Downloaded by guest on September 25, 2021 g ntecneto oiea ilg n eemn that determine and reconstruction 3D biology poriferan of morphol- context reconstructed the this in evaluate We ogy sheets. mor- that ing/merging branch- reveal Our and connectivity, meandering millimeter-thick affinity. approximately and produced proposed spacing, we measurements organism’s thickness, quantitative Here, of the comprise which . assess analyses, phological to in growing use of found buildups we reconstructions is microbial three-dimensional that on) present construction to (and encrusting affinity between modular poriferan a a sis, assign been morpho- have to demosponges apparent and coralline used Recently, affinities extant identified. with biological similarities be their logical can biomin- if analogs of especially modern provide impact Earth, environmental puta- may million on and enigmatic, 538 eralization evolution of These the to into fossils. examples million insight metazoan several (635 shell-building contain 2020) Period tive 12, [Ma]) May review ago for (received the years 2020 23, June from approved and Strata MA, Cambridge, University, Harvard Knoll, H. Andrew by Edited 91125 CA Pasadena, Technology, of Institute California Sciences, 03755; Planetary NH Hanover, College, Dartmouth Sciences, a Mehra Akshay construction microbial metazoan putative the of reconstructions Three-dimensional www.pnas.org/cgi/doi/10.1073/pnas.2009129117 reconstruc- said, That (8). modern as of way tions skeletons same formed the organism in the that suggesting phos- that biomineralizers, demonstrated has however, matrix morphology, work, phatized organic recent cup” an More in on 7). “cup carbonate (5, a precipitated 6). have of (5, to up calcified thought made was lightly organism been tubular have a to Conversely, ina, shown been (4). a has that ultrastructure also produced suggested organism, calcitic have goblet-shaped foliated Workers flexible, a unresolved. made organism , remains each degree, parts what biomin- to hard putative and Ediacaran how, the Exactly among eralizers. varied have to appear ing the are impacts growth debate. ongoing their environmental of and and subject enigmatic, affinities, be biological to habits, proven have organisms skeletal these earliest and from the tubulites, genera settings: four study water comprises shallow record to Ediacaran This record. necessary first the is metazoan with eco- it associated environmental, ramifications the biomineralizers, determine evolutionary to and as logical, well sedimentological, as that biomineralize, planet’s impact outsize the the makeup. on geochemical of Great and had biological, the indicative has (e.g., is constructions biomineralization which organic ), building largest Barrier for Earth’s responsible of are some organisms (2). biomineralizing surroundings framework Today, their producing phyla engineering were effectively and modern dwellers constructions most reef which skeletal 538.8 during emerged, to period Ma first time 538.6 a [beginning (1)], radiation Ma the By zoans. I eateto esine,PictnUiest,Pictn J08544; NJ Princeton, University, Princeton Geosciences, of Department oewtest h ria fpttv imnrlzn meta- biomineralizing putative reefs of microbe-dominated arrival the Ma), to (∼550 witness Ediacaran bore Late the n ihrsett al imnrlzto,mdso hl build- shell of modes biomineralization, early to respect With to began animals why and where, when, understand To aet etteasrinta h organism the that assertion the test to made Cloudina, Cloudina ieyi o pnegaeorganism. -grade a not is likely Namapoikia | a,b,c,1 Ediacaran hr aoatclt arcwt extant with fabric nanoparticulate a share elyA Watters A. Wesley , Namapoikia | 3.Atog opooial simple, morphologically Although (3). al life early Namacalathus, aaokarietoogen- Namapoikia d eateto srnm,WlelyClee else,M 28;and 02481; MA Wellesley, College, Wellesley Astronomy, of Department d onP Grotzinger P. John , Namapoikia Cloudina, Namacalathus Namapoikia hwta twsa was it that show Cloud- Sino- that b ekmIsiue atot olg,Hnvr H03755; NH Hanover, College, Dartmouth Institute, Neukom e n dmC Maloof C. Adam and , doi:10.1073/pnas.2009129117/-/DCSupplemental at online information supporting contains article This 1 aadpsto:Tecmuainlsuc oeue nti ae saalbei GitHub in available is paper this in used at code source computational The deposition: Data the under Published Submission.y the Direct PNAS wrote a is A.C.M. article This and interest.y A.M. competing no and declare authors per- data; The A.M. analyzed research; A.C.M. designed and A.C.M. paper.y A.M. and research; J.P.G., W.A.W., formed A.M., contributions: Author acfigognsso entv oiea fnt eepresent were more affinity for poriferan 18 definitive and Cambrian, of early 17 the organisms refs. by calcifying see Additionally, also debate). of Ediacaran 16; examples ref. the recent see of review, (for onset complete researchers and the a debated—by than described—and biomarkers, older been Spicules, have all Period, fossils, remains. body sponge full exam- even purported with replete of is record ples 635 fossil to Ma the (720 Indeed, Period Ma). Cryogenian the during evolved poriferans similarities, affinity. the poriferan these of of that basis like the is that pathway demosponges extant biomineralizing a and inferred have Chaetetid with that encrusting characteristics proposed repeatedly morphological have shares cases, Workers two (13). some another the in one over and, with competing growths, forms of microbial life surrounding Studies and (13). construction scaffold sec- of organic longitudinal and tions an transverse (2D) calcifying mate- two-dimensional polished, rapidly skeletal by produced rial construction, organic encrusting predominately labyrinthine a of up made were (12). that matrix ones but walls, tubes. furthering that (11), idea individuals the deformed aggregates and that transported revealed comprise 10), (9, frameworks wave-resistant built owo orsodnemyb drse.Eal [email protected] Email: addressed. be may correspondence whom To fseea nml.Ti eodi aeu