lable at ScienceDirect

Journal of South American Earth Sciences 81 (2018) 66e77

Contents lists avai

Journal of South American Earth Sciences

journal homepage: www.elsevier .com/locate/ jsames

Provenance analysis of the Pliocene Ware Formation in the GuajiraPeninsula, northern Colombia: Paleodrainage implications

Nicol�as P�erez-Consuegra a, b, *, Mauricio Parra a, Carlos Jaramillo b, Daniele Silvestro c, d,Sebasti�an Echeverri a, e, Camilo Montes f, g, Jos�e María Jaramillo h, Jaime Escobar b, i

a Institute of Energy and Environment, University of Sao Paulo, 055108-010 Sao Paulo, SP, Brazilb Smithsonian Tropical Research Institute, Box 0843-03092, Balboa, Anc�on, Panamac Department of Environmental Sciences, University of Gothenburg, Gothenburg, Swedend Gothenburg Global Biodiversity Centre, Gothenburg, Swedene Instituto de Geociencias, Universidade de S~ao Paulo, Rua do Lago 562, 05508-080 S~ao Paulo, SP, Brazilf Department of Geological Sciences, Yachay Tech University, San Miguel de Urcuquí, Ecuadorg Departamento de Geociencias, Universidad de los Andes, Bogot�a, Colombiah GMAS, Calle 62 # 3-24, Bogot�a DC, Colombiai Universidad del Norte, Km 5 Vía Puerto Colombia, Barranquilla, Colombia

a r t i c l e i n f o

Article history:Received 5 July 2017Received in revised form2 November 2017Accepted 2 November 2017Available online 8 November 2017

Keywords:ProvenancePlioceneColombiaPaleogeographyGuajira PeninsulaDetrital geochronologyHeavy minerals

* Corresponding author. Current affiliation: Departacuse University, Syracuse, NY 13244, USA.

E-mail address: nperezco@syr.edu (N. P�erez-Consu

https://doi.org/10.1016/j.jsames.2017.11.0020895-9811/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

The Cocinetas Basin in the Guajira Peninsula, the northernmost tip of South America, today has a dryclimate with low rainfall (<500 mm/yr), a long dry season (>ten months) and no year-long rivers orpermanent standing bodies of fresh water. In contrast, the fossil and geological record indicate that theCocinetas Basin was much wetter during the Miocene-Pliocene (~17e2.8 Ma). Water needed to sustainthe paleofauna could either have originated from local sources or been brought by a larger river system(e.g. proto Magdalena/Orinoco river) with headwaters either in Andean ranges or the Guyana shield. Wepresent a provenance study of the Pliocene Ware Formation, using petrographic analysis of conglomerateclasts and heavy minerals, and U-Pb dating of 140 detrital zircons. Clasts and heavy minerals are typicalof ensialic metamorphic and igneous sources. The detrital zircon age distribution indicates the Guajiraranges as the most probable sediment source. The overall results indicate that the fluvial system of theWare Formation drained the surrounding ranges. The water was probably derived by local precipitationonto the Guajira peninsula.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Northern South America has a long history of mountain buildingand sedimentary basin formation driven by the interaction amongthe Caribbean, Nazca and South American plates (e.g., Pindell et al.,2005). The Cocinetas Basin, located in the Guajira Peninsula (Figs. 1and 2), records an Eocene to Pliocene sedimentary sequence thatcould help us understand the landscape evolution of northwesternSouth America (Jaramillo et al., 2015). This region has a very dryclimate with low rainfall (<500 mm/yr) and a very long dry season(>ten months), and it lacks all-year-long rivers or standing bodiesof fresh water (e.g., Pab�on-Caicedo et al., 2001). In contrast, the

ment of Earth Sciences, Syr-

egra).

fossil and geological records indicate that the Cocinetas Basin wasmuch wetter during the Miocene-Pliocene (~17e2.8 Ma) (Aguileraet al., 2013; Cadena and Jaramillo, 2014; Forasiepi et al., 2014;Hendy et al., 2015; Jaramillo et al., 2015; Moreno et al., 2015;Amson et al., 2016; Carrillo-Brice~no et al., 2016; Moreno-Bernalet al., 2016; Suarez et al., 2016; Aguirre-Fern�andez et al., 2017;P�erez et al., 2017).

The Guajira Peninsula is currently isolated from the major riversystems of northern South America (e.g. Magdalena, Amazon andOrinoco rivers) by its boundingmountain ranges, which include theSierra Nevada de Santa Marta (SNSM), the Perij�a Range and theM�erida Andes (Fig. 1A). Water supplies needed to sustain theCocinetas Miocene-Pliocene fauna could either have originatedfrom local precipitation and rivers from the Guajira ranges (Figs. 1and 2) or have been brought by a larger river system, e.g. protoMagdalena or proto Orinoco rivers, with headwaters either in

Fig. 1. Topography of northern South America showing major mountain ranges and possible sediment sources including Central and Eastern cordilleras, Sierra Nevada de SantaMarta (SNSM), Perij�a range, M�erida Andes and Guiana Craton. Dashed line represents the shoreline during glacial times. (A) Yellow lines indicate drainage divides for the Mag-dalena, Orinoco and Amazon basins, and blue lines indicate main modern river systems. (B) Possible scenarios for freshwater fish dispersal in the Pliocene. (1) Proto-Orinoco riverdrains into the Caribbean through a lowland depression in the M�erida Andes. (2) The proto Magdalena River drains into the Maracaibo region. This proto Magdalena would beconnected to cis-Andean drainages through an opening in the southern Eastern Cordillera. (3) The freshwater plume of a proto Orinoco river delta would have allowed fish tomigrate into the Maracaibo region (Guajira and Falc�on). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e7768

Andean ranges or in the Guiana Craton, respectively (Fig. 1), asetting similar to themodern Nile river, which has its headwaters inthe humid mountains of Ethiopia and crosses the Sahara Desert,bringing water and humidity into its delta in the MediterraneanSea, hundreds of kilometers away (e.g., Garzanti et al., 2006;Dumont, 2009). Pliocene fossil fish found in Cocinetas have affin-ities to modern faunas of the Amazon-Orinoco river system(Aguilera et al., 2013), suggesting a riverine connection betweenthe Cocinetas/Falc�on basin and the Amazon-Orinoco drainage basinduring the Pliocene (Aguilera et al., 2013). In contrast, geologicaldata indicate that the M�erida Andes had already been uplifted bythe Pliocene, separating drainages on the two sides of the range(e.g., Parnaud et al., 1995; Audemard and Audemard, 2002;Audemard M., 2003; Bermúdez et al., 2010). The Perij�a range,SNSM and Guajira ranges also had been uplifted by the Pliocene(e.g., Kellogg, 1984; Zapata et al., 2009; Bayona et al., 2011; Cardonaet al., 2011a; Ayala et al., 2012) (Figs. 1 and 2).

Was the water needed to sustain the Cocinetas Pliocene faunalocally derived or transported to the basin from a long distance?Was there a fluvial connection between the Cocinetas Basin duringthe Pliocene and cis-Andean (e.g. drainages east of the Andes)drainages? In order to answer these questions, we performed aprovenance analysis of the fluvial strata of the Pliocene Ware For-mation (3.2e2.8 Ma). We performed a petrographic analysis ofheavy minerals and conglomerate clasts and dated 140 detritalzircons using U-Pb e Laser-ICPMS. If sediments of the Ware For-mation were locally derived, we would expect textural immaturityand Jurassic (200-150 Ma; Cardona-Molina et al., 2006), Permo-Triassic (300-250 Ma), 500-400 Ma (“Proto andean”; Weber et al.,2010) and Grenvillian (1250-850 Ma) (Cordani et al., 2005;Cardona-Molina et al., 2006) detrital zircon ages, as these are theages of the exposed basement in the peninsula. If sediments hadbeen transported for a long distance, we would expect texturalmaturity and either Cretaceous (~90-70 Ma) detrital zircon ages ifthe source is the Central/Eastern cordilleras (Villag�omez et al.,2011; Pepper et al., 2016; Jaramillo et al., 2017), or 2.5e1.5 Gaages if the source is the Craton (Gaudette et al., 1978; Priem et al.,1982; Gaudette and Olszewski, 1985; Santos et al., 2003; Ibanez-Mejia et al., 2011; Bonilla-P�erez et al., 2013; Reis et al., 2013) (Fig. 2).

1.1. Geology of the Guajira Peninsula

The Guajira Peninsula comprises three small mountain ranges(Cocinas, Jarara and Macuira) surrounded by Cenozoic sedimentarybasins (Fig. 2). These ranges are formed by three lithostratigraphicbelts that include, from south to north (Fig. 2): (1) a weaklydeformed Mesozoic volcano-sedimentary sequence similar to thesoutheastern SNSM foothills and Perij�a Range (MacDonald, 1964);(2) a Grenvillian and Late Paleozoic to Triassic metamorphicsequence (Alvarez, 1967; Cordani et al., 2005; Cardona-Molinaet al., 2006; Weber et al., 2010) intruded by Upper Jurassic plu-tons (Cardona-Molina et al., 2006); and (3) a belt of Upper Creta-ceous greenschists intercalated with mafic-ultramafic plutonicrocks (Alvarez, 1967; Weber et al., 2010). In the Cabo de la Velaregion, there are Cretaceous mafic and ultramafic rocks that cropout as isolated remnants (Fig. 2; Alvarez, 1967; Weber et al., 2009).

