 2014, Sea level and global ice volumes from the Last Glacial Maximum to the Holocene: Proceedings of the National Academy of Sciences, v. 111, p. 15296–15303, doi: 10.1073/pnas.1411762111. LINNAEUS, C., 1767, Systema naturae per regna tria naturae: secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis, 12th edition: Laurentius Salvius, Holmiae, Stockholm, Sweden, p. 533–1327. LOCARNINI, R.A., MISHONOV, A.V., ANTONOV, J.I., BOYER, T.P., GARCIA, H.E., BARANOVA, O.K., ZWENG, M.M., PAVER, C.R., REAGAN, J.R., JOHNSON, D.R., HAMILTON, M., AND SEIDOV, D., 2013, Temperature, in S. Levitus (ed.) World Ocean Atlas 2013, v. 1, A. Mishonov Technical, NOAA Atlas NESDIS 73, 40 p. Available at https://www.nodc.noaa.gov/cgibin/OC5/woa13/woa13.pl?parameter¼t. Checked April 2017. LOPES, R.P. AND BUCHMANN, F.S.C., 2011, Pleistocene mammals from the southern Brazilian continental shelf: Journal of South American Earth Sciences, v. 31, p. 17–27, doi: 10.1016/j.jsames.2010.11.003. MAHIQUES, M.M., SOUSA, S.H.M., BURONE, L., NAGAI, R.H., SILVEIRA, I.C.A., FIGUEIRA, R.C.L., SOUTELINO, R.G., PONSONI, L., AND KLEIN, D.A., 2011, Radiocarbon geochronology of the sediments of the Sa˜o Paulo Bight (southern Brazilian upper margin): Anais da Academia Brasileira de Ciˆencias, v. 83, p. 817–834, doi: 10.1590/S0001-3765201100 5000028. MAHIQUES, M.M., TESSLER, M.G., CIOTTI, A.M., SILVEIRA, I.C.A.S., SOUZA, S.H.M., FIGUEIRA, R.C.L., AND TASSINARI, C.C.G., 2004, Hydrodynamically-driven patterns of recent sedimentation in the shelf and upper slope off southeast Brazil: Continental Shelf Research, v. 24, p. 1685–1697, doi: 10.1016/j.csr.2004.05.013. MARTINS, L.R.S., 1987, Descri¸ca˜o dos testemunhos da plataforma continental sul brasileira–Geomar VII: Notas T´ecnicas, v. 4, p. 1–158. ˆ , I.C.S., AND MARTINS, I.R., 1996, Late Quaternary MARTINS, L.R.S., URIEN, C.M., CORREA processes along the Rio Grande do Sul continental shelf (southern Brazil): Notas T´ecnicas, v. 9, p. 62–68.