ffu eeafrom genera four record period: of earliest Ediacaran up made the the is study record to This animals. necessary skeletal is of it on emerged, first biomineraliza- impact metazoan tion why outsized and cycles. where, an sedimentological when, determine To and have geochemical, skeletons biological, Earth’s build that Animals Significance a hrceitc xetdo pne r o htmte,an matter, that for or, sponge, animal. a of expected that characteristics cal find We sponge. calcifying of structions and https://github.com/giriprinceton/namapoikia oeua lc n hlgntcetmts(5 ugs that suggest (15) estimates phylogenetic and clock Molecular that proposed have researchers Recently, ee emauetredmninlrecon- three-dimensional measure we Here, Namapoikia. nte uua raim lohdplastic had also organism, tubular another Sinotubulites, Namapoikia Cloudina NSlicense.y PNAS Namapoikia Vaceletia ugs ope nepa ewe the between interplay complex a suggest Namacalathus, rdcdweakly-to-non-biomineralized produced a ots h yohssta ti a is it that hypothesis the test to and Namapoikia Namapoikia . Acanthochaetetes y . y e iiino elgcland Geological of Division https://www.pnas.org/lookup/suppl/ Cloudina, NSLts Articles Latest PNAS a enasge a assigned been has c eateto Earth of Department ak h physi- the lacks Sinotubulites, a Namapoikia, Namapoikia 1,1) On 14). (13, | f7 of 1

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES on Earth, and some sponges, namely the Archaeocyathids, ramp in the northern Zaris subbasin coincident with conver- were even responsible for the world’s first framework reefs (2). gence along the Damara and Gariep orogens (20, 21). There It stands to reason that poriferans, having appeared during the are no direct radiometric dates from the Driedoornvlakte stratig- Cryogenian and eventually becoming the dominant engineers raphy. However, uranium–lead zircon ages from the Kuibis of the Early Cambrian, may have first evolved the ability to constrain the maximum depositional age to 548.8 ± 1 Ma build calcified skeletons during the Late Ediacaran. To test this (20). Additionally, a uranium–lead zircon date in the overlying idea, we seek to determine whether Namapoikia is, in fact, a Schwarzrand subgroup (22) provides a minimum deposition age sponge-grade organism. of 545.41 ± 1 Ma. Like many other Ediacaran fossils, specimens of Namapoikia The reef at Driedoornvlakte Farm, which is 500 m thick, 10 km lack soft tissue preservation and exhibit signs of diagenetic alter- long, and dips 25 to 40◦ to the southeast, sits on Precambrian ation (e.g., recrystallization). As a result, it is necessary to analyze quartzite. The carbonate ramp was created over the course of the gross morphological characteristics of Namapoikia speci- three distinct accommodation cycles; in the final stage, just prior mens, such as the size, shape, and distribution of structures, to drowning by shales of the Urikos Member, pinnacle buildups in order to describe growth habit, identify possible analogs, developed on the platform margins (21). These pinnacles com- and evaluate biological affinity. Since 2D measurements (e.g., prise microbial mounds made up of a combination of columnar made on polished slabs or on bedding planes in outcrop) or encrusting and columnar or massive thrombo- are subject to misinterpretation and measurement error (11), lites. Both Cloudina and Namacalathus can be found in the three-dimensional (3D) data are required for accurate analysis. fill between microbial buildups and in clinoformal grainstones. Unfortunately, Namapoikia skeletons are preserved as carbonate Namapoikia is found encrusting the walls and tops of microbial minerals within carbonate rock, precluding isolation via acid dis- buildups in decimeter-wide neptunian dykes (Fig. 1 D and E), solution or imaging with traditional, density-sensitive techniques. which are shallow fractures in the reef that opened to the To address this problem, we utilize serial grinding and imag- seafloor. ing, a method which relies on color and texture to differentiate Individual Namapoikia can be up to 1 m wide and up to between features of interest (e.g., fossils) and the surrounding 0.25 m tall (13, 23). Namapoikia specimens comprise regularly matrix (6, 11, 19). The resulting 3D reconstructions, combined spaced, millimeter-thick structural elements, here referred to as with field observations, enable us to quantitatively assess the “partitions.” These elements split and merge in transverse and affinity and paleoecology of Namapoikia. longitudinal directions (see Fig. 2 for a visual reference to the spatial conventions used in this work). In certain specimens, par- Geologic Setting, Field Observations, and Reconstructions titions have been said to be intersected by perpendicular tabulae In situ examples of Namapoikia are found in a pinnacle reef (13). A detailed survey of the pinnacle reef at Driedoornvlakte outcropping on Driedoornvlakte Farm near Reitoog, Namibia reveals that Namapoikia is rare, occurring in 3% of observed (WGS84 UTM 33K 667792E 7358951N; Fig. 1 A and B). These locations (Fig. 1C). To the best of our knowledge, Namapoikia rocks are part of the Kuibis Subgroup (Omkyk Member) of has not been described in other, contemporaneous rocks in the Neoproterozoic . They formed on a carbonate Namibia or, for that matter, anywhere else on Earth.

A B D

C

E

Fig. 1. Field observations. (A) Location of the study area in Namibia. The red square is Driedoornvlakte Farm, while the light gray fill depicts the geo- graphic extent of Nama Group rocks. (B) View of the reef complex at Driedoornvlakte Farm, with the study area marked by the yellow rectangle. (C)(Top) Interpolated survey data showing the occurrence of Namapoikia in the reef. The contour interval is 5 m; contours were generated from a drone imagery- derived digital surface model (11). (Bottom) Pie charts depicting the percent of observed locations exhibiting a feature (presence of feature denoted in black). (D and E) Field photographs of in situ Namapoikia. Map coordinates in B and C are with reference to WGS84 UTM 33K.