During the Eocene, the interaction between the Caribbean andSouth American plates caused strike-slip displacement of the Cuisafault (Fig. 2; Hern�andez et al., 2003) and normal faulting of blockswhich led to the uplift of the Guajira Peninsula ranges and theformation of the pull-apart Cocinetas Basin (Zapata et al., 2014;Moreno et al., 2015). The youngest strata of the Cocinetas Basincorrespond to the 25 m thick, late Pliocene (3.5e2.8 Ma), WareFormation (Hendy et al., 2015; Moreno et al., 2015), which consistsof conglomerates at the base, coarse sandstones and mudstones in

the middle and a coquina at the top. The Ware Formation wasdeposited in a fluvio-deltaic setting (Hendy et al., 2015; Morenoet al., 2015).

2. Methods

Several samples were collected for a variety of analyses. Weused six-digit identification numbers (IDs) to label field localities.Each individual rock collected also has a unique ID number from theSmithsonian Tropical Research Institute (STRI), together with thelocality ID (Table S1, SOM). Information on the stratigraphic sec-tions, locality, and sample ID data can be consulted from Morenoet al. (2015). We did conglomerate clast counting at four differentlocalities (100 clasts per locality; Table S1 and S2, SOM). Sampleswere collected from basal conglomerates at four localities of theWare Formation: Flor de la Guajira (STRI locality 290080; N11.8801 W �71.4772), Yotojoro area (STRI locality 290839; N11.9471 W �71.3085), Bahia Cocinas (STRI locality 640006, N11.8314 W �71.3853) and Puerto Lopez (STRI locality 290836; N11.9587 W �71.2399). We analyzed eight samples for heavy min-erals petrography (Table S1 and S3, SOM). Six of these samples werecollected from different stratigraphic levels of the Ware Formationstratotype section (STRI locality 390075; N 11.8487 W �71.3243).The two remaining samples were collected from a basal conglom-eratic sandstone in the Flor de la Guajira locality and from amodernriver sandy channel (STRI locality 640008; N 11.9572 W �71.3032).Two samples from the Ware stratotype section were used to obtainzircons for U-Pb dating (Table S1, S4, SOM).

2.1. Conglomerate clast counts

A total of 100 clasts were collected at each locality using theribbon method (Howard, 1993). This technique consists in countingindividual clasts within a specified area. The area must have di-mensions significantly larger (5�) than the longest axis of thelargest clast in the sample (Howard, 1993). Clasts <2 cm in sizewere excluded from the analysis. Clasts were observed and iden-tified under a binocular microscope. Thin sections of some of theclasts were done to aid in the identification. Clasts were classifiedinto seven categories including quartzite, mica-schist, quartz are-nites, lithic arenites, mudstone, micrite and sparitic limestone.

2.2. Heavy mineral analysis

Heavy mineral separation was carried out by first sieving 200 gof sample into the 63e125 mm fraction. Samples were subsequentlydried and passed through 2.9 g/cm3 Methyl iodide (diluted) toseparate the lighter minerals from the heavier suite. Heavy mineralanalysis was carried out through optical microscopy identificationof individual grains. The ribbon method was used for counting 300grains per sample (Mange and Maurer, 1992). Heavy minerals werecategorized into three tectonic domains following Feo-Codecido(1956) and Pettijohn et al. (1987): (1) Magmatic arc domain:apatite, biotite, hornblende, sphene; (2) Ultramafic and maficdomain: augite, serpentine, olivine, picotite, pleonaste, spinel anddiopside; (3) Ensialic metamorphic domain: garnet, staurolite,sillimanite, kyanite and andalusite.

2.3. U-Pb detrital geochronology

We dated 145 zircons and obtained 140 concordant ages fromtwo samples (STRI 39929 [n ¼ 68] and 38930 [n ¼ 72]; Table S4,SOM) at two different stratigraphic levels of the lower PlioceneWare Formation. Zircon crystals were extracted fromheavymineralsamples using the Franz magnetic separator and 3.4 g/cm3 Methyl

Fig. 2. Geological map of the Guajira Peninsula with sample localities (modified G�omez et al., 2015b; Irving, 1972) and major faults (Cuisa and MF: Macuira fault). Topography of theGuajira Peninsula in the upper right corner, with the location of the Macuira, Jarara and Cocinas ranges and the Cocinetas Basin. Lithology: met: Metamorphic; sed: Sedimentary;volc: Volcanic and intr: Intrusive.

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e77 69

iodide. Zircon grains were mounted into epoxy resin discs andobserved under transmitted light (TL) and cathodoluminescence(CL) (Figure S1, SOM). UePb age determinations by laser-ablationmulticollector inductively coupled plasma mass spectrometer(LAICPMS) were carried out at the Institute of Geosciences, Uni-versity of S~ao Paulo. The choice of spot sites was guided by CLimaging. The U-Pb isotope analyses were performed on zircongrains using a Thermo-Fisher Neptune laser-ablation multicollectorinductively coupled plasma mass spectrometer equipped with a193 Photon laser system. The analytical routine followed a pro-cedure that starts with three measurements of the Gem zircon (GJ-1) standard, followed by measurements of two blanks and twoNational Institute of Standards and Technology (NIST) 612 stan-dards. After these steps, a sequence of 13 analyses of unknownzircons was performed, then two more analyses of GJ-1, two blanksand two NIST 612. These raw data were then corrected for back-ground, instrumental mass bias drift and common Pb (Siqueiraet al., 2014). Grain ages were filtered using standard methodolo-gies (Gehrels et al., 2008). Due to the magnitude of uncertaintiesinherent to U-Pb ages, a cut-off of 1000Mawas chosen for selectingthe 206 Pb/238U age in younger grains and the 206Pb/207 Pb in

older grains. In U-Pb ages younger than 300 Ma, the discordance orinverse discordance was not evaluated, whereas for grains olderthan 300Ma, those with discordances greater than 25% and inversediscordance greater than 10%, or error greater than 10%, were dis-carded. The interpretations on the detrital zircon ages are based inidentifying populations with at least three grains to prevent overinterpreting outliers that may arise from Pb loss.

We created a dataset of 228 published U-Pb zircon ages frombedrock samples of six regions that could have been sedimentsources for the Ware Formation sediments, including the Centraland Eastern cordilleras, the M�erida Andes, the SNSM, the Guajiraranges, and the Guiana Craton (Fig. 1 and Table S5, SOM). This dataset includes 160 U-Pb ages compiled by G�omez et al. (2015a) and 68U-Pb ages compiled in this study (Table S5, SOM) (Gaudette et al.,1978; Priem et al., 1982; Gaudette and Olszewski, 1985; D€orret al., 1995; Restrepo, 1995; Restrepo-Pace et al., 1997; SchneiderSantos et al., 2000; Ordo~nez-Carmona, 2001; Cardona-Molina,2003; Jimenez Mejia, 2003; Cordani et al., 2005; Correa et al.,2006; Cardona-Molina et al., 2006; Vinasco et al., 2006; Iba~nez-Mejia et al., 2007; Mejía Herrera et al., 2008; García et al., 2009;Mantilla F. et al., 2009; Duque, 2010; Giraldo-Arroyave, 2010;

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e7770

Horton et al., 2010; Jim�enez, 2010; Bustamante et al., 2010;Restrepo-Pace and Cediel, 2010; Cardona et al., 2010, 2011b; Weberet al., 2010, 2014; Ibanez-Mejia et al., 2011; Leal-Mejía, 2011;Mantilla Figueroa et al., 2011, 2012, 2013; Restrepo et al., 2011;Villag�omez et al., 2011; Bayona et al., 2012; Bonilla-P�erez et al.,2013; Cochrane, 2013; Reis et al., 2013; van der Lelij, 2013; Cuadroset al., 2014; Baquero et al., 2015; Zuluaga et al., 2015; van der Lelijet al., 2016; Jaramillo et al., 2017).

We decided to compare the detrital ages of the Ware Formationwith bedrock U-Pb samples, because modern-river data is notavailable from all of the possible sources (i.e., mountain ranges).There are published detrital zircon U-Pb ages for 12 modern riversin the region of interest (Goldstein et al., 1997; Bande et al., 2012;Nie et al., 2012; Pepper et al., 2016). Only eight of these riverswere sampled at points that drain catchments of a single sourceregion including one detrital sample from a river that drains thenorthern Central Cordillera (Sinu river; Pepper et al., 2016), threedetrital samples draining the eastern flank of the Central Cordillera(Nie et al., 2012), two detrital samples draining the western flank ofthe Eastern Cordillera (Nie et al., 2012) and two detrital samplesdraining the eastern flank of the Eastern Cordillera (Bande et al.,2012). Four detrital samples belong to catchments that containsseveral of our defined sources and thus are not suitable for theanalysis: three detrital samples come from the Magdalena river ator near the delta (Pepper et al., 2016) and one detrital samplecomes from the Orinoco river (Goldstein et al., 1997). The detritalsamples from the Magdalena river are within a catchment thatincludes the Sierra Nevada de Santa Marta and the Central andEastern cordilleras. The detrital sample from the Orinoco river iswithin a catchment that includes the Merida Andes, Guiana Cratonand the Eastern Cordillera. Furthermore, there are no detritalmodern river samples from catchments of three of our possiblesource areas including Guajira ranges, Sierra Nevada de SantaMartaand the Merida Andes. Using a probabilistic analysis with anincomplete set of sources is not informative for our researchquestion.