PA L A I O S

¨ MOLLER , O.O., PIOLA, A.R., FREITAS, A.C., AND CAMPOS, E.J.D., 2008, The effects of river discharge and seasonal winds on the shelf off southeastern South America: Continental Shelf Research, v. 28, p. 1607–1624, doi: 10.1016/j.csr.2008.03.012. NAGAI, R.H., SOUSA, S.H.M., AND MAHIQUES, M.M., 2014, The southern Brazilian shelf, in F.L. Chiocci and A.R. Chivas (eds.), Continental Shelves of the World: Their Evolution During the Last Glacio-Eustatic Cycle: Geological Society Memoirs, v. 41, p. 305–313, doi: 10.1144/M41.5. OLSZEWSKI, T.D., 1999, Taking advantage of time-averaging: Paleobiology, v. 25, p. 226– 238. OLSZEWSKI, T.D., 2004, Modeling the influence of taphonomic destruction, reworking, and burial on time-averaging in fossil accumulations: PALAIOS, v. 19, p. 39–50, doi: 10.1 669/0883-1351(2004)019,0039:MTIOTD.2.0.CO;2. OLSZEWSKI, T.D., 2012, Remembrance of things past: modelling the relationship between species’ abundances in living communities and death assemblages: Biology Letters, v. 8, p. 131–134, doi: 10.1098/rsbl.2011.0337 OLSZEWSKI, T.D. AND KAUFMAN, D.S., 2015, Tracing burial history and sediment recycling in a shallow estuarine setting (Copano Bay, Texas) using postmortem ages of the bivalve Mulinia lateralis: PALAIOS, v. 30, p. 224–237, doi: 10.2110/palo.2014.063. R CORE TEAM, 2016, R: A language and environment for statistical computing: R Foundation for Statistical Computing, Vienna, www.R-project.org (July 2016). REIMER, P.J., BARD, E., BAYLISS, A., BECK, J.W., BLACKWELL, P.G., RAMSEY, B.C., BUCK, C.E., CHENG, H., EDWARDS, R.L., FRIEDRICH, M., GROOTES, P.M., GUILDERSON, T.P., HAFLIDASON, H., HAJDAS, I., HATTE´, C., HEATON, T.J., HOFFMANN, D.L., HOGG, A.G., HUGHEN, K.A., KAISER, K.F., KROMER, B., MANNING, S.W., NIU, M., REIMER, R.W., RICHARDS, D.A., SCOTT, E.M., SOUTHON, J.R, STAFF, R.A., TURNEY, C.S.M., AND VAN DER PLICHT, J., 2013, IntCal13 and Marine13 radiocarbon age calibration curves 0– 50,000 years cal BP: Radiocarbon, v. 55, p. 1869–1887, doi: 10.2458/azu_js_rc.55.1 6947. RITTER, M.N., DE FRANCESCO, C.G., ERTHAL, F., HASSAN, G.S., TIETZE, E., AND MART´INEZ, S., 2016, Manifesto of the South American School of (actualistic) taphonomy: PALAIOS, v. 31, p. 20–24, doi: 10.2110/palo.2015.065. RITTER, M.N., ERTHAL, F., AND COIMBRA, J.C., 2013, Taphonomic signatures in molluscan fossil assemblages from the Holocene lagoon system in the northern part of the coastal plain, Rio Grande do Sul State, Brazil: Quaternary International, v. 305, p. 5–14, doi: 10.1016/j.quaint.2013.03.013. RIVERS, J.M., JAMES, N.P., KYSER, T.K., AND BONE, Y., 2007, Genesis of palimpsest coolwater carbonate sediment on the continental margin of southern Australia: Journal of Sedimentary Research, v. 77, p. 480–494, doi: 10.2110/jsr.2007.046. SCARPONI, D., KAUFMAN, D.S., AMOROSI, A., AND KOWALEWSKI, M., 2013, Sequence stratigraphy and the resolution of the fossil record: Geology, v. 41, p. 239–242, doi: 10.1130/G33849.1. SMITH, E.A., 1915, Mollusca Part I, Gastropoda Prosobranchia, Scaphopoda and Pelecypoda. British Antarctic (‘‘Terra Nova’’) Expedition, 1910: Natural History Report, Zoology, v. 2, p. 61–112. SOUZA, R.B. AND ROBINSON, I.S., 2004, Lagrangian and satellite observations of the Brazilian Coastal Current: Continental Shelf Research, v. 24, p. 241–262, doi: 10.1016/j. csr.2003.10.001. STUIVER, M., REIMER, P.J., AND REIMER, R.W., 2005, CALIB 7.1, Radiocarbon calibration program. http://calib.qub.ac.uk/calib/. Checked December 2016. TERRY, R.C. AND NOVAK, M., 2015, Where does the time go?: Mixing and the depthdependent distribution of fossil ages: Geology, v. 43, p. 487–490, doi: 10.1130/G3 6483.1. ˇ YCH ¨ ´ TOMASOV , A., FURSICH , F.T., AND OLSZEWSKI, T.D., 2006, Modeling shelliness and alteration in shell beds: variation in hardpart input and burial rates leads to opposing predictions: Paleobiology, v. 32, p. 278–298, doi: 10.1666/0094-8373(2006)32[278: MSAAIS]2.0.CO;2. ˇ YCH ´ TOMASOV , A. AND KIDWELL, S.M., 2010, Predicting the effects of increasing temporal scale on species composition, diversity, and rank-abundance distributions: Paleobiology, v. 36, p. 672–695, doi: 10.1666/08092.1. ˇ YCH ´ TOMASOV , A., KIDWELL, S.M., AND BARBER, R.F., 2016a, Inferring skeletal production from time-averaged assemblages: skeletal loss pulls the timing of production pulses towards the modern period: Paleobiology, v. 42, p. 54–76, doi: 10.1017/pab.2015.30. ˇ YCH ´ , A., KIDWELL, S.M., BARBER, R.F., AND KAUFMAN, D.S., 2014, Long-term TOMASOV accumulation of carbonate shells reflects a 100-fold drop in loss rate: Geology, v. 42, p. 819–822, doi: 10.1130/G35694.1. ˇ YCH ¨ ˇ ´ ´ , N., 2016b, Temporal and , A., SCHLOGL , J., KAUFMAN, D.S., AND HUDA´CKOVA TOMASOV bathymetric resolution of nautiloid death assemblages in stratigraphically condensed oozes (New Caledonia): Terra Nova, v. 28, p. 271–278, doi: 10.1111/ter.12218. TOMAZELLI, L.S., DILLENBURG, S.R., AND VILLWOCK, J.A., 1996, Late Quaternary geological history of Rio Grande do Sul coastal plain, southern Brazil: Revista Brasileira de Geociˆencias, v. 30, p. 474–476. TOMAZELLI, L.J. AND VILLWOCK, J.A., 2000, O Cenoz´oico no Rio Grande do Sul: Geologia da Plan´ıcie Costeira, in M. Holz and L.F. De Ros (eds.), Geologia do Rio Grande do Sul: CIGO/UFRGS, p. 375–406. VICALVI, M.A., 1977, Sedimentos quaterna´rios da plataforma continental e talude do sul do Brasil: estratigrafia e curvas paleoclima´ticas, in Evolu¸ca˜o sedimentar holocˆenica da