2 of 7 | www.pnas.org/cgi/doi/10.1073/pnas.2009129117 Mehra et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 edrn ftesml eosrcin Saebra h otmlf fec ae,05c. h ieto fsrtgahcu sdenoted is up stratigraphic of direction The cm.) al. panel. 0.5 et each Mehra panel, in each bar of labeled scale left the marker bottom above circle the figure The at axis analyses. bar the morphological (Scale on for reconstruction. arrow used red B is the sample square, by of the red (C slice of outlined Single . Rendering the (B) blocky by (D) rock. of marked the latitudinal. within subregion, example fracture a an a line), denotes filling dotted dolomite 1 yellow to a pointing is using 2 (denoted labeled fabrics marker circle the while sample, Namapoikia the throughout distributed is that calcite 735/1,221/1,881 of thickness a have voids 539/831/1,122 a of have thickness a voids have interpartition tions the 1,559/2,402/3,547 while of text), through thickness the used of is 640/1,030/1,393 remainder convention of the this thickness a percentiles; have (25th/50th/75th partitions µm A, sample In ples. S2 i.2. Fig. tabu- of evidence 2 no (Fig. with are lae sections longitudinal phases and matrix transverse in and fossil the that reveals is A recrystallized. B sample sec- thin volume, of Sample petrographic A tion by 2B). planes). (Fig. textures 1.6% fracture microbial by bounded (approximately along dolomite predominately between occurring yellow fill replace of 2 to amount (Fig. appears partitions and occurs the samples spar both calcite black Blocky throughout structures. fine-grained, pinnacles sedimentary of cal- 2 the depositional matrix (Fig. white, from a fill comprise B) by micrite samples surrounded sample partitions Both and cified Farm. A Driedoornvlakte sample at as to (referred CD A hnrcntutd attosmadr rnh n merge and branch, meander, partitions reconstructed, When of specimens two measure and reconstruct We .Priintikesadsaigvr ewe h w sam- two the between vary spacing and thickness Partition ). eosrcin of Reconstructions apeB rcse sn aulsra rnigadiaigpoeue(fe e.6.A the As 6). ref. (after procedure imaging and grinding serial manual a using processed B, sample C and i.S1 F Fig. Appendix, (SI D) A and Namapoikia A and .Tefilcnan oeietsyn- evident no contains fill The B). .I apeB parti- B, sample In 3B). (Fig. µm .Sml nldsasmall a includes A Sample B). igesieof slice Single (A) samples. edrn ftesml eosrcin (Inset reconstruction. A sample the of Rendering ) ,adteinterpartition the and µm, .Atwo-sample A µm. and oisS1 Movies Namapoikia Namapoikia and B hcnse lorjcstenl yohss(P hypothesis null distri- interpartition the on rejects single also test thicknesses a same The from statistic). are Kolmogorov–Smirnov the populations (P rejects the bution that thicknesses hypotheses partition null of test Kolmogorov–Smirnov .We pligabs-tline best-fit a applying When 3E). (Fig. sample each in height (P thicknesses interpartition of 0.44). test (P = a D distribution does same as 0.14), hypothesis the = from null D are the populations rejects the thicknesses that partition Kolmogorov– of two-sample test A Smirnov 671/1,005/1,435. of thickness 566/781/1,020 a of have thickness a have 1,285/1,811/2,377 of thickness ness a have 430/856/1,204 partitions A, the of measured subvolume between In and disparities persist. sub- regions, samples chosen selected two B—are these representative Within subvolume 3D). and (Fig. samples, A voids interpartition subvolume to two referred these and as and partitions of partition the spaced regularly presence in volumes—comprising between the differences whether the thicknesses test to To contributes voids. tributed apeA rcse sn II h icemre aee hw blocky shows 1 labeled marker circle The GIRI. using processed A, sample h eintikeso attosvre ihrsetto respect with varies partitions of thickness median The dis- irregularly large, contain B and A samples both Notably, < iga lutaigtetrstases,lniuia,and longitudinal, transverse, terms the illustrating Diagram ) .0,D=01;Drfr otemgiueo the of magnitude the to refers D 0.19; = D 0.001, ,aditratto od aeathick- a have voids interpartition and µm, Namapoikia ,wie nsboueB partitions B, subvolume in while, µm, pcmni one ythrombolite by bounded is specimen ,aditratto voids interpartition and µm, NSLts Articles Latest PNAS < .0,D=0.41). = D 0.001, < < | 0.001, 0.001, f7 of 3

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES Fig. 3. Measurements made on reconstructions of Namapoikia specimens. (A) The size of various skeletal elements (and voids) from four different hyper- calcified sponges discussed in this study. For V. crypta, A. seunesi, and G. discoforma, bars represent the range of values as reported in literature (refs. 14, 31, and 40, respectively), sometimes from multiple specimens. In the case of A. perforata (SI Appendix, Fig. S2 A and B), bars represent the 25th and 75th percentile bounds of measured data from a single specimen, while the white dots depict median values. (B and C) Histograms illustrating the thickness of partitions and interpartition voids in samples A and B, respectively. The thickness of interpartition voids can be considered equivalent to spacing between partitions. The blue highlighted area depicts the region shown in A. (D) A comparison of partitions and interpartition void thicknesses from subvolumes selected for their regularly spaced partitions (as described in the text). (E) Median partition thickness, mean subtracted, versus height in both samples A and B. Stratigraphic up is in the direction of increasing height, which also is interpreted to be the direction of growth. For each sample, two best-fit trends are shown as dashed lines: in black, the fit takes into account all data, while, in light blue, the fit excludes outliers. Trends are similar even after excluding outliers, so only the r2 of the dotted black line is denoted. (Inset) A box plot showing the range of median partition thicknesses (i.e., as calculated at each sampled height) in both samples.