We developed a probabilistic framework to infer the provenanceof a set of zircons of unknown origin given a set of potential sourceregions (i.e., differentmountain ranges). A sample is defined here asthe radiometric age of an individual zircon crystal. The term sourcerefers to the mountain range from which a zircon could have beenderived.

The main outputs of this analysis are as follows: (1) The likeli-hood of a set of samples (i.e., detrital zircon ages from a sedimen-tary unit, in this case the Ware Formation) given each sourceregion. Likelihoods can be directly compared to assess which sourcebest explains the distribution of ages measured in the analyzed setof samples. (2) The relative probability for each sample to haveoriginated in each of the possible sources. (3) The identification ofsamples that are not adequately explained by any of the samples inthe range of sources. (4) The likelihood of a mixed model that al-lows each sample to have a different origin. This model does notassign each sample to a source, but rather it integrates out theuncertainties around the provenance of the samples and estimatesthe relative contribution of each source to the analyzed set ofsamples. All inferences about the age ranges explicitly incorporatethe uncertainty around the dating of each sample (error for theindividual age of a zircon crystal). The approach is implemented inProvenanceFinder, an R script available at: https://github.com/dsilvestro/ProvenanceFinder.

Each sample (either from the Ware Formation or one of thesources) is assigned a temporal range defined by a minimum andmaximum age (amin, amax), which reflects the dating uncertainty.We consider the minimum and maximum ages of a sample as the0.025 and the 0.975 quantiles of a normal distribution, so that its

true age falls within the temporal range with 95% probability. Thus,the probability density of the age of a sample is modeled by anormal distribution with mean equal to the midpoint of the rangemx ¼ (amax e amin)/2 and standard deviation set to sx ¼ (amax e

amin)/(2 � 1.96).We indicate the N samples of unknown origin as Z ¼ {z1,…, zN},

and the M samples within a source as S ¼ {s1, …, sM}. The functiong(t) describing the probability of obtaining a sample of age t thatoriginated from a source S (Fig. 3A and B) derives from the sum ofthe normal densities associated with each of the samples from thesource S:

gðtÞ ¼ 1Mþ 1

ðXMi¼1

½fðt;mi;siÞ� þ yðt;Tmin; TmaxÞ (1)

where f(mi, si) is the probability density function of a normal dis-tribution with mean mi and standard deviation si, describing theage range of sample i. To account for the possibility that the sam-ples within a source do not represent the full set of ages charac-terizing the source, we add a uniform distribution with densityy(Tmin, Tmax), with arbitrary boundaries encompassing the entiretime range of interest. This uniform distribution provides a back-ground probability that can help us to identify samples with un-known source (see below) and to avoid numerical problems due tonear-zero probabilities. The function g(t) is normalized by dividingby the sum of the integrals of the normal and uniform densities (i.e.M þ 1).

The probability that a sample zi (e.g., Ware Formation detritalzircon) originated from S is calculated by integrating the productbetween the function g(t) describing the samples in S (Eq. (1)) andthe normal density modeling the age of the sample zi (Fig. 3C andD):

PðzjSÞ ¼ZTmax

Tmin

gðtÞ � fðmz;szÞdt (2)

We emphasize that this approach enables us to infer the prov-enance probabilities, while explicitly incorporating dating un-certainties associated to both the unknown sample and the set ofsamples included in the source. Based on Eq. (2), we can computethe likelihood of a set of samples Z given a source S as:

PðZjSÞ ¼YNi¼1

PðzijSÞ (3)

The likelihoods of the set of samples of unknown origin condi-tional on different sources are then compared to assess which ofthe different sources (e.g., mountain ranges) is more likely to havegenerated the samples.

After calculating the likelihood of a sample z conditional of allavailable sources using Eq. (2), we compute the relative probabilityof origin from each source. For example, the relative probabilitythat a sample z (a single detrital zircon) originated in source Sj(where j represents different sources regions e.g., Merida Andes) isgiven by:

relP�zjSj

� ¼ P�z��Sj�PK

i¼1P�z��Sj� (4)

where K is the number of available sources.Based on Eqs. (1) and (2), the probability of a sample z will be

similar to the background probability, if there is negligible temporaloverlap between z and the samples in the source Sj. In this case, the

Fig. 3. Visualization of the probability analysis. (A) and (B) show the likelihood surfaces or Probability Density functions (black lines) of two potential sources, each with three zirconages (samples). The age ranges of the zircons (which define the width of the normal densities combined in the likelihood surface) are shown as blue bars at the bottom of the plots.The red dashed line shows the Probability Density Function describing the estimated age of a zircon z of unknown provenance. The product of the source (black) and the sample(red) Probability Density functions is displayed in (C) and (D) (blue shaded areas). The likelihood of z conditional on a source is computed by integrating the functions displayed in(C) and (D). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e77 71

contribution of the normal densities is close to zero:

XMi¼1

½fðz;mi; siÞ�z0

and therefore the probability of the sample z is mostly determinedby the density of the background uniform distribution y(Tmin,Tmax):

PðzjSÞz 1M þ 1

ZTmax

Tmin

vðt; Tmin; TmaxÞdt (5)

Here we consider as unidentifiable the origin of a sample with aprobability that differs by less than 5% from the backgroundprobability (Eq. (5)), because none of the source samples are suf-ficiently similar to explain its age.

We implemented a mixedmodel to allow samples in the set Z tohave different sources of origin. Although we can use the relativeprobabilities of the samples (Eq. (4)) to assess themost likely sourcefor each sample, we expect to observe quite frequently substantialuncertainties around the assignment of a sample to a particularsource. These uncertainties are reflected in similar relative

probabilities for different sources. The aim of the mixed model istherefore to calculate the likelihood of the entire set of sampleswhile accounting for the possibility that samples have differentprovenances, but without assigning the samples to any particularsource. Instead, we calculate the probability that a sample origi-nated from any of the sources, thus integrating out the un-certainties. This probability is equal to a sum of probabilitiesweighted by a vector of parameters w ¼ {w1, …, wK}:

PðzjS1;…; SK ;w1;…;wKÞ ¼XKj¼1

�P�zjSJ

�

� wj�; where

XKj¼1

wj

¼ 1 (6)

The likelihood of the set of samples Z is computed as a productof probabilities over all samples (as in Eq. (3)), assuming the sameweights across all samples. The vector of weights w is optimized tomaximize the likelihood of the entire set of samples and it thusexpresses the relative contribution of each source to the set ofsamples Z. In other words, the estimated weights indicate theproportions of samples that are inferred to have originated from

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e7772

each source.We implemented an additional test to verify the effect of sample

size on the robustness of the results. We used a Jackknife sub-sampling algorithm to randomly reduce the size of both the WareFormation data and data of all the potential sources. The number ofsamples kept in the Jackknife analysis matched that of the smallestdata set in the considered set (n ¼ 20). After subsampling, werepeated the likelihood calculation (Eqs (2) and (3)) and ranked thesources by their probability to explain the results from the WareFormation. We repeated this procedure 10,000 times andcomputed for each source the frequency of being selected as thebest region of origin of the detrital zircon ages.

3. Results

3.1. Petrography

Conglomerate clast samples (n ¼ 4) are dominated by quartzite(mean ¼ 61.2, SD ¼ 11.1), quartz arenites (mean ¼ 17.7, SD ¼ 3.8),micrite (mean¼ 11.7, SD¼ 7), lithic arenites (mean¼ 4.7, SD¼ 9.5),mica-schist (mean ¼ 3, SD ¼ 3), and very small amounts spariticlimestone fragments (mean ¼ 1, SD ¼ 2) and mudstones(mean ¼ 0.5, SD ¼ 1) (Fig. 4A and Table S2, SOM). Some quartziteclasts contained accessory garnet crystals.

Heavy minerals from theWare Formation (n¼ 7) are dominatedby epidote (mean ¼ 30.38%, SD ¼ 6.96%), zircon (mean ¼ 17.76%,SD ¼ 4.81%), garnet (mean ¼ 12.04%, SD ¼ 9.78%) and apatite(mean ¼ 8.76%, SD ¼ 5.75%) (Fig. 4B and Table S3, SOM). Themodern sample is dominated by epidote (33.6%) hornblende (33%),and garnet (16%) (Fig. 4B). Heavy minerals are mostly from anensialic metamorphic domain (mean ¼ 69.1%; SD ¼ 19.4%) andmagmatic arc domain (mean ¼ 30.8%; SD ¼ 19.4%) (Fig. 4C). Thereare no mafic or ultramafic heavy minerals such as augite, serpen-tine, olivine, picotite, pleonaste and diopside.

3.2. U-Pb geochronology

The age distribution of the 140 zircons of the Ware Formationshows Jurassic (200-150 Ma), Permo-Triassic (350-200 Ma), Cry-ogenianeCarboniferous (“proto Andean”; 800-350 Ma), Ectasian e

Cryogenian (“Grenvillian”; 1250e800 Ma) and OrosirianeEctasian(Mesoproterozoic and Paleoproterozoic; 2000e1250 Ma) pop-ulations (Fig. 5A). Only one zircon out of 140 has a Cretaceous age,therefore we do not consider it as a statistically significant popu-lation. Paleogene zircons are absent (Fig. 5 and Table S4, SOM).

The provenance analysis indicates that the Guajira ranges arethe single source that best explains the set of the Ware Formationzircons, with a log likelihood of �1198.361 (Table 1, Fig. 6). Thisvalue of log likelihood is 66.5 units higher than the next best fittingsource (i.e. SNSM), indicating that it is strongly preferred over theother sediment source alternatives (Table 1, Fig. 5, SOM). TheGuajira ranges were selected as the most likely source over 65% ofthe times with the use of the Jackknife technique. The relativeprobabilities show that there is substantial uncertainty around theexact assignment of samples to one particular source and only 3.6%of the samples could be attributed to a source with a certaintygreater than 95%. Furthermore, we identified four zircon ages (3%)that could not be adequately explained by any of the sampledsources (Table S6, SOM).

The mixed model obtained the best fit, with a log-likelihood126.1 units higher than that of the Guajira ranges alone. Based onthe weights estimated for each source, we infer that 33.6% of thesamples have a Guajira origin, whereas the provenance of theremaining 66.4% is scattered among the other sources in smallerproportions (Table 1). The SNSM source, which obtained the

second-best likelihood score among individual sources, seems tohave contributed as little as 6.8% of the Ware Formation samples.

4. Discussion

The conglomerate clasts are dominated by metamorphic andsedimentary (siliciclastic and calcareous) clasts, which suggestsmetamorphic sources as well as reworking of older sedimentaryunits. The composition of the heavy mineral suite suggests ensialicmetamorphic and igneous rock sources (Fig. 4B and C), typical ofthe eastern side of the Guajira ranges (Fig. 2) (MacDonald, 1964;Lockwood, 1965). The general high textural immaturity of theheavy mineral samples of the Ware Formation indicates that sedi-ments had low abrasion and short transport from nearby sources.The modern heavy mineral sample from an eastern Guajira catch-ment shows a similar distribution to the samples from the WareFormation except for the absence of zircon (Fig. 4B). Garnet andsillimanite are present in both the modern and Ware Formationsamples, indicating metamorphic sources (Feo-Codecido, 1956).The U-Pb age distribution of the Ware Formation detrital zircons(Fig. 5) also suggests that the most likely sediment source for theCocinetas Basin during the Pliocene was the Guajira ranges (Figs. 5and 6; log likelihood of �1198). A local source of sediments for thefluvial system of the Ware Formation suggests that the waterneeded to sustain the abundant paleofauna of the Cocinetas Basin(Moreno et al., 2015) was probably precipitated from water vaporcondensation above the Guajira peninsula, rather than brought by amajor river system from distant headwaters.

Aguilera et al. (2013) suggested, on the basis of fish fossil taxawith Orinoco/Amazon affinities found in the late Pliocene Ware(Cocinetas Basin) and San Gregorio (Falc�on Basin) formations(Fig. 1A), that a paleo-Orinoco could have flowed across the M�eridarange, connecting the Orinoco region with the Caribbean Falc�onand Cocinetas basins. We see no evidence, however, of texturallymature sediments indicating a long-distance transport of sedi-ments, and the zircon age distribution indicates that the WareFormation has a low probability of being derived from the Guianacraton or M�erida Andes (Figs. 5 and 6). Furthermore, paleogeo-graphic reconstructions indicate the M�erida Andes had beenuplifted enough to block the flow of the proto-Orinoco by the lateMiocene (~10 Ma) (e.g., Díaz de Gamero, 1996; Guzm�an and Fisher,2006; Escalona and Mann, 2011) or even since the Oligocene e

early Miocene (e.g., Rod, 1981; Rinc�on et al., 2014). The flow of theOrinoco river into the Atlantic Ocean since the Oligocene is sup-ported by provenance studies in Trinidad (Xie and Mann, 2014).

How to explain, then, the presence of Orinoco-derived fresh-water fish in the Cocinetas Basin if the Ware Formation fluvialsystem was local and mostly restricted to the Guajira ranges? Wepropose two hypotheses: (1) Pliocene marine currents could havepushed the freshwater plume of the Orinoco delta (Warne et al.,2002) to the west, allowing freshwater fish to continually movefrom the Orinoco drainage into Maracaibo drainages (Guajira andFalc�on; Fig. 1B). This setting could be similar to the modern coastalrivers in the Guiana region (e.g., Essequibo river), which are isolatedfrom the Amazon drainage basin by ~1200 km but hold severalspecies of freshwater fish that are also present in the Amazon riverdrainage (e.g., Arapaima gigas, Osteoglossum bicirrhosum, Pygocen-trus nattereri, Colomesus asellus and Phractocephalus hemioliopterus;(Hardman and Lundberg, 2006; Lima and Ribeiro, 2011). Fresh-water fish colonized Guiana either by migrating through the ma-rine currents that deflect northward the massive plume of theAmazon river delta or by the capture of a tributary river from theAmazon basin by a coastal river of Guiana (e.g., Cardoso andMontoya-Burgos, 2009). 2) A proto Magdalena could have flowedinto the Maracaibo region (Fig. 1B; Duque-Caro, 1990) and formed

Fig. 4. Summary of petrographic results. (A) Conglomerate clast counts of four samples from the Ware Formation (100 clasts counted per sample). (B) Heavy mineral counts of onemodern river sample and seven samples from the Ware Formation (300 grains per sample; results are normalized to 100). (C) Provenance domain discrimination of heavy mineralsof Ware. Categories include (1) Magmatic arc domain: apatite þ biotite þ hornblende þ sphene; (2) Mafic and ultramafic domain:augite þ serpentine þ olivine þ picotite þ pleonaste þ diopside; (3) Ensialic metamorphic domain: garnet þ staurolite þ sillimanite þ kyanite þ andalusite. (D) Pictures of selectedclast from the conglomerate samples. (E) Selected common heavy minerals.

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e77 73

an extensive coastal plain where multiple small rivers, includingthe Cocinetas Basin drainage, would have converged. This coastalplain would have expanded during the sea level drops of the Plio-cene (Fig.1B), some of which reached 50m (Miller et al., 2005). Thismodel would require the proto Magdalena River to be connected inthe south to the Amazon/Orinoco basins (Fig. 1B; Ujueta, 1999). Themodern Magdalena has a marked elbow near El Banco (N 8.97�, W73.86�) where it turns west, draining into the Momposinadepression and afterward into the Caribbean (Smith, 1986). Duringthe late Pliocene, the proto Magdalena River could have followed apath toward the northeast, flowing through the flat valley in be-tween the SNSM and Perij�a Range (Fig. 1B) and draining into theMaracaibo region (Fig. 1B; Duque-Caro, 1990). Post-Pliocene (<2.8Ma) deformation in the Cerrej�on thrust (e.g., Kellogg, 1984; Monteset al., 2010) or Oca Fault (e.g., AudemardM.,1996) would have tiltedthe valley, causing drainage reversal and blocking the incursion ofthe proto Magdalena in the Maracaibo region and changing its

course toward the west, leaving its sedimentary load in the lowerMagdalena Valley Basin (Bernal-Olaya et al., 2015) as in the modernsetting (Smith, 1986).

5. Conclusion

Our integrated provenance analysis indicates that sediments ofthe Pliocene Ware Formation were derived mostly from localsources within the Guajira Peninsula ranges. Precipitation abovethe Guajira Peninsula would have allowed a local fluvial system tothrive and sustain the diverse paleofauna recorded in the Cocinetasbasin.

Acknowledgments

This work was part of We thank GMAS Lab for sample prepa-rations and Miguel Basei at CPGeo for his help with the

Fig. 5. (A) Probability density function (PDF) plots created with ProvenanceFinder (R). PDF's were built using the bedrock U-Pb ages dataset for the six possible source areas (blacklines): Guajira ranges n ¼ 23, Sierra Nevada de Santa Marta n ¼ 27, Eastern Cordillera n ¼ 65, Merida Andes n ¼ 22, Central Cordillera n ¼ 71 and Guiana Craton n ¼ 20. Theprobability density function for the detrital zircon U-Pb ages of the Ware Formation n ¼ 140 (single ages) is shown in the background of each source plot (red dashed lines). (B)Likelihood of the Ware Formation samples (i.e., U-Pb detrital zircon ages) conditional on different potential sources (i.e., U-Pb bedrock zircon ages; Methods, Eqs (2) and (3)). Eachblue curve represents the product between the PDF of one of the sources with the PDF of the Ware Formation. The integral of the blue curve represents the probability that thedetrital zircon ages of the Ware Formation originated from a given source (Figure C, D). Vertical axis is in log-likelihood units. The plots are arbitrarily truncated at a log-likelihoodof �12 and thus display only the highest peaks (while the entire area was integrated to calculate the likelihood values reported in Table 1). (For interpretation of the references tocolour in this figure legend, the reader is referred to the web version of this article.)

Table 1Summary of the probability analysis. Log likelihoods are computed for each sourceand for the mixed model. The weights estimated under the mixed model indicatethe fraction of samples of the Ware Formation, which are likely to have originatedfrom each source.

Source Log likelihood Estimated weights

(mixed model)

Guajira ranges �1198.361 33.6%SNSM �1264.902 6.80%Eastern Cordillera �1279.95 15.4%Merida Andes �1332.285 14.8%Central Cordillera �1370.773 12.8%Guiana Craton �1408.765 16.3%Mixed model �1072.245

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e7774

geochronology. We thank Ana Maria G�oes, Paulo C. Giannini, Fed-erico Moreno, Sergio And�o, Andr�es Cardenas, Sebasti�an Zapata,German Bayona, Austin Hendy and Felipe Lamus for helpful dis-cussions and recommendations. We thank the team of paleontol-ogist and geologist who helped us in the field during the2013e2014 field trips to the Guajira Peninsula. NP was financed byFondo Corrigan ACGP-ARES. JE was partially funded by Colcienciasgrant 121566044585 “Causas y consecuencias de la aridificaci�on deecosistemas húmedos tropicales. Una mirada clim�atica y geol�ogicadurante el Neogeno de Suram�erica”. CJ was funded by the Smith-sonian Institution, the National Geographic Society, the AndersFoundation, 1923 Fund, Gregory D. and Jennifer Walston Johnson,and the National Science Foundation (Grant EAR 0957679). DSreceived funding from the Swedish Research Council (2015-04748)

Central

Craton

EC

Guajira

Merida

SNSM

Ware

Fig. 6. Likelihood distance plot. Likelihood of the sediments of the Ware Formation ofbeing derived from the different sources. Greater line thickness, darkness and prox-imity of the circles indicates higher probability.

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e77 75

and from the Knut and Alice Wallenberg Foundation. Field trips tothe Guajira Peninsula were sponsored by the Smithsonian TropicalResearch Institute, Universidad del Norte, Colombia, and Universityof Zurich. U-Pb analytical procedures at USP were financed byFAPESP JP Project 2013/30265-5 to MP. We thank Editor FranckAudemard, as well as two reviewers for their valuable comments.

Appendix A. Supplementary data

Supplementary data related to this article can be found athttps://doi.org/10.1016/j.jsames.2017.11.002.

References

Aguilera, O., Lundberg, J., Birindelli, J., Sabaj P�erez, M., Jaramillo, C., S�anchez-Villagra, M.R., 2013. Palaeontological evidence for the last temporal occurrenceof the ancient western amazonian river outflow into the caribbean. PLoS One 8,e76202. https://doi.org/10.1371/journal.pone.0076202.

Aguirre-Fern�andez, G., Carrillo-Brice~no, J.D., S�anchez, R., Amson, E., S�anchez-Villagra, M.R., 2017. Fossil cetaceans (mammalia, cetacea) from the neogene ofColombia and Venezuela. J. Mamm. Evol. 24, 71e90. https://doi.org/10.1007/s10914-016-9353-x.

Alvarez, W., 1967. Geology of the Simarua and Carpintero Areas, Guajira Peninsula,Colombia. PhD Thesis. Princeton University.

Amson, E., Carrillo, J.D., Jaramillo, C., 2016. Neogene sloth assemblages (mammalia,pilosa) of the Cocinetas basin (La Guajira, Colombia): implications for the greatamerican biotic interchange. Palaeontology 59, 563e582. https://doi.org/10.1111/pala.12244.

Audemard M., F.A., 1996. Paleoseismicity studies on the Oca-Anc�on fault system,northwestern Venezuela. Tectonophysics 259, 67e80. https://doi.org/10.1016/0040-1951(95)00144-1.

Audemard, F.E., Audemard, F.A., 2002. Structure of the M�erida Andes, Venezuela:relations with the south AmericaeCaribbean geodynamic interaction. Tecto-nophysics 345, 1e26. https://doi.org/10.1016/S0040-1951(01)00218-9.

Audemard M., F.A., 2003. Geomorphic and geologic evidence of ongoing uplift anddeformation in the M�erida Andes. Venezuela. Quat. Int. 101e102, 43e65.https://doi.org/10.1016/S1040-6182(02)00128-3.

Ayala, R.C., Bayona, G., Cardona, A., Ojeda, C., Montenegro, O.C., Montes, C.,Valencia, V., Jaramillo, C., 2012. The paleogene synorogenic succession in thenorthwestern Maracaibo block: tracking intraplate uplifts and changes insediment delivery systems. J. South Am. Earth Sci. 39, 93e111. https://doi.org/10.1016/j.jsames.2012.04.005.

Bande, A., Horton, B.K., Ramírez, J.C., Mora, A., Parra, M., Stockli, D.F., 2012. Clasticdeposition, provenance, and sequence of Andean thrusting in the frontalEastern Cordillera and Llanos foreland basin of Colombia. Bull. Geol. Soc. Am.124, 59e76. https://doi.org/10.1130/B30412.1.

Baquero, M., Grande, S., Urbani, F., Cordani, U., Hall, C., Armstrong, R., 2015. New

evidence for putumayo crust in the basement of the falcon basin and Guajirapeninsula, northwestern Venezuela. AAPG Mem. Pet. Geol. Potential Colomb.Caribb. Margin 108, 105e138. https://doi.org/10.1306/13531933M1083639.

Bayona, G., Cardona, A., Jaramillo, C., Mora, A., Montes, C., Valencia, V., Ayala, C.,Montenegro, O., Iba~nez-Mejia, M., 2012. Early paleogene magmatism in thenorthern Andes: insights on the effects of oceanic plateau-continent conver-gence. Earth Planet. Sci. Lett. 331e332, 97e111. https://doi.org/10.1016/j.epsl.2012.03.015.

Bayona, G., Montes, C., Cardona, A., Jaramillo, C., Ojeda, G., Valencia, V., Ayala-Calvo, C., 2011. Intraplate subsidence and basin filling adjacent to an oceanicarc-continent collision: a case from the southern Caribbean-South Americaplate margin. Basin Res. 23, 403e422. https://doi.org/10.1111/j.1365-2117.2010.00495.x.

Bermúdez, M.A., Kohn, B.P., Van Der Beek, P.A., Bernet, M., O'Sullivan, P.B.,Shagam, R., 2010. Spatial and temporal patterns of exhumation across thevenezuelan Andes: implications for cenozoic caribbean geodynamics. Tectonics29. https://doi.org/10.1029/2009TC002635.

Bernal-Olaya, R., Mann, P., Escalona, A., 2015. Cenozoic tectonostratigraphic evo-lution of the lower Magdalena basin, Colombia: an example of an under- tooverfilled forearc basin. AAPG Mem. 108, 345e397. https://doi.org/10.1306/13531943M1083645.

Bonilla-P�erez, A., Frantz, J.C., Char~ao-Marques, J., Cramer, T., Franco-Victoria, J.A.,Mulocher, E., Amaya-Perea, Z., 2013. Petrografía, geoquímica y geocronologíadel granito de Parguaza en Colombia. Bol. Geol. Colomb. 35, 83e104.

Bustamante, C., Cardona, A., Bayona, G., Mora, A., Valencia, V., Gehrels, G.,Vervoort, J., 2010. U-Pb la-ICP-MS geochronology and regional correlation ofmiddle jurassic intrusive rocks from the garz�on massif, upper Magdalena Valleyand central cordillera, southern Colombia. Bol. Geol. Colomb. 32, 93e109.

Cadena, E., Jaramillo, C., 2014. Early to middle Miocene turtles from the northern-most tip of South America: giant testudinids, chelids, and podocnemidids fromthe castilletes formation, Colombia. Ameghiniana 52, 188e203. https://doi.org/10.5710/AMGH.10.11.2014.2835.

Cardona-Molina, A., 2003. Correlaç~oes entre fragmentos do embasamentopreeMesozoico da terminaç~ao setentrional dos Andes Colombianos. com baseem dados isotopicos e geocronologicos. Ms Thesis. Universidade de Sao Paulo.

Cardona-Molina, A., Cordani, U.G., MacDonald, W.D., 2006. Tectonic correlations ofpre-Mesozoic crust from the northern termination of the Colombian Andes,Caribbean region. J. South Am. Earth Sci. 21, 337e354. https://doi.org/10.1016/j.jsames.2006.07.009.

Cardona, A., Valencia, V.A., Bayona, G., Duque, J., Ducea, M., Gehrels, G., Jaramillo, C.,Montes, C., Ojeda, G., Ruiz, J., 2011a. Early-subduction-related orogeny in thenorthern Andes: turonian to Eocene magmatic and provenance record in theSanta Marta massif and rancheria basin, northern Colombia. Terra Nov. 23,26e34. https://doi.org/10.1111/j.1365-3121.2010.00979.x.

Cardona, A., Valencia, V., Garz�on, A., Montes, C., Ojeda, G., Ruiz, J., Weber, M., 2010.Permian to Triassic I to S-type magmatic switch in the northeast Sierra Nevadade Santa Marta and adjacent regions, Colombian Caribbean: tectonic settingand implications within Pangea paleogeography. J. South Am. Earth Sci. 29,772e783. https://doi.org/10.1016/j.jsames.2009.12.005.

Cardona, A., Valencia, V., Weber, M., Duque, J., Montes, C., Ojeda, G., Reiners, P.,Domanik, K., Nicolescu, S., Villagomez, D., 2011b. Transient Cenozoic tectonicstages in the southern margin of the Caribbean plate: U-Th/He thermochro-nological constraints from Eocene plutonic rocks in the Santa Marta massif andSerranía de Jarara, northern Colombia. Geol. Acta 9, 445e466.

Cardona, A., Weber, M., Valencia, V., Bustamante, C., Montes, C., Cordani, U.,Mu~noz, C.M., 2014. Geochronology and geochemistry of the Parashi granitoid,NE Colombia: tectonic implication of short-lived Early Eocene plutonism alongthe SE Caribbean margin. J. South Am. Earth Sci. 50, 75e92. https://doi.org/10.1016/j.jsames.2013.12.006.

Cardoso, Y.P., Montoya-Burgos, J.I., 2009. Unexpected diversity in the catfish Pseu-dancistrus brevispinis reveals dispersal routes in a Neotropical center ofendemism: the Guyanas region. Mol. Ecol. 18, 947e964. https://doi.org/10.1111/j.1365-294X.2008.04068.x.

Carrillo-Brice~no, J.D., Argyriou, T., Zapata, V., Kindlimann, R., Jaramillo, C., 2016.A new early miocene (aquitanian) elasmobranchii assemblage from the laGuajira Peninsula, Colombia. Ameghiniana 53, 77e99. https://doi.org/10.5710/AMGH.26.10.2015.2931.

Cochrane, R., 2013. U-pb Thermochronology, Geochronology and Geochemistry ofNW South America: Rift to Drift Transition, Active Margin Dynamics and Im-plications for the Volume Balance. PhD Thesis. Universit�e de Gen�eve.

Cordani, U., Cardona, A., Jim�enez, D.M., Liu, D., Nutman, A., 2005. Geochronology ofProterozoic basement inliers in the Colombian Andes: tectonic history ofremnants of a fragmented Grenville belt. Geol. Soc. Lond. Spec. Publ. 246,329e346. https://doi.org/10.1144/GSL.SP.2005.246.01.13.

Correa, A.M., Pimentel, M., Restrepo, J.J., Nilson, A., Ordo~nez, O., Martens, U.,Laux, J.E., Junges, S., 2006. U-Pb Zircon ages and Nd-Sr isotopes of the AltavistaStock and de San Diego Gabbro: new insights on Cretaceous arc magmatism inthe Colombian Andes. In: V South American Symposium on Isotope Geology,pp. 84e86.

Cuadros, F.A., Botelho, N.F., Ord�o~nez-Carmona, O., Matteini, M., 2014. Mesoproter-ozoic crust in the San Lucas Range (Colombia): an insight into the crustalevolution of the northern Andes. Precambrian Res. 245, 186e206. https://doi.org/10.1016/j.precamres.2014.02.010.

Díaz de Gamero, M.L., 1996. The changing course of the Orinoco River during theNeogene: a review. Palaeogeogr. Palaeoclimatol. Palaeoecol. 123, 385e402.

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e7776

https://doi.org/10.1016/0031-0182(96)00115-0.D€orr, W., Gr€osser, J.R., Rodriguez, G.I., Kramm, U., 1995. Zircon U-Pb age of the

paramo rico tonalite-granodiorite, santander massif (cordillera oriental,Colombia) and its geotectonic significance. J. South Am. Earth Sci. 8, 187e194.https://doi.org/10.1016/0895-9811(95)00004-Y.

Dumont, H.J., 2009. A description of the nile basin, and a synopsis of its history,ecology, biogeography, hydrology, and natural resources. In: Dumont, H.J. (Ed.),The Nile. Monographiae Biologicae, vol. 89. Springer, Dordrecht, pp. 1e21.https://doi.org/10.1007/978-1-4020-9726-3_1.

Duque-Caro, H., 1990. Neogene stratigraphy, paleoceanography and paleobiogeog-raphy in northwest South America and the evolution of the Panama seaway.Palaeogeogr. Palaeoclimatol. Palaeoecol. 77, 203e234. https://doi.org/10.1016/0031-0182(90)90178-A.

Duque, J., 2010. Geocronología (U/Pb y 40Ar/39Ar) y geoquímica de los intrusivospale�ogenos de la Sierra Nevada de Santa Marta y sus relaciones con la tect�onicadel Caribe y el arco magm�atico circun-caribe~no. Universidad NacionalAut�onoma de M�exico.

Escalona, A., Mann, P., 2011. Tectonics, basin subsidence mechanisms, and paleo-geography of the Caribbean-South American plate boundary zone. Mar. Pet.Geol. 28, 8e39. https://doi.org/10.1016/j.marpetgeo.2010.01.016.

Feo-Codecido, G., 1956. Heavy-mineral techniques and their application to Ven-ezuelan stratigraphy. Am. Assoc. Pet. Geol. Bull. 40, 984e1000.

Forasiepi, A.M., Soibelzon, L.H., Gomez Suarez, C., S�anchez, R., Quiroz, L.I.,Jaramillo, C., S�anchez-Villagra, M.R., 2014. Carnivorans at the great americanbiotic interchange: new discoveries from the northern neotropics. Natur-wissenschaften 101, 965e974. https://doi.org/10.1007/s00114-014-1237-4.

García, C.A., Correa, K., Mantilla, F.,L.C., Bernal, L., 2009. Aspectos petrogr�aficos ygeoquímicos de las rocas metam�orficas del sector Barranco de Loba (Serranía deSan Lucas). Bol. Geol. Colomb. 31, 45e60.

Garzanti, E., And�o, S., Vezzoli, G., Ali Abdel Megid, A., El Kammar, A., 2006. Petrologyof Nile River sands (Ethiopia and Sudan): sediment budgets and erosion pat-terns. Earth Planet. Sci. Lett. 252, 327e341. https://doi.org/10.1016/j.epsl.2006.10.001.

Gaudette, H., Olszewski, W., 1985. Geochronology of the basement rocks, AmazonasTerritory, Venezuela and ectonic evolution of the western Guiana Shield. Geol.Mijnb 64, 131e143.

Gaudette, H.E., Mendoza, V., Hurley, P.M., Fairbairn, H.W., 1978. Geology and age ofthe Parguaza rapakivi granite, Venezuela. Bull. Geol. Soc. Am. 89, 1335e1340.https://doi.org/10.1130/0016-7606(1978)89<1335:GAAOTP>2.0.CO;2.

Gehrels, G.E., Valencia, V.A., Ruiz, J., 2008. Enhanced precision, accuracy, efficiency,and spatial resolution of U-Pb ages by laser abla-tionemulticollectoreinductively coupled plasmaemass spectrometry. Geo-chem. Geophys. Geosystems 9, 1e13. https://doi.org/10.1029/2007GC001805.

Giraldo-Arroyave, M.I., 2010. Esquema geodin�amico de la parte noroccidental de laCordillera Central de Colombia. Ms Thesis. Universidad Nacional de Colombia.

Goldstein, S., Arndt, N.T., Stallard, R.F., 1997. The history of a continent from U-Pbages of zircons from Orinoco River sand and Sm-Nd isotopes in Orinoco basinriver sediments. Chem. Geol. 139, 271e286. https://doi.org/10.1016/S0009-2541(97)00039-9.

G�omez, J., Montes, N.E., Alc�arcel, F.A., Ceballos, J.A., 2015a. Cat�alogo de datacionesradiom�etricas de Colombia en ArcGIS y Google Earth. In: G�omez, J.,Almanza, M.F. (Eds.), Compilando La Geología de Colombia: Una Visi�on a 2015.Servicio Geol�ogico Colombiano. Publicaciones Geol�ogicas Especiales No. 33.Bogot�a, pp. 63e419.

G�omez, J., Nivia, �A., Montes, N.E., Diederix, H., Almanza, M.F., Alc�arcel, F.A.,Madrid, C.A., 2015b. Mapa Geol�ogico de Colombia. In: G�omez, J., Almanza, M.F.(Eds.), Compilando La Geología de Colombia: Una Visi�on a 2015. ServicioGeol�ogico Colombiano. Publicaciones Geol�ogicas Especiales No. 33. Bogot�a,pp. 35e60.

Guzm�an, J.I., Fisher, W.L., 2006. Early and middle Miocene depositional history ofthe Maracaibo Basin, western Venezuela. Am. Assoc. Pet. Geol. Bull. 90,625e655. https://doi.org/10.1306/10110505035.

Hardman, M., Lundberg, J.G., 2006. Molecular phylogeny and a chronology ofdiversification for “phractocephaline” catfishes (Siluriformes: pimelodidae)based on mitochondrial DNA and nuclear recombination activating gene 2 se-quences. Mol. Phylogenet. Evol. 40, 410e418. https://doi.org/10.1016/J.YMPEV.2006.03.011.

Hendy, A.J.W., Jones, D.S., Moreno, F., Zapata, V., Jaramillo, C., 2015. Neogene mol-luscs, shallow marine paleoenvironments, and chronostratigraphy of the Gua-jira Peninsula, Colombia. Swiss J. Palaeontol. 134, 45e75. https://doi.org/10.1007/s13358-015-0074-1.

Hern�andez, R., Ramírez, V., Reyes, J.P., 2003. Evolucion Geohist�orica de las Cuencasdel Norte de Colombia. In: VIII Simp. Boliv. - Explor. Pet. en las Cuencas Sub-andinas, pp. 256e263.

Horton, B.K., Saylor, J.E., Nie, J., Mora, A., Parra, M., Reyes-Harker, A., Stockli, D.F.,2010. Linking sedimentation in the northern Andes to basement configuration,Mesozoic extension, and Cenozoic shortening: evidence from detrital zircon U-Pb ages, Eastern Cordillera, Colombia. Bull. Geol. Soc. Am. 122, 1423e1442.https://doi.org/10.1130/B30118.1.

Howard, J.L., 1993. The statistics of counting clasts in rudites: a review, with ex-amples from the upper Palaeogene of southern California, USA. Sedimentology40, 157e174. https://doi.org/10.1111/j.1365-3091.1993.tb01759.x.

Ibanez-Mejia, M., Ruiz, J., Valencia, V.A., Cardona, A., Gehrels, G.E., Mora, A.R., 2011.The Putumayo Orogen of Amazonia and its implications for Rodinia re-constructions: new UePb geochronological insights into the Proterozoic

tectonic evolution of northwestern South America. Precambrian Res. 191,58e77. https://doi.org/10.1016/J.PRECAMRES.2011.09.005.

Iba~nez-Mejia, M., Tassinari, C.C.G., Jaramillo-Mejia, J.M., 2007. U/Pb zircon ages ofthe antioquian batholith: geochronological constraints of late cretaceousmagmatism in the central Andes of Colombia. In: 11th Colombian GeologicalCongress, Extended Abstracts.

Irving, E.M., 1972. Mapa geol�ogico de la Península de la Guajira, Colombia(compilaci�on). 1:100,000. In: Inst. Nac. Investig. Geol�ogico-mineras, Bogot�a,Colomb.

Jaramillo, C., Moreno, F., Hendy, A.J.W., S�anchez-Villagra, M.R., Marty, D., 2015.Preface: La Guajira, Colombia: a new window into the cenozoic neotropicalbiodiversity and the great american biotic interchange. Swiss J. Palaeontol. 134,1e4. https://doi.org/10.1007/s13358-015-0075-0.

Jaramillo, J.S., Cardona, A., Le�on, S., Valencia, V., Vinasco, C., 2017. Geochemistry andgeochronology from Cretaceous magmatic and sedimentary rocks at 6�350 N,western flank of the Central cordillera (Colombian Andes): magmatic record ofarc growth and collision. J. South Am. Earth Sci. 76, 460e481. https://doi.org/10.1016/j.jsames.2017.04.012.

Jim�enez, C.A., 2010. Vulcanismo y mineralizaciones cupríferas en la Serranía delPerij�a. Undergraduate Thesis. Universidad Nacional de Colombia.

Jimenez Mejia, D.M., 2003. Caracterizaç~ao metam�orfica e geocronol�ogica das rochasproteroz�oicas do Maciço de Garz�on - Sudeste dos Andes da Colombia. MsThesis. Universidade de Sao Paulo. https://doi.org/10.11606/D.44.2003.tde-18092015-172653.

Kellogg, J.N., 1984. Cenozoic tectonic history of the Sierra de Perij�a, Venezuela-Colombia, and adjacent basins. In: Bonini, W.E., Hargraves, R.B., Shagam, R.(Eds.), Geological Society of America Memoirs 162, pp. 239e261. https://dx.doi.org/10.1130/MEM162-p239.

Leal-Mejía, H., 2011. Phanerozoic Gold Metallogeny in the Colombian Andes: aTectono-magmatic Approach. PhD Thesis. Universitat de Barcelona.

Lima, F.C.T., Ribeiro, A.C., 2011. Continental-scale tectonic controls of biogeographyand Ecology. In: Albert, J.S., Reis, R.E. (Eds.), Historical Biogeography ofNeotropical Freshwater Fishes. University of California Press, pp. 145e164.https://doi.org/10.1525/california/9780520268685.003.0009.

Lockwood, P.J., 1965. Geology of the Serrania de Jarara area, Guajira Peninsula,Colombia. PhD Thesis. Princenton University.

MacDonald, W.D., 1964. Geology of the Serranía de Macuira area, Guajira Peninsula,Colombia. PhD Thesis. Princeton University.

Mange, M.A., Maurer, H.F.W., 1992. Heavy Minerals in Colour. Chapman & HallLondon.

Mantilla F., L.C., Valencia, V.A., Barra, F., Pinto, J., Colegial, J., 2009. Geocronología U-Pb de los cuerpos porfiriticos del distrito aurífero de Vetas-California (Dpto deSantander, Colombia). Bol. Geol. Colomb. 31, 31e43.

Mantilla Figueroa, L.C., Mendoza, H., Bissig, T., Hart, C.J.R., 2012. Remains of earlyordovician mantle-derived magmatism in the santander massif (colombianeastern cordillera). J. South Am. Earth Sci. 38, 1e12.

Mantilla Figueroa, L.C., Mendoza, H., Bissig, T., Hart, C.J.R., 2013. The magmatichistory of the Vetas-California mining district, santander massif, easterncordillera, Colombia. J. South Am. Earth Sci. 45, 235e249. https://doi.org/10.1016/j.jsames.2013.03.006.

Mantilla Figueroa, L.C., Mendoza, H., Bissig, T., Hart, C.J.R., 2011. Nuevas EvidenciasSobre el Magmatismo Mioc�enico en el Distrito Minero de Vetas-California(Macizo de Santander, Cordillera Oriental, Colombia). Bol. Geol. Colomb. 33,43e58.

Mejía Herrera, P., Santa Escobar, M., Ordo~nez Carmona, O., Pimentel, M., 2008.Consideraciones petrogr�aficas, geoquímicas y geocronol�ogicas de la parteoccidental del Batolito de Santa Marta. Dyna 75, 223e236.

Miller, K.G., Kominz, M.A., Browning, J.V., Wright, J.D., Mountain, G.S., Katz, M.E.,Sugarman, P.J., Cramer, B.S., Christie-Blick, N., Pekar, S.F., 2005. The Phanerozoicrecord of global sea-level change. Sci. (80-. ) 310, 1293e1298. https://doi.org/10.1126/science.1116412.

Montes, C., Guzman, G., Bayona, G., Cardona, A., Valencia, V., Jaramillo, C., 2010.Clockwise rotation of the Santa Marta massif and simultaneous paleogene toneogene deformation of the plato-san jorge and cesar-ranchería basins. J. SouthAm. Earth Sci. 29, 832e848. https://doi.org/10.1016/j.jsames.2009.07.010.

Moreno-Bernal, J.W., Head, J., Jaramillo, C.A., 2016. Fossil crocodilians from the highGuajira peninsula of Colombia: neogene faunal change in northernmost southAmerica. J. Vertebr. Paleontol. 36, e1110586. https://doi.org/10.1080/02724634.2016.1110586.

Moreno, F., Hendy, A.J.W., Quiroz, L., Hoyos, N., Jones, D.S., Zapata, V., Zapata, S.,Ballen, G.A., Cadena, E., C�ardenas, A.L., Carrillo-Brice~no, J.D., Carrillo, J.D., Del-gado-Sierra, D., Escobar, J., Martínez, J.I., Martínez, C., Montes, C., Moreno, J.,P�erez, N., S�anchez, R., Su�arez, C., Vallejo-Pareja, M.C., Jaramillo, C., 2015. Revisedstratigraphy of neogene strata in the Cocinetas basin, La Guajira, Colombia.Swiss J. Palaeontol. 134, 5e43. https://doi.org/10.1007/s13358-015-0071-4.

Nie, J., Horton, B.K., Saylor, J.E., Mora, A., Mange, M., Garzione, C.N., Basu, A.,Moreno, C.J., Caballero, V., Parra, M., 2012. Integrated provenance analysis of aconvergent retroarc foreland system: U-Pb ages, heavy minerals, Nd isotopes,and sandstone compositions of the Middle Magdalena Valley basin, northernAndes, Colombia. Earth-Science Rev. 110, 111e126. https://doi.org/10.1016/j.earscirev.2011.11.002.

Ordo~nez-Carmona, O., 2001. Caracterizaç~ao isot�opica Rb-Sr e Sm-Nd dos principaiseventos magm�aticos nos Andes Colombianos. PhD Thesis. Universidade deBrasilia.

Pab�on-Caicedo, J.D., Eslava-Ramírez, J.A., G�omez-Torres, R.E., 2001. Generalidades

N. P�erez-Consuegra et al. / Journal of South American Earth Sciences 81 (2018) 66e77 77

de la distribuci�on espacial y temporal de la temperatura del aire y de la pre-cipitaci�on en Colombia. Meteorol. Colomb. 4, 47e59.

Parnaud, F., Gou, Y., Pascual, J.-C., Capello, M. a, Truskowski, P., Passalacqua, H., 1995.Stratigraphic synthesis of western Venezuela. AAPG Mem. 62, 681e698.

Pepper, M., Gehrels, G., Pullen, A., Ibanez-Mejia, M., Ward, K.M., Kapp, P., 2016.Magmatic history and crustal genesis of western South America: constraintsfrom U-Pb ages and Hf isotopes of detrital zircons in modern rivers. Geosphere12, 1532e1555. https://doi.org/10.1130/GES01315.1.

P�erez, M.E., Vallejo-Pareja, M.C., Carrillo, J.D., Jaramillo, C., 2017. A new Pliocenecapybara (rodentia, caviidae) from northern South America (Guajira, Colombia),and its implications for the great american biotic interchange. J. Mamm. Evol.24, 111e125. https://doi.org/10.1007/s10914-016-9356-7.

Pettijohn, F.J., Potter, P.E., Siever, R., 1987. Production and provenance of sand. In:Pettijohn, F.J., Potter, P.E., Siever, R. (Eds.), Sand and Sandstone. Springer, NewYork, NY, pp. 251e273. https://doi.org/10.1007/978-1-4612-1066-5_7.

Pindell, J., Kennan, L., Maresch, W.V., Stanek, K.-P., Draper, G., Higgs, R., 2005. Plate-kinematics and crustal dynamics of circum-Caribbean arc-continent in-teractions: tectonic controls on basin development in Proto-Caribbean margins.Geol. Soc. Am. Spec. Pap. 394, 7e52. https://doi.org/10.1130/SPE394.

Priem, H.N.A., Andriessen, P.A.M., Boelrijk, N.A.I.M., De Boorder, H., Hebeda, E.H.,Huguett, A., Verdurmen, E.A.T., Verschure, R.H., 1982. Geochronology of theprecambrian in the amazonas region of southeastern Colombia (western Gui-ana shield). Geol. Mijnb 229e242.

Reis, N.J., Teixeira, W., Hamilton, M.A., Bispo-Santos, F., Almeida, M.E., D'Agrella-Filho, M.S., 2013. Avanavero mafic magmatism, a late Paleoproterozoic LIP in theGuiana Shield, Amazonian Craton: U-Pb ID-TIMS baddeleyite, geochemical andpaleomagnetic evidence. Lithos 174, 175e195. https://doi.org/10.1016/j.lithos.2012.10.014.

Restrepo-Pace, P.A., Cediel, F., 2010. Northern South America basement tectonicsand implications for paleocontinental reconstructions of the Americas. J. SouthAm. Earth Sci. 29, 764e771. https://doi.org/10.1016/j.jsames.2010.06.002.

Restrepo-Pace, P.A., Ruiz, J., Gehrels, G., Cosca, M., 1997. Geochronology and Ndisotopic data of grenville-age rocks in the colombian Andes: new constraints forlate proterozoic-early paleozoic paleocontinental reconstructions of the amer-icas. Earth Planet. Sci. Lett. 150, 427e441. https://doi.org/10.1016/S0012-821X(97)00091-5.

Restrepo, J.J., Ord�o~nez-Carmona, O., Armstrong, R., Pimentel, M.M., 2011. Triassicmetamorphism in the northern part of the Tahamí Terrane of the centralcordillera of Colombia. J. South Am. Earth Sci. 32, 497e507. https://doi.org/10.1016/j.jsames.2011.04.009.

Restrepo, P., 1995. Late Precambrian to Early Mesozoic Tectonic Evolution of theColombian Andes, Based on New Geochronological, Geochemical and IsotopicData. PhD Thesis. University of Arizona.

Rinc�on, A.D., Sol�orzano, A., Benammi, M., Vignaud, P., McDonald, H.G., 2014.Chronology and geology of an early Miocene mammalian assemblage in northof South America, from cerro La cruz (castillo formation), lara state, Venezuela:implications in the “changing course of Orinoco river” hypothesis. Andean Geol.41, 507e528. https://doi.org/10.5027/andgeoV41n3-a02.

Rod, E., 1981. Notes on the shifting course of the ancient Rio Orinoco from lateCretaceous to Oligocene time. In: GEOS. Esc. Geol. Y Minas, vol. 26. UCV,Caracas, pp. 54e56.

Santos, J.O.S., Potter, P.E., Reis, N.J., Hartmann, L.A., Fletcher, I.R., McNaughton, N.J.,2003. Age, source, and regional stratigraphy of the Roraima Supergroup andRoraima-like outliers in northern South America based on U-Pb geochronology.Bull. Geol. Soc. Am. 115, 331e348. https://doi.org/10.1130/0016-7606(2003)115<0331:ASARSO>2.0.CO;2.

Schneider Santos, J.O., Hartmann, L.A., Gaudette, H.E., Groves, D.I., Mcnaughton, N.J.,Fletcher, I.R., Santos, J.O.S., Hartmann, L.A., Gaudette, H.E., Groves, D.I.,Mcnaughton, N.J., Fletcher, I.R., 2000. A new understanding of the provinces ofthe Amazon craton based on integration of field mapping and U-Pb and Sm-Ndgeochronology. Gondwana Res. 3, 453e488. https://doi.org/10.1016/S1342-

937X(05)70755-3.Siqueira, R., Hollanda, M.H.B.M., Basei, M.A.S., 2014. A novel approach to (LA-ICP-MS

acquired) U-Th-Pb data processing. In: 9th South Am. Symp. Isot. Geol. Abstr,p. 306.

Smith, D.G., 1986. Anastomosing river deposits, sedimentation rates and basinsubsidence, Magdalena River, northwestern Colombia, South America. Sedi-ment. Geol. 46, 177e196. https://doi.org/10.1016/0037-0738(86)90058-8.

Suarez, C., Forasiepi, A.M., Goin, F.J., Jaramillo, C., 2016. Insights into the neotropicsprior to the great american biotic interchange: new evidence of mammalianpredators from the Miocene of northern Colombia. J. Vertebr. Paleontol. 36,e1029581. https://doi.org/10.1080/02724634.2015.1029581.

Ujueta, G., 1999. La Cordillera Oriental colombiana no se desprende de la CordilleraCentral. Geol. Colomb. 24, 3e28.

van der Lelij, R., 2013. Reconstructing North-western Gondwana with Implicationsfor the Evolution of the Iapetus and Rheic Oceans: a Geochronological, Ther-mochronological and Geochemical Study. PhD Thesis. Universit�e de Gen�eve.

van der Lelij, R., Spikings, R., Ulianov, A., Chiaradia, M., Mora, A., 2016. Palaeozoic toearly jurassic history of the northwestern corner of gondwana, and implicationsfor the evolution of the iapetus, rheic and pacific oceans. Gondwana Res. 31,271e294. https://doi.org/10.1016/j.gr.2015.01.011.

Villag�omez, D., Spikings, R., Magna, T., Kammer, A., Winklerd, W., Beltr�an, A., 2011.Geochronology, geochemistry and tectonic evolution of the Western and Cen-tral cordilleras of Colombia. Lithos 125, 875e896. https://doi.org/10.1016/j.lithos.2011.05.003.

Vinasco, C.J., Cordani, U.G., Gonz�alez, H., Weber, M., Pelaez, C., 2006. Geochrono-logical, isotopic, and geochemical data from Permo-Triassic granitic gneissesand granitoids of the Colombian Central Andes. J. South Am. Earth Sci. 21,355e371. https://doi.org/10.1016/j.jsames.2006.07.007.

Warne, A.G., Meade, R.H., White, W.A., Guevara, E.H., Gibeaut, J., Smyth, R.C.,Aslan, A., Tremblay, T., 2002. Regional controls on geomorphology, hydrology,and ecosystem integrity in the Orinoco Delta, Venezuela. Geomorphology 44,273e307. https://doi.org/10.1016/S0169-555X(01)00179-9.

Weber, M., Cardona, A., Valencia, V., García-Casco, A., Tob�on, M., Zapata, S., 2010. U/Pb detrital zircon provenance from late cretaceous metamorphic units of theGuajira Peninsula, Colombia: tectonic implications on the collision between theCaribbean arc and the South American margin. J. South Am. Earth Sci. 29,805e816. https://doi.org/10.1016/j.jsames.2009.10.004.

Weber, M.B.I., Cardona, a, Paniagua, F., Cordani, U., Sepulveda, L., Wilson, R., 2009.The Cabo de la Vela Mafic-Ultramafic Complex, Northeastern ColombianCaribbean region: a record of multistage evolution of a Late Cretaceous intra-oceanic arc. Geol. Soc. Lond. Spec. Publ. 328, 549e568. https://doi.org/10.1144/SP328.22.

Xie, X., Mann, P., 2014. U-Pb detrital zircon age patterns of Cenozoic clastic sedi-mentary rocks in Trinidad and its implications. Sediment. Geol. 307, 7e16.https://doi.org/10.1016/j.sedgeo.2014.04.001.

Zapata, S., Cardona, A., Montes, C., Valencia, V., Vervoort, J., Reiners, P., 2014.Provenance of the Eocene soebi blanco formation, bonaire, leeward antilles:correlations with post-eocene tectonic evolution of northern South America.J. South Am. Earth Sci. 52, 179e193. https://doi.org/10.1016/j.jsames.2014.02.009.

Zapata, S., Weber, M., Cardona, A., Valencia, V., Guzm�an, G., Tob�on, M., 2009.Provenance of Oligocene conglomerates and associated sandstones from theSiamana Formation, Serranía de Jarara, Guajira, Colombia: implications forOligocene Caribbean-South American Tectonics. Boletín Ciencias la Tierra.Medellín. Colomb 27, 7e24.

Zuluaga, C.A., Pinilla, A., Mann, P., 2015. Jurassic silicic volcanism and associatedcontinental-arc Basin in northwestern Colombia (southern boundary of thecaribbean plate). In: Bartolini, C., Mann, P. (Eds.), AAPG Memoir 108: PetroleumGeology and Potential of the Colombian Caribbean Margin, pp. 137e160.https://doi.org/10.1306/13531934M1083640.