PA L A I O S

SPATIAL VARIATION IN TIME-AVERAGING

plataforma continental e do talude do sul do Brasil, Rio de Janeiro: PETROBRAS, CENPES, DINTEP, Projeto REMAC Series, v. 2, p. 28–76. VILLWOCK, J.A. AND TOMAZELLI, L.J., 1995, Geologia costeira do Rio Grande do Sul: Notas T´ecnicas, v. 8, 1–45. VILLWOCK, J.A., TOMAZELLI, L.J., LOSS, E.L., DEHNHARDT, E.A., HORN FILHO, N.O., BACHI, F.A., AND DEHNHARDT, B.A., 1986, Geology of the Rio Grande do Sul Coastal Province: Quaternary of South America and Antarctica Peninsula, v. 4, p. 79–97. WALKER, K.R. AND BAMBACH, R.K., 1971, The significance of fossil assemblages from finegrained sediments: time-averaged communities: Geological Society of America Annual Meeting Program with Abstracts, v. 3, p. 783–784. WEHMILLER, J.F. AND MILLER, G.H., 2000, Aminostratigraphic dating methods in Quaternary Geology, in J.S. Noller, J.M. Sowers, and W.R. Lettis (eds.), Quaternary

583

Geochronology: Methods and Applications: American Geophysical Union, Washington, DC, p. 187–222, doi: 10.1029/RF004p0187. ZEMBRUSCKI, S.G., 1979, Geomorfologia da margem continental sul brasileira e das bacias oceˆanicas adjacentes, in H.A.F. Chaves (ed.), Geomorfologia da margem continental brasileira e das bacias oceˆanicas adjacentes, Rio de Janeiro: PETROBRAS, ENPES, DINTEP, Projeto REMAC Series, v. 7, p. 129–177. ZUSCHIN, M., STACHOWITSCH, M., AND STANTON, R.J., 2003, Patterns and processes of shell fragmentation in modern and ancient marine environments: Earth-Science Reviews, v. 63, p. 33–82, doi: 10.1016/S0012-8252(03)00014-X.

Received 10 January 2017; accepted 13 July 2017.