to the data, the two specimens exhibit different trends, with tive to a sample’s median partition thickness and interpartition partitions thinning toward stratigraphic up in sample A and spacing. In sample A, the change is 117 µm (or 11% and partitions thickening toward stratigraphic up in sample B. 6% of median partition and interpartition thicknesses, respec- In addition to exhibiting low coefficients of determination tively), while, in sample B, the change is 131 µm (or 16% (Fig. 3E), the absolute change of these trends is small rela- and 13%).

4 of 7 | www.pnas.org/cgi/doi/10.1073/pnas.2009129117 Mehra et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 er tal. et Mehra ape obnd.Tecmie atto n interpartition and partition of combined thickness The combined). samples 200 from voids, range subvolumes large examining exclude only that when even and, combined), samples 200 from range 1,200 canal and and skeletal 300 combined 1,000 a from ranging has thickness (31), sponge hypercalcified 14). ref. and 600 from 650 ranging of chambers thickness combined a 950 recent have to spaces the void and with in walls properties (13), example, biological and For morphological share thick. sub- millimeter are demosponge that hypercalcified a features have to generally millimeter Structures sponges 30). calcifying ref. also by (see built peri- 29) (28, as sponges function within to pumps staltic thought gas are densely chambers for small, choanocyte losses), required pump populated effectively resistance/frictional processes to overcome order (and diffusion In water (27). the capture nutrient with and exchange deal to systems is poriferans. that other structure to compared a when to large and anomalously diagenesis—speaks partition postmortem of of effects thickness in combined void the recon- interpartition in said, determine void to That interpartition difficult be of structions. can skeletal expense which of of the effects the at spacing space), elements and thicken- calcareous (e.g., size of processes by the diagenetic ing by controlled Both impacted be tissue. are may observed not living elements dimensions in of are their scale sponges voids the calcareous systems, and of partitions canal skeletons the necessarily the of While scale specimens. poriferan the a from result, comes a osculum, As an fibers. ostia, or system, for tissues, assignment aquiferous spongin, clear any a and/or of of chambers, Specimens (26). remnants choanocyte-bearing osculii from any more through or water one ostia, moves out and that as system known exact aquiferous to While pores branching unique poriferan. traits a group debated, include stem are a synapmorphies represents poriferan fossil the that Namapoikia (see superficial are Appendix similarities S1 SI morphological Fig. apparent Appendix, any (SI that chambers or reconstruc- 3D tabulae However, ). S2 of and S1 tions as Figs. Appendix, such (SI record, Inozoa rock the (18)]. in Archaeocyathids sponges the [e.g., Cambrian Namapoikia standpoint, early morphological the of a see sponges From however; debated, counterpoint). single This is a a Namapoikia (24). prediction for with (this 25 tree walled, (24) thin ref. metazoan spicules was the of likely sponges layer of ancestor common modern last base of the and be (LCA) the Cryogenian, the phylogeny—should near in diverged and group or evolution at of monophyletic placed a terms be that—in to group sponges believe increasingly Researchers Discussion xlie yslcieclicto,ee hncnrligfor be controlling cannot when that even calcification, thicknesses selective void by interpartition explained and partition Namapoikia sponges. known of largest thickness the canal of and skeletal one combined discoforma, the G. as large as times 300 2.9 only: subvolumes considering µm; norreconstructions, our In canal high-density small-diameter, produce typically Sponges for affinity poriferan a to challenge additional An between similarities morphological of lack the Given nadto opouiglreadwdl pcdstructures, spaced widely and large producing to addition In n 1,000 and µm wt h al en 50 being walls the (with µm ihissetlk attosadlc of lack and partitions sheet-like its with Namapoikia, ndaee;Fg 3A). Fig. diameter; in µm o uniaiecmaaieanalysis). comparative quantitative a for sulk h rpsdLAo vntethin-walled the even or LCA proposed the unlike is Namapoikia osba asn eebac ohypercalcified to resemblance passing a bear does ape loehbtsgicn ainei both in variance significant exhibit also samples n te ecie pne,i a eargued be may it sponges, described other and necpinlylarge exceptionally an discoforma , Gigantospongia Namapoikia o2,905 to µm 1,650 to µm hc n uua aasbten700 between canals tubular and thick µm amti hc eae the negates which metric Namapoikia—a cnieigaldt:483 data: all (considering Namapoikia hc a enpooe to proposed been has which crypta, V. sdoubtful. is 5ht 5hpretlsfrboth for percentiles 95th to (5th µm both for percentiles 95th to (5th µm o900 to µm o2,200 to µm Namapoikia hc n tissue-bearing and thick µm aeei crypta Vaceletia nepriinthicknesses interpartition o4,452 to µm ndaee;Fg 3A Fig. diameter; in µm wt al between walls (with µm atto thicknesses partition C–F Namapoikia ,mk tclear it make ), )i pto up is µm) o6,425 to µm Namapoikia Namapoikia n various and Porifera lack µm µm utpe nrmna vnstruhtime. through Simply events incremental walls. multiple, skeletal of 4B) metazoan expression final (Fig. coalescing the ridges of put, microbial result of the evolution not to shape branching and the transverse support of and provide longitudinal indicative eas- of is to topple character or the as baffling deform Thus, so would ily. with otherwise 4), filled that structures be (Fig. microbial would sediment partitions and/or between cement space branch- the and and migration pro- via of 4 flux) (Fig. series ing sediment a (i.e., and/or conditions as changing nutrient, appear to light, time, respond would in model, which moment ridges, our truding given In any at 4). would, (Fig. expression morphological which in niche ecological of surprising. not lack is this twin construction, formal microbial of expression morphological compare and reconstruct to to structure mediated microbially alent interpartition thrombolites. and of attributes partitions these with of consistent are like variance individual also within and voids especially Thrombolites thicknesses, scale distributed metazoans. The than irregularly (36). size large, in contain regular and less thick produce much millimeter-to-centimeter could are characteristically construction by bolites mediated expressed microbially morphologies the a that lows ie o hi ieest h al fgre ae;see mazes; garden of “mace- walls as the to to S3 (referred likeness Fig. curtains Appendix, their vertical for calcified branching riae” create produce fabrics—can also as such clotted thrombolites, Certain of forms. up bolites—made 35). 34, (5, ina tutosotnaergoa 3) hspoet ol account the would is property con- why This hypothesis microbial (33). our for regional of with are expressions often time Consistent structions morphological through reef. that the conditions explained observation on environmental be space local could and in thicknesses In variations interpartition (32). by and flux partition sediment and in of levels, case light the bio- depth, such conditions, water environmental growth, at of by including controlled incremental morphologies reef are of The partly the constructions calcification. result up and the make aggregation, are that Farm, thrombolites Driedoornvlakte and stromatolites of the growth the structures. by such mediated explained microbially that totally be or suggest can partially expression We result, morphological characteristic). a a proposed As sides). this their illustrates on up cards playing Namapoikia and stand much (i.e., to transversely support trying external like without both upright that stay S1F merge to integrity Fig. tural suggest and mean- Appendix, split spaced, (SI we widely longitudinally that as animals, partitions summarized be dering generally, can which more expression, or, Namapoikia sponges of construction. metazoan regular too and a feeding) be filter to (and and that variable bearing pumping actively choanocyte suggest of be capable to scale therefore large too of structures 300 created only observations likely of range our in total seen together, variation a the show of specimens, (∼11% µm single structures. on made sized in both voids and seunesi of spaced thochaetetes measurements regularly example, produce For metazoa) 3 all (Fig. voids irregular h rcue nwhich in fractures The equiv- exactly other no knowledge, our of best the to is, There ohsrmtltscmrsn n aiain—n throm- laminations—and fine stromatolites—comprising Both irbal eitdsdmnaycntutos including constructions, sedimentary mediated Microbially expected regularity the lack which reconstructions, our Given ie h nuneta niomn a on has environment that influence the Given Namapoikia. and Namacalathus Namapoikia B, h bevditrapedifferences intersample observed the Namapoikia, rbbyhdlweegn yotcrle Fg 4 (Fig. relief synoptic emergent low had probably a o metazoan. a not was i–iii). and Namapoikia aohrproposed (another oi S4 Movie .I otat pne adindeed, (and sponges contrast, In B–D). sntfudcocrigwith cooccurring found not is sebae nohrpaleocontinents other on assemblages Namapoikia Namapoikia Namapoikia ol xii o yotcrelief, synoptic low exhibit would arc nthrom- in Fabrics Namapoikia. o u eosrcin.I fol- It reconstruction). our for smorphological Namapoikia’s .Taken 3A). Fig. Namapoikia; aoaaei cooperi Favosamaceria sfudpoie unique a provided found is ,lkl akdtestruc- the lacked likely ), ol rwwt distinct a with grow could ersnsacleto of collection a represents NSLts Articles Latest PNAS Namapoikia Namapoikia .Crypta V. Namapoikia Namapoikia Namapoikia specimens, and analog), Cloud- | Acan- (33), f7 of 5 SI

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES Fig. 4. Diagrams illustrating the proposed model for Namapoikia.(A) Cartoons depicting the presence of Namapoikia in the reef at Driedoornvlakte Farm. Namapoikia grew incrementally in syndepostional fissures, interacting with the thrombolites and encrusting stromatolites that made up the reef. (B)A model of incremental growth, with only several partitions illustrated. (i) Illustration of partitions and void fill at some time 0, with the partitions having low emergent synoptic relief. (ii) A second moment in time, where a partition has grown upward and split and migrated. The locations of splits are marked in red. Transverse migrations and meanders of partitions lead to drift in longitudinal cross-section, such that sheet-like partitions may not be perfectly vertical. (iii) Partitions have both merged and spilt. The location of the merger is marked in yellow.

Our analyses of Namapoikia demonstrate that, in order to reconstruction. In the case of sample A, the rock’s location was recorded assign organismal affinity of problematic fossils based on mor- with a handheld Trimble GeoXH6000 GPS unit, and its field orientation (i.e., phology, 3D observations are required. Quantitative measure- bedding plane direction up) was denoted with arrow markings on multiple ments made on reconstructions can both elucidate the presence faces. or absence of structures for use in identification (e.g., test- Samples A and B both were slabbed and then mounted on to steel plates ing for tabulae and/or chambers; SI Appendix, Fig. S1 C–F) using epoxy adhesive. Each sample then was serially sectioned and imaged and give clues about functional morphology (e.g., the sugges- (final ground dimensions for samples A and B: 41.0 × 57.9 × 19.7 mm and tion that Namapoikia grew with low synoptic relief). The 3D 126.9 × 116.8 × 29.4 mm, respectively). data provide insights about basic organismal form and func- Sample A was processed using the Grinding, Imaging, and Reconstruc- tion that cannot confidently be extracted using 2D observations tion Instrument (GIRI) at Princeton University (11). GIRI comprises a com- puter numerical control surface grinder that has been retrofitted with alone. misting, wiping, and imaging stages. The imaging stage is made up of an 80-megapixel Phase One IQ180 digital back equipped with a 120-mm Materials and Methods Schneider Kreuznach macro lens. This imaging system is positioned verti- Survey Data. To map the spatial extent of various facies, a detailed survey cally so as to attain a 1:1 reproduction ratio at a resolution of 5.73 µm of the pinnacle reef at Driedoornvlakte Farm was conducted. Using a hand- per pixel. For sample A, GIRI programmatically 1) ground away 30 µm of held Trimble GeoXH6000 GPS unit (excluding the external antenna), a total material from the sample surface, 2) wiped off any excess coolant, 3) took of 1,254 GPS points, arrayed on an orthogonal grid with 10 × 20 m spacing, an image, 4) evaluated image quality, and 5) then repeated the grinding were collected. At each point, a set of discrete keywords—corresponding process a total of 658 times. to visible sedimentological, lithological, and/or physical characteristics— Sample B was processed at Massachusetts Institute of Technology (MIT) were recorded. The GPS data were differentially corrected using the GPS using a manual serial grinding and imaging procedure. After removing Pathfinder Office software package at Princeton University. Differential cor- 100 µm of material with a surface grinder, the sample was placed (pol- rections were made with data from the TrigNet Springbok base station, ished side down) on an EPSON flatbed scanner, and data were recorded located 654 km from Driedoornvlakte Farm. The corrected data have a mean at 600 dots per inch resolution (corresponding to a per-pixel resolution of horizontal accuracy of 0.596 m (SD = 0.126 m) and a mean vertical accuracy 42.33 µm). This method was repeated 319 times. of 0.706 m (SD = 0.256 m). For comparative analysis, an A. perforata specimen [originally collected Next, the corrected data—along with associated field observations— by Rigby and Senowbari-Daryan (37)] and an F. cooperi specimen were were examined for the presence or absence of Namapoikia. Points with no selected for 3D reconstruction. The A. perforata sample was processed with evidence of Namapoikia were filtered out, and then multiple modeled semi- GIRI, following the same grinding and imaging procedures as for sample A, variograms were fit to the resulting empirical (culled) dataset. The best-fit with the only difference being a smaller step size (i.e., 20 µm as opposed to semivariogram (in this case, an exponential with range = 127.69, sill = 0.14, 30 µm). A total of 1,624 images of the A. perforata specimen were collected, and nugget = 0.10) was used to perform indicator kriging, which resulted in and approximately 1 mm of the sample was preserved and redeposited at a map of the spatial distributions of Namapoikia. the Smithsonian National Museum of Natural History. The F. cooperi sample also was processed with GIRI, following the same grinding and imaging pro- Sample Collection and Serial Grinding. Two Namapoikia-bearing samples, cedures as for sample A (with a step size of 30 µm). A total of 391 images referred to as A and B, were collected at Driedoornvlakte Farm for 3D of the F. cooperi sample were collected.

6 of 7 | www.pnas.org/cgi/doi/10.1073/pnas.2009129117 Mehra et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 2 .Ce,S egsn .M hu .Hn,Y ho uesrcueadoriginal and structure Tube Zhao, Y. Hong, H. Zhou, M. C. Bengtson, S. Chen, Z. 12. that reveals approach Multiscale Maloof, C. A. Mehra, A. structures attachment and 11. branching Multiple Bowyer, F. Curtis, A. Wood, R. Shore, A. 10. 3 .Wo,A en,Sbtaegot yaisadboieaiaino nEdiacaran an of biomineralization and dynamics growth Substrate Penny, A. Wood, R. 13. 6 .B ncif,R .T alw .D rse,Gvn h al oslrcr fsponges of record fossil early the Giving Brasier, D. M. Callow, T. H. R. Antcliffe, B. J. Biomarkers, 16. glass? the Where’s Peterson, J. K. Pisani, D. Robinson, M. J. Sperling, A. E. Stein- Verticillitidae 15. family Verticillitida, order “Sphinctozoa”, “Recent Vacelet, J. 14. 0 .Gozne,E .Aas .Schr S. Adams, W. E. Grotzinger, J. 20. origins. Maloof sponge C. for A. Searching 19. Nettersheim, J. B. Botting, P. J. 18. Nettersheim J. B. 17. opoueapeito,ad3 paignuo egt vastochastic (via weights neuron updating 3) net- and network the the prediction, through within data a training neuron produce running 2) each to networks, weight, initializing random both a For 1) with class. by work assigned accomplished its to was corresponding training directory a in TIFF a (11 neighborhood class. square given perforata a a A. to pixel, belonging painted as them each marking For thereby user, in B, a stored sample by of and painted case were calculated the were In as superpixel structure. well chosen data as values, a each channel for Blue term) and entropy Green, the Red, an (i.e., a of statistics via covariance of classes and user collection predefined SD, a a mean, of completion, which set Upon a after interface. of For user image, one graphical Matlab. to each in superpixels for assigned written calculated and scripts selected were training custom Next, superpixels of series selected. A, a B—were sample using sample compiled for were five A, data process, sample grinding for the images—three out representative of and number perforata, a First, cation. (38)—was texture and color of applied. basis was B, network the super- sample on on For network—operating made neural used. clusters layer pixel hidden or a In A, pixels, task. sample classification dolomite). of this case and for leveraged the calcite, were blocky networks elements, neural different calcified Two matrix, additional (e.g., no classes required tinct files, JPEG Sample 8-bit conversion. before as adjusted processing. created before not were conversion value were which balance data white images, raw further same B the before the files images, applying TIFF all of RGB to exception 16-bit the to converted With be processing. must which (.IIQ), mat Processing. Image er tal. et Mehra .A .Penny M. A. 9. for fossil Gilbert index A. potential P. a U. P. Cloudina, of 8. distribution and metazoans, structure early Shell calcified Grant, W. of S. reconstruction 7. Digital thrombolite- Grotzinger, P. in J. Watters, metazoans A. Calcified W. Knoll, 6. H. A. Watters, terminal A. W. the Grotzinger, in as interpreted P. metazoan animals J. skeletal Ediacaran biomineralizing 5. Penny, M. Diverse A. Wood, Hong, A. R. H. Zhuravlev, Y. A. Li, 4. G. Xiao, reefs. S. Cambrian Cai, lower of Y. paleoecology and 3. Structure Gangloff, A. R. Rowland, M. S. 2. Linnemann Ulf. 1. nalisacs h erlntokhdt etandpirt classifi- to prior trained be to had network neural the instances, all In dis- into segmented were images analysis, and visualization 3D to Prior opsto fSntblts hlyfsisfo h aenortrzi nsouthern in neoproterozoic late the China. from Shaanxi, fossils Shelly Sinotubulites: of composition constructions. reef (2018). situ E2527 in not and detritus Namibia. Group, Nama cloudinomorphs, in (2014). 1504–1506 344, U.S.A. Sci. Acad. Natl. Proc. Proterozoic. terminal the Namibia. Group, Nama Proterozoic terminal Namibia. Group, Nama (2000). Proterozoic 334–359 terminal 26, the of reefs lophophorate. a explosion. Cambrian the herald Period Ediacaran Palaios explosion. Cambrian the of onset Nova driven ecologically rapid, indicate boundary nrsigporiferan. encrusting squeeze. a record fossil spicules. Precambrian sponge missing siliceous 200-myr of a suggest microRNAs and clocks, molecular 1097–1098. pp. 2002), in 1882” mann, rtrzi aaGop(.5053M) Namibia. Ma), 550–543 (c. Group Nama proterozoic Australia. South (2018). 1685–1686 animals. of rise early an 95 (2019). 49–58 31, 1–3 (1988). 111–135 3, il Rev. Biol. n 33 and daaa eaonresfo h aaGop Namibia. Group, Nama the from reefs metazoan Ediacaran al., et osbeaia-oyfsisi r-aionlmsoe from pre-Marinoan in fossils animal-body Possible al., et .cooperi F. Lethaia ytm Porifera Systema e ihrslto g aafo h Ediacaran–Cambrian the from data age high-resolution New al., et a.Geosci. Nat. rc il Sci. Biol. Proc. uaiesog imresi nclua hzraquestion Rhizaria unicellular in biomarkers sponge Putative al., et imnrlzto ypril taheti al animals. early in attachment particle by Biomineralization al., et IIotusdt naporeayrwiaefiefor- file image raw proprietary a in data outputs GIRI × 7–04(2014). 972–1004 89, and perforata, A. rc il Sci. Biol. Proc. 74 (2008). 37–45 41, 3for 33 aueEo.Evol. Ecol. Nature m .Sci. J. Am. n,det aitosi mg ult through- quality image in variations to due and, 75–76 (2019). 17659–17665 116, 5–5 (2010). 653–659 3, 0580(1818). 20151860 282, .cooperi F. .N .Hoe,R .M a os,Es (Springer, Eds. Soest, Van M. W. R. Hooper, A. N. J. , dr irba–eaonreso h terminal the of reefs Microbial–metazoan oder, ¨ 0798(2018). 20171938 285, 6–9 (1989). 261–294 290, 43 (2010). 24–36 8, 7–8 (2019). 577–581 3, Geology a xrce n hnsoe as stored then and extracted was ) ovltoa neural convolutional a cooperi, F. rc al cd c.U.S.A. Sci. Acad. Natl. Proc. and perforata, A. Paleobiology Geology 013/4471(2020). 10.1130/G47447.1 , el Mag. Geol. × 5–7 (2001). 159–171 27, 8–8 (2019). 380–384 47, Cloudina 1frsml and B sample for 11 9–1 (2005). 499–517 142, aueEo.Evol. Ecol. Nature .cooperi F. grgtsare aggregates Paleobiology E2519– 115, pixels , Science Terra. A. 2, 4 .P otn,L .Mi,Erysog vlto:Arve n phylogenetic and review A evolution: sponge Early Muir, biomineralized A. L. modular Biostratigraphic Proterozoic Botting, Dickson, Kaufman, P. D. J. A. J. J. 24. A. Grotzinger, P. J. Saylor, Wood, A. Z. R. B. 23. Bowring, A. S. Grotzinger, P. J. 22. Schr S. Adams, W. E. 21. 5 .Lo .Za,Z a,Tefis eoto axi pnefo h Cambrian the from sponge vauxiid a of report first The Han, Z. Zhao, F. Luo, C. 25. 9 .U Riisg U. H. Riisg U. 29. H. Larsen, and S. P. sponges 28. of biology hidden The Haddock, D. H. S. Leys, P. S. Dunn, W. C. 26. 2 .Bsk .H nl,A .Ptof h enn fstromatolites. of meaning The Petroff, P. A. Knoll, H. A. Bosak, T. 32. Senowbari-Daryan, B. Rigby, Keith J. 31. Leys P. S. 30. Hammel U. J. 27. 9 .Hlern,P R P. Hildebrand, T. 39. Achanta R. Rocknest 38. the Senowbari-Daryan, of B. thrombolites Rigby, Ga) Keith (1.9 J. Proterozoic Early 37. Grotzinger, P. J. Kah, C. L. 36. Amthor E. J. 35. Neoprotero- in assemblage Namacalathus-Cloudina Mountjoy, W. E. Hofmann, J. H. 34. Awramik, M. S. Shapiro, S. R. 33. 0 .R K. 40. .Mnu,adE emn etakJ tas o edako the on Invertebrate feedback Tuttle Grant Princeton the Sciences for from Earth funding NSF thought- Fund. Strauss by by and their supported Maloof J. for was A. to work reviewers 1028768 thank This anonymous input. two We and and critique Geyman. ful Howes, Knoll A. B. E. Getraer, and A. and manuscript with of Manzuk, discussions sample from R. benefited a analyses with phological us permis- gave kindly provided the Erwin analyze D. to the destructively time while to study, in his sion for dedicated instrumental Florence specimens invaluable M. Inozoan were provided Smithsonian, locate the Samuels, Bartolucci help At R. B. field. and the sample especially Hart in Tasistro of assistance but A. reconstruction, GIRI. Studio, early of Situ an development created of Mason All T. grind- with B. performed and us Ren assisted imaging, B. and MIT, and Namibia At Schneider ing in respectively. G. process, working permitting for Namibia, export permits of the us granted Survey Eiseb Geological J. the and At Farm. at Driedoornvlakte repository public a ACKNOWLEDGMENTS. in located is paper this in that data https://github.com/giriprinceton/namapoikia. process includes to and used void) code that or sphere partition largest (i.e., the object at of the calculates, diameter in voxel. 39, the contained ref. voxel), is by (or both mod- defined pixel The as Avizo. volumetric thickness within each local module implements Map which values Thickness thickness ule, the particular, using In datasets. generated visu- volumetric were for of designed These package analysis TIFFs. software and a classified Avizo, alization into create loaded to were then thresholded files were TIFF then which maps, ability cooperi network improving of intent the with accuracy. momentum) with descent gradient l a mg aaaeaalbeuo eus.Tecmuainlsource computational The request. upon available are data image raw All B, sample A, sample of images training, Following framework. Namibia. Group, Nama the from metazoan evolution. animal early on (1995). constraints geochronologic and plat- carbonate isolated Namibia). microbial-dominated, Group, (2004). Nama a Proterozoic, of (terminal form evolution stratigraphic and hnjagBiota. Chengjiang edn,esneo urn knowledge. current of essence feeding, morpho- and microtomography. reconstruction X-ray 3D and (2012). by casting corrosion revealed using system metrics aquiferous sponge leuconoid ctenophores. h pnebd plan. body sponge the 4–5 (1996). 347–355 Mexico. 70, New Mountains, Guadalupe , Capitan sponge, ftikesi he-iesoa images. three-dimensional in thickness of Intell. Mach. Anal. Pattern Trans. IEEE Tunisia Tebaga Djebel from Sponges Hexactinellid Canada. Territories, Northwest Formation, Oman. in boundary Cambrian fossils. shelly oldest Canada’s Columbia: British Geology Formation), (Byng Group Miette zoic thrombolite. time-restricted dispersed, Sci. Planet .N .Hoe,R .M a os,Es Srne,20) p 275–278. pp. 2002), (Springer, Eds. Soest, Van M. W. R. Hooper, A. N. J. tlr .Vclt Fml cnhcattdeFshr 90 in 1970” Fischer, Acanthochaetetidae “Family Vacelet, J. utzler, ¨ eerntruhterrsetv erlntokt rdc prob- produce to network neural respective their through run were 0119 (2001). 1091–1094 29, h pnepm:Terl fcretidcdflwi h einof design the in flow induced current of role The pump: sponge The al., et r,P .Lre,Cmaaieeohsooyo ciezoeti filter zoobenthic active of ecophysiology Comparative Larsen, S. P. ard, 14 (2013). 21–44 41, ˚ LCspriescmae osaeo-h-r uepxlmethods. superpixel state-of-the-art to compared superpixels SLIC al., et Palaeoworld xicino luiaadNmcltu ttePrecambrian- the at Namacalathus and Cloudina of Extinction al., et rnsEo.Evol. Ecol. Trends h o-irrhcl o-nfrl rnhn oooyo a of topology branching non-uniformly non-hierarchical, The al., et esge,Anwmto o h oe-needn assessment model-independent the for method new A uegsegger, ¨ .Paleontol. J. dr .P rtigr .S comc,Dgtlreconstruction Digital McCormick, S. D. Grotzinger, P. J. oder, ¨ d h pnepump. sponge The ad, ˚ lSOne PloS etakC uslanfrgatn sacs to access us granting for Husselmann C. thank We –9(2018). 1–29 27, aoaaei cooperi Favosamaceria Geology 83 (2020). 28–33 94, 8–9 (2015). 282–291 30, 277(2011). e27787 6, .perforata A. .cooperi F. e eu,telretknown largest the genus, new Gigantospongia, 2428 (2012). 2274–2282 34, .Paleontol. J. pe ema nzi,Dmsogd and Demospongid, Inozoid, Permian Upper 3–3 (2003). 431–434 31, .SaRes. Sea J. .Microsc. J. Palaios Science .Ter Biol. Theor. J. o eosrcin u mor- Our reconstruction. for ape .Saiogenerously Shapiro R. sample. 0–1 (1992). 305–315 7, 6–9 (2000). 169–193 44, Uiest fKna,1995). Kansas, of (University 3328 (2002). 2383–2386 296, 1–2 (2006). 411–422 80, NSLts Articles Latest PNAS 77 (1997). 67–75 185, e ru n om widely A form: and group new .Sdmn.Res. Sediment. J. 36 (1994). 53–63 168, and perforata, A. caZool. Acta 598–604 Science270, nu e.Earth Rev. Annu. ytm Porifera, Systema 479–497 74, .Paleontol. J. 160–170 93, | f7 of 7 F.

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES