Cainozoic Research November 2002 , 1(1-2) (2001), pp. 35-81,

Lake Pebas: a palaeoecological reconstruction of a Miocene,

long-lived lake complex in western Amazonia

1,7 4 4, F.P. M.E. Räsänen G. Irion H.B. Vonhof R. W. Renema¹, Wesselingh , ², ³, , Kaandorp 5 6 L.+Romero Pittman & M. Gingras

1 NationaalNatuurhistorischMuseum, P.O. Box 9517, NL-2300 RA Leiden,, the Netherlands: e-mail: [email protected]

department ofGeology, University of Turku, SF 20014 Turku, Finland

3 Senckenberg Meeresforschungsinstitut, D 26382 Wilhelmshaven, Germany

4 Department ofEarth Sciences, Vrije Universiteit, de Boelelaan 1085, NL-1081 HV Amsterdam, the Netherlands

S INGEMMET, Avenida Canada 1470, San Borja, Lima 12, Peru

6 Departmentof Geology, University ofNew Brunswick, P.O. Box 4400, Fredericton-NB E3B#5A3, Canada 1 Biodiversity Centre, University of Turku, SF20014 Turku, Finland

Received 20 February 2001; revised version accepted 19 February 2002

The taxonomic composition and palaeoecological signature of molluscan faunas from the Miocene Pebas Formation of Peruvian

Amazonia are assessed. The Pebas fauna is almost entirely made up of extinct, obligate aquatic taxa, and is dominatedin numbers of

species and specimens by endemic cochliopine hydrobiid gastropods and pachydontine corbulid bivalves. Molluscan assemblages

are defined and linked to depositional environments. Isotope data from the shells indicate freshwater settings during deposition of with the Pebas Formation, the exception of a few incursion levels that were deposited under oligohaline-mesohaline conditions.

Faunal and data to a freshwater lake at level with and to isotope geochemical point large, long-lived system sea swamps deltas, open marine settings in the north (Llanos Basin). Sedimentological data and ichnofossils point to (restricted) marine settings. These dif-

ferent interpretations are discussed, and it is concluded that faunas (including ichnofabrics) from evolutionary isolated and long- lived be assessed systems cannot in a straightforward actualistic mode, using taxa from non-long-lived environments for compari-

son. Aspects of Lake Pebas are compared with modem depositional environments. Lake Pebas is among the largest and longest-

lived lake complexes in Phanerozoic history; it was an important stage for the evolution of endemic molluscan and ostracod faunas.

It may have played some role in the transition of marine biota to Amazonian freshwater environments during the Miocene, and

likely was an important, hitherto unrecognised, dispersal barrier for terrestrial organisms in northwest South America during the Miocene.

Key words Palaeoecology, molluscs, Miocene, Amazonia, Pebas Formation, community palaeoecology, long-lived lake, Lake

Pebas.

Introduction away from modern coastal areas, was questioned. Gabb’s

work was the first of some fifteenpapers dealing with the

In 1869, the American palaeontologist W.M. Gabb re- remarkable fossil molluscs from the Pebas Formation of

remarkable fossil molluscan fauna from Pe- western Amazonia. Different authors with dif- ported on a came up bas in Peruvian Amazonia. All but one of the species ferent interpretations of the environment that sustained

described Gabb and the bore by were new, assemblage these unusual faunas (terrestrial, freshwater, brackish but little resemblance to any modem South American and/or marine settings) and the origin of possible marine

fauna. Almost all but the species were extinct, exquisite influence (Pacific, southern Atlantic, eastern Atlantic

preservation of the fauna led Gabb to believe that the approximately through the course of the present-day

fossils were ofrelatively recent age. Gabb described sev- Amazon River, and Caribbean).

eral taxa whose relatives at present occur in settings with In 1990, Patrick Nuttall published his seminal mono-

fluctuating salinities, others that are marine, and still oth- graph on the Neogene molluscs from western Amazonia

with freshwater The in what ers a affinity. question arose and adjacent regions (Nuttall, 1990a). He clarified much

kind of environment the Pebas molluscs had lived, and of the systematic uncertainties that complicated age esti-

the inferred of marine environmental given presence taxa (the gastropod mates and interpretations. Based on a me-

genera Turbonilla and Mesalia and the bivalve Corbula), ticulous region-wide stratigraphic correlation he assigned

the origin of marine settings in lowland Amazonia, far a Middle Miocene age to the fauna. Nuttall (1990a) in- -36-

terpreted the depositional environment in Miocene west- employed here. Finally, implications for landscape evo- ern Amazonia as a continually shifting pattern of lution and biotic evolution in Miocene western Amazonia streams, swamps and lakes of varying salinity, with ma- are discussed. rine influence from the north (Caribbean). Nuttall’s age assignments were confirmed and refined by Hoorn

(1993, 1994a, b) on the basis of pollen. Hoorn concluded Material and methods that the Pebas Formation included at least three pollen

between late Miocene and sediment zones, spanning an age Early Fossil-bearing samples, collected (in 1996) in early Late Miocene (c. 17-10 Ma). Based on pollen and outcrops of Pebas Formation deposits in Peruvian sediments, Hoorn (1994a) interpreted the depositional Loreto, were weighed and washed (minimum sieve mesh environment ‘fluviolacustrine with marine influ- 0.3 locations in as some mm). Sampling are given Appendix 1. ence’. Samples typically weighed about 1 kg. A total of 285

The discovery of supposed Miocene tidal deposits in samples were assessed for taxonomic composition. The

Brazilian led Webb distribution of ‘Pebas’ and Acre (Rasanen et al., 1995) (1995) geographic contemporanous to Miocene western Amazonia of non-marine faunas is shown in interpret as part an Figure 1. The age of the intracontinental seaway that would have occupied all samples is between late Early Miocene and early Late

basins of inland South America. Miocene major sedimentary (Figure 2; see Hoom 1993, 1994a).

Webb’s views severe criticism Hoom, The Pebas Formation is referred lithostrati- provoked (ie.g., to as a

1996; Paxton et al., 1996; L.G. Marshall & Lundberg, graphic unit comprising predominantly turquoise-blue

1996). Nevertheless, more tidal deposits have been dis- smectitic clays; immature, feldspar-rich, usually grey covered since then in Peruvian Amazonia (Rebata, 1997; sands, and brown-black organic clays and lignites. The

Rasanen et al., 1998), and in the Pebas Formation. The formation is characterised by 3-7 m thick, predominantly predominantly endemic molluscan and ostracod faunas coarsening-upwards (CU) cycles, the brilliantly turquoise

(Nuttall, 1990a; Wesselingh, 1993; Whatley et al., 1996; smectite-rich clays, as well as the common occurrence of

Munoz-Torres et al., 1998; Vonhof et al., 1998) conflict lignitic intervals (Rasanen et al., 1998). The Pebas For- with interior envi- mation is an interpretation as a possible seaway extremely rich in fossils, often beautifully pre- ronment Rasanen et al. and Webb served. wood well as proposed by (1995) e.g. molluscs, ostracods, fragments as

(1995), as well as with fluvio-lacustrine conditions as as fish, amphibian and reptilian remains. proposed by Hoorn (1993, 1994a, b). Based on stron- In Peru, this formation is properly known as the Pe-

carbon and well bas Formation tium, oxygen isotopes, as as on fauna, (see e.g., Rasanen et al., 1998). The Cura-

Vonhof et al. (1998) concluded that the Pebas Formation ray Formation of eastern Ecuador (Tschopp, 1953; Bal- had been deposited in a lacustrine system, occasionally dock, 1982) is the time equivalent of the Pebas Forma- reached by marine incursions. A Caribbean (northerly) tion. In ColombianAmazonia, deposits formerly referred

of the incursions Nuttall ‘Terciario inferior Amazonico’ attributed origin as proposed by (1990a) to as were by and Hoom (1994b; see also Hoorn et al., 1995) was sup- Hoom (1994b) to the Pebas Formation. The La Tagua

these faunal and data. The ostracod Beds ported by isotope (Nuttall, 1990a; Hoom, 1994b) are included in the fauna from the Pebas Formation was studied by Whatley Pebas Formation as well. In Brazil, Pebas Formation et al. (1996), who proposed a low energetic athalassic deposits have usually been included in the Solimoes

based the taxonomic (inland-sea) environment, on com- Group/Formation, an informal unit comprising a plethora position and exceptional preservation of the ostracod of western Brazilian Amazonian Cainozoic strata (eg., fauna. Compilations of Pebas molluscan taxa autecologi- Maia et al., 1977), but some authors (Costa, 1981; Petri cal such Nuttall table signatures, as provided by (1990a, & Fulfaro, 1983) have indeed distinguished the Pebas

2) and Vonhof et al. (1998, table 1) show an admixture Formation as such in Brazil. of (inferred) ecological preferences. Almost all species In the present paper we use the terms ‘lowermost’, from the Pebas Formation several of are extinct, and the ‘lower’, ‘middle’ and ‘upper’ Pebas Formation loosely; genera dominating the fauna are extinct or have survived all in lower case, in order not to be confused with as relics only (Wesselingh, 2000), limiting their actualis- ‘Lower’, ‘Middle’ and ‘Upper’ in lithostratigraphic us- tic This has into the For the Pebas ecological use. complicated insight age. Formation, ‘lower’, ‘middle’ and ‘up- environmental system in which these faunas lived. per’ correspond to the Psiladiporites/Crototricolpites

The aim of the is the present paper to assess palaeo- Concurrent Range, Crassoretitriletes Acme and environment in which the Pebas Formation was depos- Grimsdalea Interval zones of Hoom (1993), respectively

based of molluscan ited, on a palaeoecological analysis (Figure 2). ‘Lowermost’ refers to all deposits deemed to samples from individual shell-bearing horizons. Palaeo- be older than the Peruvian localities studied by Hoom, environments should be reconstructed by integrating fau- and includes the La Tagua Beds of southern Colombia. nal, sedimentological and isotope data, which method is -37-

Figure 1. Distributionof Pebas molluscan faunas and the probable maximum limitsof the Pebas system (dashed line). Filled circles

are faunas checked by the authors, open circles are unconfirmed reports. Black triangles are contemporaneousnon-marine mol-

luscan occurrences checked by the authors, open triangles are unconfirmed contemporaneous non-marine molluscs. Encircled

triangles are contemporaneous molluscan faunas that have species in common with the Pebas fauna. Distribution of Pebas For-

mation deposits in Colombia is adapted from Floom (1994b), with a westerly extension into the Putumayo Basin. Approximate distributionof Curaray Formation in Ecuador follows Tschopp (1953) and Baldock (1982); distribution of the Pebas Formation

in Brazil follows Petri & Fulfaro (1983). Peruvian localities are those from Nuttall (1990a), Wesselingh (Appendix 1 here) and

Romero-Pittman (1997). -38 -

2. and in this Ol M Figure Stratigraphic framework for the Pebas Formation geological units discussed paper. Oligocene,

T S L B A C 1 Messinian, Tortonian, Serravallian, Langhian, Burdigalian, Aquitanian, Chattian, u upper, m middle, lower, 1m

lowermost; the timescale is after Berggren et al. (1995).

The systematic status of numerous molluscan taxa in estimated abundances for each of the taxa groups. Cate- our material is uncertain. More than half of the fauna is gories were logarithmically chosen (0%, >0-1%, >1-2%, undescribed The Pebas molluscan fauna >2-4%. >4-8% The of this of (Appendix 2). etc.). use type categories, a also contains several species flocks, further complicating slight modificationof the so-called ‘octave scale’, is dis-

taxonomic cussed in Smith The scale’ has the species-based analyses. By using groups, (1999). ‘octave great some of which contain single species while others in- advantage that data do not suffer the closed sum effect in clude several have the genera, we attempted to overcome statistical analyses and that inherent noise and minor systematic uncertainties. Classification procedures are variation within the data set are suppressed, leaving only outlined in Appendix 2. The senior author (FPW) has the basic major signals (Smith, 1999). It cannot be over- -39-

that these methods deliver crude in- sition of each of the emphasised only a assemblages were calculated using

sight into faunal composition and abundances. The sam- SPSS version 8.0.

ples were subjected to a variety of statistical analyses. In Subjecting the taxa/sample matrix to Ward clustering

results of methods the present paper Ward clustering using Euclidean distances produced five clusters of sam-

using Euclidean distances were subjected to indicator- ples that are described as assemblages below. Indicator-

species analysis as described in Duffene & Legendre species analysis (Duffene & Legendre, 1997) was ap-

(1997). Following those authors, indicator values over 30 plied. An R-mode cluster analysis was also performed. A

were considered to be meaningful. Analyses were run on few species-groups were found to cluster tightly. Pear-

version PC-ORD, 4 (McCune & Mefford, 1999). In or- son’s correlation coefficients were computed for these

der to discuss between combinations. This three that taxa (dis-)similarity assemblages, produced groups yield

Pearson’s similarity coefficients for the average compo- that cluster together and show a correlationof over 0.4.

genera species species% abundance %

Neritidae 1 4 3,0 0,9

Ampulariidae 1 1 0,7 0,0

Cochliopinae 15 75 56,0 28,2

Pachychilidae 2 7 5,2 0,6

Thiaridae 3 5 3,7 0,7

perimarine taxa 0,1

Melongenidae 1 1 0.7

Nassariidae 1 1 0,7

Pyramidellidae 1 3 2,2

Pulmonata 0,7

Planorbidae 2 2 1,5

Ferrissidae 1 1 0,7

Acavidae 1 1 0,7

Bulimulidae 1 1 0,7

indet. family 1 1 0,7

Corbiculidae 1 1 0,7 0,1

Sphaeriidae 2 2 1.5 0,1

Dreissenidae 1 2 1,5 1,6

Hyriidae 2 3 2,2 0,4

Mycetopodidae 1 2 1,5 0,3

Tellinidae 1 1 0,7 0,0

Pachydontinae 6 20 14,9 66,5

Table 1. Taxonomic composition (estimated numbers of species per (sub)family) and estimated molluscan abundance (percentages)

ofthe Pebas fauna.

The groups are (A) Pebasia group + Ostomya group, (B) 1874 (Nuevo Horizonte, Loreto dept., Peru, level 366;

Neritina roxoi group + Pachychilidae + Thiaridae and M. Rasanen Colin), and Pachydon erectus Conrad, 1871

(C) Corbicula + perimarine taxa. These three groups are (Santa Rosa de Pichana, Loreto dept., Peru; collected

used additional for the from as descriptors assemblages. surface of outcrop at the fossiliferous interval from

Stable shells from which isotope analyses were run on 109 samples 533-536 were taken, F. Wesselingh Colin,

seven different levels following analytical procedures 1996) were sampled along growth increments. The

described al. Horizonte in Vonhof et (1998). Five of the levels are Nuevo specimen is of late Middle to early

from Santa Rosa de Pichana, the remaining two are from Late Miocene age (Grimsdalea Interval Zone). Within-

shell outcrop Tamshiyacu (see Appendix 1). Both outcrops are isotope profiles of the fossil shells were compared

located in the of the Crassoretitriletes Acme of bivalve shells from upper part to isotope profiles two the present- characterised Zone, by the occurrence of the gastropod day Amazonian floodplains. These are Anadontites

Sioliella bella and of Middle Mio- (Conrad, 1874), are trapesialis (Lamarck, 1819) from a floodplain lake on

cene age. Entire or slightly damaged shells were ho- Isla Indiana (F. Wesselingh Colin, 1996; 73°30’ W,

to in order to ‘mean’ and from mogenised prior analyses, provide a 3°32’S) Diplodon sp. the Itaya River, southwest isotopic signature for the shells. of Iquitos (Kaandorp & Vonhof Colin, 1998; 73°16’W,

within-shell For exploring isotopic variation a speci- 3°45’S). Salinities are given in practical salinity units

men each of the bivalves Diplodon longulus Conrad, (psu). -40-

Figure 3. Morphological diversity in Pebasian cochliopines. Scale bar represents 1 mm -41 -

A- RGM 456 Amazonas 000; Dyris sp., Macedonia, dept., Colombia;

B- RGM 456 001; Nanivitrea hauxwelli (Nuttall, 1990a), Indiana,Loreto dept., Peru;

C- RGM 456 002; Sioliella sp., Santa Teresa, Loreto dept., Peru;

D- Santa Amazonas RGM 456 003; Tryonia sp., Sofia, dept., Colombia;

E- RGM 456 004; Dyris ortoni (Gabb, 1869), Macedonia, Amazonas dept., Colombia;

F- RGM 456 005; Dyris sp., Chimbote, Loreto dept., Peru; G- RGM 456 006; Nanivitrea degrevei (Nuttall, 1990a),Santa Julia, Loreto dept., Peru;

H- Loreto RGM 456 007; genus and species indet., Nuevo Horizonte, dept., Peru;

I- RGM 456 Santa Amazonas 008; Onobops sp., Sofia, dept., Colombia;

J- RGM 456 009; Toxosoma sp., Mocagua, Amazonas dept., Colombia;

K- RGM Santa Loreto 456 010; Pyrgophorus sp., Teresa, dept. Peru;

L- RGM 456 011; Dyris sp., Nuevo Horizonte, Loreto dept., Peru;

M- RGM 456 012; Sioliellabella (Conrad, 1874), Puerto Almendras, Loreto dept., Peru; N- RGM 456 013; Tryonia tuberculata (de Greve, 1938), Santa Julia, Loreto dept., Peru;

O- RGM 456 014; Tropidobora tertiana (Conrad, 1874), Iquitos Puerto Ganso Azul, Loreto dept., Pem.

Clay mineralogy analyses were run on twenty-seven grazers and browsers as well as scavengers (Gittenberger

samples from stratigraphic intervals covering the three & Janssen, 1998). Pebasian Neritina ortoni Conrad, 1871

pollen zones defined by Hoom (1993); analytical proce- is characterised by its flared outerlip and velatiform out-

dures are described in Rasanen et al. (1998). The abbre- line (Nuttall, 1990a). Possibly these were adaptations for

viation collections of the bottoms. RGM refers to the Department living on muddy of Cainozoic Mollusca, Nationaal Natuurhistorisch Mu-

* seum, Leiden (the Netherlands; formerly Rijksmuseum Ampullariidae (1 species)

voor Geologic en Mineralogie).

Ampullariid snails are very rare in the Pebas Formation.

snails Today, these amphibious occur abundantly in

Composition ofthe fauna floodplains of Amazonia, where they litter areas that

experience regular flooding. Ampullariids feed on

The estimated number of and abundance of species aquatic vegetation, and are extremely salt-intolerant (von

families in the Pebas fauna are given in Appendix 3 and Cosel, 1986).

summarised in Table 1. The entire Pebas fauna contains

* at least 135 species, 59 of which have been described Cochliopinae (c. 75 species)

previously. The taxonomic work is in progress, so these

indications of numbers serve as only. In numbers species The Cochliopinae outnumbers any other group in the

the Cochliopinae (55.6%) and Pachydontinae (14.1%) Pebas fauna in number of species. Many of these, and

dominate. Other common groups in numbers of species some of the genera as well, are undescribed and appear

include Neritidae (3.0%), Pachychilidae (5.2%), Thiari- endemic to Miocene deposits of western Amazonia

dae (3.7%) and Unionoidea (Hyriidae and Mycetopodi- (Nuttall, 1990a; Wesselingh, 1993). Some speciose gen-

dae, 3.7%). era, such as Dyris and Sioliella(referred to as Eubora by

In estimated abundance, Pachydontinae (66.5%) earlier authors) currently occur as relics only (Dyris

dominate, followed by Cochliopinae (28.2%). All other amazonicus (Haas, 1949) and Sioliella effusa Haas, 1949

groups together make up only an estimated 5.3% of the in the lower Tapajos River of central Brazil; Wesselingh,

molluscan fauna. 2000), restricting their value as uniformitarian ecological

markers. Other low numbers of genera, represented by

Trophic and life habit characteristics offaunal elements species in the Pebas Formation, such as Pyrgophorus and

Nanivitrea, are widespread in modem freshwater settings

* Neritidae(4 species) of the tropical Americas (Hershler & Thompson, 1992).

Extant species of Neritina are mostly hard-substrate Some cochliopine genera are known only from the Pebas

mud-dwellers. obli- Formation dwellers, some are Some species are (e.g., Toxosoma, Tropidobora, Liosoma).

gate freshwater inhabitants, whereas others are found in Other Pebasian species are tentatively assigned to extant

brackish lagoons, and still others occur in tidal pools on neotropical genera known from outside Amazonia (e.g

rocky shores (Rodriguez, 1963; Russell, 1941; von Co- Lithococcus and Onobops), but this is in need of system-

sel, 1986). The latter tolerate salinity changes from atic confirmation.The Pebas Formation is unusually rich

freshwater to hypersaline. Mangrove-inhabiting Neritina in hydrobiid taxa as well as morphotypes (Figure 3), in-

cling to roots and may experience periodic drought. cluding planorbiform, naticiform, trochiform, turritelli-

Various coastal neritids withstand and and large temperature form, turbonilliform, hydrobiform species. Such an

salinity fluctuations (Russell, 1941). Neritids are algal excessive morphological diversity is known from long- -42-

lived lake environments (e.g Lake Ohrid, Lake Baikal, Melongena woodwardi (Roxo, 1924) occurs in marine

Miocene Lake incursion levels in Caspian Sea, Pannon; see Boss, 1978; the upper Pebas Formation only

Muller et al, 1999, amongst others). Morphological (Vonhof et al, 1998; Vermeij & Wesselingh, 2002). features found in the Pebasian cochliopines, such as the Melongenid gastropods are marine to perimarine carni- trend towards open coiling, miniaturisation as well as vores that feed on barnacles, other molluscs or carrion. gigantism (the smallest hydrobiids are barely 2 mm, They can withstand reduced salinities but are not known

in the di- from freshwater whereas the largest exceed 21 mm height) and settings.

of elements known in verse array sculptural are to occur

* long-lived lake gastropod faunas (see e.g.. Boss, 1978; Nassariidae (1 species)

Gorthner, 1992). The functionality of these is not always clear (Gorthner, 1992). The denticulation seen in Tox- A species of Nassarius is known from marine incursion osoma is known only from one other hydrobiid genus intervals in the upper Pebas Formation (Vonhof et al,

(Hemistomia Crosse, 1872), known from New Caledonia 1998, fig. 2-1; Vermeij & Wesselingh, 2002). Nassariids and Lord Howe Island; Haase & Bouchet, 1998). To our are primarily carnivores or scavengers living in marine knowledge, siphon-like structures, as seen in most spe- and brackish environments, and have been reported to

cies of and occasion- tolerate conditions Sioliella, Tropidobora Toxosoma, meso- to oligohaline (Cemohorsky, ally including a second adapical sinus in some species of 1984; Gittenberger & Janssen, 1998). They are not

Sioliella, are a unique feature for hydrobiid snails. Extant known from freshwater environments. cochliopine snails occur in aquatic habitats ranging from

* high-altitude freshwater streams to brackish coastal set- Pyramidellidae (3 species) tings (Hershler & Thompson, 1992). Most species are

detritivores, includes Three some are grazers. Cochliopinae species of Odostomia have been found in the up- epifaunal and semi-infaunaldwellers. per part ofthe Pebas Formation in levels containing other

indicators of marine influence (van Aartsen & Wessel-

* Pachychilidae (7 species) ingh, 2000). Pyramidellids are parasitic snails, living on

and in molluscs and other invertebrates. Pyramidellid

The freshwater known from of salinities low Pachychilidae comprises neotropical species are a range (as as

dwellers that graze on algae. They are dioecious(Nuttall, oligohaline), but have not been reported from freshwater

1990a; Glaubrecht, 1996). Extant pachychilids (eg environments.

Doryssa and Pachychilus) are mostly substrate depend-

* ent, which might be related to feeding and egg- Planorbidae (3 species)

deposition. Five species of Sheppardiconcha and two undescribed species tentatively assigned to Doryssa are Helisoma is an extremely rare constituent of the Pebas

Pebas Formation representatives of this family. The su- fauna, represented by two species. Specimens tentatively

is praspecific classification of Doryssa uncertain (Glau- attributed to Drepanotrema were found in a single sam- brecht, 1996). ple only. Planorbids are limited to freshwater environ-

ments, both permanent and ephemeral. They feed on or-

* Thiaridae(5 species) ganic detritus, algae, and decaying plant material (Baker,

1945). The scarcity of planorbids in the Pebas Formation

Thiaridae are represented by two genera in the Pebas contrasts markedly with the abundance of this group in

and third modem South American freshwater Formation, viz. Aylacostoma Hemisinus. A ecosystems.

genus, Charadreon (to which Longiverena eucosmius

* Pilsbry & Olsson, 1935 should be assigned), is tenta- Ferrissiidae (1 species) tively placed in this family, but its assignment needs further investigation. Extant species of Hemisinus are Only a single species (?JHebetancylus sp.) is known from

widely distributed in freshwater ecosystems of the Carib- the Pebas Formation (Nuttall, 1990a, p. 261). Some bean and northern South America (Nuttall, 1990a; Glau- fragments attributable to this species have been recog- brecht, 1996). Some species of Hemisinus have been nised in two of the samples assessed for the present pa-

to tolerate saline conditions Ferrissiid and reported temporal (Glau- per. gastropods (Laevapex Hebetancylus brecht, 1996). The genus Aylacostoma comprises various spp.) are common in aquatic environments of white-

extant species living in rivers and other freshwater habi- water floodplains (floodplains of rivers draining the An- tats in eastern Brazil (Nuttall, 1990a). One species of des) of central Amazonia (Irmler, 1975), where they feed

Charadreon, C. ruginosum (Morelet, 1849), lives in on detritus and occasionally on fungi.

Guatemalan lakes. Like pachychilids, thiarid snails are

* (sub-)tropical freshwater dwellers that graze on algae and Acavidae (1 species) organic detritus.

Pebasiconcha immanis Wesselingh & Gittenberger,

* Melongenidae (1 species) 1999 is a huge, extinct terrestrial snail (up to 26 cm in -43 -

whose relatives height), extant live on trees and on the cies of Corbicula in South America tolerate wide may a

in forest where of salinities ground neotropical environments, they range (Darrigan, 1992). It is somewhat sur- feed material Pebasiconcha on plant (e.g., leaves). is prising that the rare Corbicula valves found in the Pebas

restricted to the Miocene of and is Formation western Amazonia, usually occur in samples containing indicators

not rare in Pebas Formation of marine wheras deposits. influence, strontium isotope measure-

ments on the corbiculid shells (H. Vonhof, unpublished

* Orthalicidae freshwater (2 species) data) point to a habitat. The few samples in

which Coribula have been found also contain freshwater

Orthalicus linteus is known from (Conrad, 1874) only taxa, such as cerithioideans, pointing to a mixed origin of the Pebas Formation it is (Nuttall, 1990a); a rare con- faunas in them. Corbiculids are mainly filter feeders but stituent of the fauna. A second, unidentified, species have been also reported to be facultative deposit feeders

attributed to this tentatively family has been found in a (Way et al, 1990). Corbicula is common in Miocene

Extant of Orthalicus wide- single sample. species are fossiliferous deposits of the tropical Andean region (see

spread in neotropical forests, where occur on the they e.g., Nuttall, 1990a), but a very rare constituent of the

forest floors and in trees and feed on leaves, lichens etc. Pebas fauna.

* Dreissenidae * (2 species) Sphaeriidae (2 species)

Species of Mytilopsis are hard substrate- epifaunal, Sphaeriids are very rare in the Pebas Formation. They attached filter feeders. dependent, byssally Often, speci- are represented by two species only, one assigned to the form mens aggregates or clumps (Nuttall, 1990b). genus Eupera, the other to Pisidium. These tiny nutclams of found few occasions in Clumps Mytilopsis were on a are obligate freshwater inhabitants. The habitat of the Pebas Formation deposits. Extant species of Mytilop- sphaeriids ranges from ephemeral and small-bodied

sis are found in fresh, brackish and freshwater where occasionally hyper- settings to large lakes, they live at or in saline Some settings (Nuttall, 1990b). Mytilopsis occur- bottoms, or clogged on the vegetation (Gittenberger & rences have been from far inland reported (e.g., Janssen, 1998). Sphaeriids are filter feeders, but Amazo-

Figueiredo-Alvarenga & an ob- Ricci, 1989), indicating nian sphaeriids have been reported to feed on small de-

freshwater existence for these. These mussels tritus ligate are particles and occasionally on fungi (Irmler, 1975). able to colonise both and fast-running relatively stagnant Sphaeriids do not survive in salt water.

waters.

* Tellinidae(1 species)

* Pearly freshwater mussels: Mycetopodidae (2 species) and Hyriidae (3 species) Macoma sp. has been found at a single marine incursion

level in the Pebas Formation only (Vonhof et al, 1998; Pearly freshwater mussels are semi-infaunal filter feeders as Psammotreta sp.). Today, species of Macoma live in that restricted to freshwater such are ecosystems, as marine and marginal marine environments as infaunal lakes, streams and rivers. During reproduction, juveniles deposit feeders, capable of facultative suspension feeding fish. This undergo a parasitic stage on way they gain as well (Cadee, 1984). Minimum salinities reported for their distribution to rather immobile adult life prior a Macoma are 3 psu (Kuiper, 2000), comparing well with these mussels stage. Nowadays, are widespread in many inferred maximum salinities from strontium analyses of freshwater habitats of South America, including Amazo- 3-5 for the interval in which psu Macoma was encoun- nia (Nuttall, These taxa are not able to withstand 1990a). tered (Vonhof et al, 1998). Macoma was not encoun-

salt water. Pearly freshwater mussels are a common con- tered in the studied for samples the present study. stituent of the Pebas fauna; they were found in nearly one

thirdof the samples, but their numbers are low (0.8% of * Corbulinae(1 species) estimated abundance).

An undescribed species of Panamicorbula was encoun- * Corbiculidae (1 species) tered in a ‘marine’ interval in the Buenos Aires section of

southern Colombia (Vonhof et al, 1998). Those authors

Species of the semi-infaunal Corbicula occur in genus listed it as Pachydon cebada (F.M. Anderson, 1928) and

many freshwater systems in South America, but to P cf. ovalis 1990a. The appear Nuttall, same species was illus- have been lacking in the Amazon until intro- system, trated by Hedberg (1936, pi. 8, fig. 4) from the Miocene ducted recently (pers. obs.). Unlike the corbiculid of northern Venezuela. Two species of Panamicorbula

Polymesoda, which lives in oligohaline and mesohaline are known from Pacific mangroves, from Mexico to Peru habitats in northern South America (Rodriguez, 1963; (L. Anderson, 1996). Panamicorbula has been reported von Cosel, 1978), neotropical Corbicula appears entirely from “soft mud’ in the impalpable upper (freshwater- restricted tofreshwater ecosystems (e.g., W.B. Marshall, reaches of dominated) mangroves on the Pacific side of 1927; de Morretes, 1949). Recent Lange immigrant spe- Panama (Olsson, 1961). Fossils are known from the -44-

Miocene and Pliocene of Venezuela and Trinidad, as Ecology of the Pebas fauna well as from the Miocene of Costa Rica and the Domini- can Republic (L. Anderson, 1996).

tolerance * Pachydontinae (c. 19 species) salinity terrestrial taxa 0,4%

obligatefreshwater taxa 2,5%

The Pachydontinae is subfamily of the Corbulidae. Cor- freshwater to oligohaline taxa 97,1%

to taxa bulids are semi-infaunal filter feeders (Beesley et al, oligohaline hypersaline 0,0%

1998) and are well known to tolerate dysoxic settings. In feeding characteristics abundance the Pachydontinae dominatethe Pebas fauna, grazers/browsers 2,4% in number of species they are second after the cochliopi- suspension feeding 68,8% nes (Table 1). Four of the six pachydontine genera that detritus feeding 28,7% scavengers/carnivores 0,1% occur in the Pebas fauna (Pachydon, Pebasia, Ostomya parasites 0,1% extinct 1990a; see and undescribed genus 1) are (Nuttall,

2 These four endemic to Appendix here). genera appear life habit

Oligo-Miocene inland basins of northwest South Amer- infaunal sessile 67,4%

and semi-infaunal ica. One species is attributed to a second undescribed epifaunal vagile 28,2% epifaunal vagile 2,6% genus, and tentatively placed in the Pachydontinae. The epibyssate 1,6% sixth genus, Anticorbula, is a rare constituent of the Pe- bas fauna. Extant Anticorbulafluviatilis (Adams, 1860) lives in freshwater habitats of the central and lower

Amazon well in rivers the Table 2. region, as as draining Guyana Inferred gross ecology, feeding ecology and life habit

Pebas Shield (up into the upper reaches of estuaries), mostly for the fauna (based on estimated abundance of taxa byssally attached to hard substrates, often nestling (Nut- in the Pebas fauna).

tall, 1990a; Leistikow & Janssen, 1997; Simone, 1999).

The attached mode of life of Anticorbula is atypical of other Pebasian pachydontines with the possible exception Endemic elements make up 90.2% of the Pebas fauna, of Pebasia. The shell torsion seen in Pebasia has been which may be characterised as almost exclusively

Savazzi indicate endemic and extinct. Extant relatives of Peba- suggested by & Peiyi (1992) to a pleu- aquatic, rothetic life habit and possibly a strong byssal attach- sian taxa occur in a range of salinities, from freshwater to ment. Specimens of Pachydon tenuis Gabb, 1869 show hypersaline. Strictly freshwater and brackish-marine taxa

forms described in the Pebas fauna the moderate shell torsion comparable to are rare (Table 2). Despite rarity from extant species of Cuneopsis in China. This shell of freshwater taxa (Ampullariidae, Pachychilidae, Thi- torsion has been attributed to a shallow burrowing be- aridae, Corbicula, Sphaeriidae, Hyriidae, Mycetopodi- haviour (Savazzi & Peiyi, 1992). In the field, articulated dae) in terms of estimated abundance (2.5% of the

valves ofPachydon were quite often found preserved in fauna), these groups were found (in low numbers) in

have studied. with marine life position. All species of Pachydon appear to 54% of the samples Taxa a affinity

which is indicated the survive in freshwater make been shallow borrowers, by pres- (unknown to settings) up only

of shallow sinus in most 0.1% ofthe faunas studied and found in 5% of ence a very pallial specimens. were only Pachydon obliquus Gabb, 1869, the commonest species the samples.

to have of extant taxa of Pachydon in the Pebas Formation, appears The paucity or absence pachydontine adapted especially well to dysoxic settings. Articulated and many of the cochliopine gastropod genera that domi-

of salin- specimens, often in situ or with signs of minimal trans- nate the Pebas fauna complicates an assessment

dominate the faunas of based uniformitarian which is illus- port, are found to organic-rich ity on a approach, clayish intervals, where other pachydontine (and trated in the large portion in Table 2 assigned to fresh-

less abundant. faunas. For similar the cochliopine) species are considerably water to oligohaline reasons as-

The Pebas fauna is dominated by aquatic molluscs, sessment of trophic characteristics and life habit of the which constitute in excess of an estimated 99.3% in faunal members are subject to uncertainty. Trophic char- abundance (the pulmonate snails accounting for the other acterisation is furthermore complicated by the ability of

0.7% includeboth terrestrial and freshwater taxa). Nearly some species to switch between feeding modes, so-called

all of the species are restricted to the Pebas Formation, ‘opportunistic feeding’ (Cadee, 1984). The procedure of

and do not occur in contemporaneous fossiliferous de- inferring ecological characteristics of the fauna is out- posits in nearby Ecuador and Venezuela (Nuttall, 1990a). lined in Appendix 2. The Pebas fauna is characterised by

Although the taxonomic work is far from complete, it the predominance of sessile infaunal Pachydontinae and

semi-infaunal and Sub- appears that only two species found in the Pebas Forma- vagile epifaunal Cochliopinae. tion survive to the present day, viz. Hemisinus kochi strate-dependant taxa (Neritidae, Thiaridae, Pachychili-

? (Bemardi, 1856) and Mytilopsis sallei (Recluz, 1849) dae, Dreissenidae, Pebasia and Anticorbula) make up

(see Nuttall, 1990a). an estimated 4% in abundance, but representatives of -45 -

these found in 92% of the groups are samples, indicating firmgrounds or wood) was commonly available. that substrate in the Pebas system (e.g., shelly substrate,

Estimated abundance Relative abundance average Relative abundance Relative frequency

assemblage I1 II III IV V I II III IV V I II III IV V

n 64 52 78 83 8 64 52 78 8383 8 64 52 7878 83 8

Neritina ortoni group 0,2 0,6 0,2 0,5 0,0 16 33 20 31 0 39 56 5353 64 0

NehtinaNeritina roxoi group 0,7 0,0 0,00,0 0,1 3,4 31 5 4 7 53 44 12 12 14 50

Ampulariidae 0,0 0,0 0,0 0,0 0,40.4 17 0 0 0 83 6 0 0 0 13

19 Tall DyrisDyhs group 2,6 30,4 0,5 2,6 0,1 19 62 4 13 3 69 100 27 65 13

Small Dyris group 6,7 15,4 4,5 22,8 0,2 20 29 16 3232 3 98 98 92 100 38

Tryonia group 10,5 4,9 1,2 3,8 11,9 31 18 8 18 25 98 83 67 92 88

Sioliella group 2,2 1,61.6 1,3 3,4 4,94,9 19 18 14 2525 24 81 83 83 95 75

Other Other Cochliopinae 0,7 0,6 0,7 2,2 1,2 18 16 13 3333 19 73 63 59 95 50

Pachychilidae 0,9 0,6 0,1 0,1 7,6 27 11 5 4 54 50 25 17 10 38

Thiaridae 0,7 0,3 0,2 0,3 12,5 19 7 5 6 63 41 13 15 13 50

Perimarine taxa 0,0 0,0 0,1 0,1 0,0 0 0 32 6868 0 0 0 4 16 0

Pulmonata 0,1 0,2 0,00,0 0,0 23,2 4 6 1 0 88 13 13 3 1 63

Corbiculidae 0,0 0,0 0,0 0.20,2 0,4 2 8 7 33 51 2 4 4 12 13

Sphaeriidae 0,0 0,0 0,0 0,0 3,0 0 0 2 0 98 0 0 1 0 13

Dreissenidae 0,4 1,5 1,0 1,9 14,1 8 16 15 18 43 33 63 60 65 75

Hyriidae 0,6 0,1 0,1 0,2 7,6 20 6 5 11 58 34 15 12 23 50

Mycetopodidae 0,1 1,0 0,0 0,1 1,9 10 32 1 5 53 17 23 1 6 38

Pachydon obliquus 33,8 24,7 53,4 22,2 1,5 26 20 28 2323 3 100 92 96 99 25

other Pachydon group 39,1 17,4 35,4 37,637,6 6,0 26 20 22 2626 6 100 100 96 100 38

14 Pebasia group 0,3 0,4 0,60,6 1,1 0,0 14 18 22 42 4 34 48 49 83 13

Ostomya group 0,3 0,4 0,5 0,8 0,0 19 16 25 40 0 44 31 5050 77 0

Table 3. and indicator values of Pebasian molluscan Abundance, frequency (IV) assemblages. Estimated average abundance: sum

of the abundance of in all Relative mean a taxon samples of an assemblage, corrected for 100% total. abundance: percentageof

abundanceof a taxon in an the abundance ofthat average given assemblage over average taxon in all assemblages, expressed as %. Note that the relative abundance a is based on an octave-scale classification, and therefore occasionally differs from the esti- mated abundance Relative of average sample signature. frequency, percentage perfect indication (percentage of samples in an as-

where the taxon is semblage given present). Indicator value (IV): percentageof perfect indication based on combining the val-

ues for relative abundance and relative frequency (from Dufrene & Legendre, 1997).

Assemblage I: Tryonia assemblage; Assemblage II:tall Dyris assemblage; Assemblage III: small Dyris assemblage; Assemblage

IV: Pachydon obliquus assemblage; Assemblage V: Thiaridae-Pulmonataassemblage.

assemblage I1 II IIIIII IV V

n 64 52 78 83 8

endemics 86,0 91,6 97,3 93,3 14,4

substrate-dependent taxa 3.23,2 3,4 2,12.1 4,0 37,6

infaunal infaunal taxa 73,9 43,6 89,4 61,1 17,4

obligate freshwater taxa 2,4 2,2 0,4 0,90.9 56,6

obligate saline taxa 0,0 0,0 0,1 0,1 0,0

Pebasia group + Ostomya group 0,7 0,7 1,11.1 1,9 0,1

Neritina roxoi group + Pachychilidae + Thiaridae 2.22,2 0,9 0,3 0,5 23,3

Corbiculidae and perimarine taxa 0 0 0,2 0,30.3 0,4

Table 4. Some average properties ofthe assemblages. Numbers refer to estimated percentages. -46-

Figure 4. Example of a Tryonia assemblage residue. Length of valve in the centre (no. 7) is c. 10 mm. Sample 895 (Nueva Paleta,

the eucosmius & Napo: late Early-early Middle Miocene), containing snails Sheppardiconcha spp. (1), Charadreon (Pilsbry

de 1938 Olsson, 1935)(2), Neritina roxoi Greve, (3), the small Tryonia aff. scalarioides (Etheridge, 1879) (4), and Sioliella sp.

(5), as well as the bivalves Pachydon obliquus Gabb, 1869 (6) and P. carinatus Conrad, 1871 (7).

Middle domi- Figure 5. Example of a tall Dyris assemblage residue. Scale bar equals 10 mm. Sample 685 (Tamshiyacu: Miocene),

and P. natedby various species of Pachydon, eg., P. tenuis (1) obliquus (2). Many of thePachydon valves and fragments retain

remains of periostracum (skin). Furthermore, small cochliopines (Toxosoma eboreum (3), Sioliella crassilabra (Conrad, 1871)

(4)) and various Dyris ortoni (5) are present. -47-

Figure 6. Example of a Pachydon obliquus assemblage residue. Scale bar represents 10 mm. Sample 681 (Indiana: late Early-early MiddleMiocene), dominatedby single valves and articulated specimens of Pachydon obliquus (1). The elongate valves of other

of tenuis and P. cf. amazonensis species Pachydon (P. (2) (Gabb, 1869)are present. Gastropods, such as Dyris spp. (3) and Sio-

liellasp., (4) are rare.

Figure 7. Example of a small Dyris assemblage residue. Scale bar equals 10 mm. Sample 707 (Porvenir: late Middle-early Late

1874 Miocene), being very rich in species. Diplodon longulus Conrad, (1), Pachydon carinatus (Conrad, 1871) (2), P. tenuis

Neritina 1 ortoni 1871 (3), Pebasia dispar (Conrad, 1874) (4), Mytilopsis cf. sallei (Recluz, 1849), sp. (6), N. Conrad, (7), Dyris

Toxosoma is spp. (8) and sp. (9) are seen. Diplodon usually not present in samples assigned to the small-Dvra assemblage. -48 -

that sustained this Synecology appear therefore the most likely setting assemblage.

cluster Five assemblages were recognised by analysis.

Properties of the assemblages are listed in Tables 3 and 3 - Pachydon obliquus Assemblage (Figure 6)

4.

Composition — This assemblage is dominatedby species

of Pachydon, and P. obliquus in particular. Strictly

1 - Tryonia Assemblage (Figure 4) speaking, there are no indicator taxa (IV >30), but close Pachydon obliquus comes very to qualifying as

Composition — This assemblage is dominated (>10% such (IV 27). Cochliopine gastropods are rare compared abundance) by species of Pachydon (P. obliquus and to the other assemblages. Absent from this assemblage tall others) and of Tryonia. The only indicator taxon (indi- (IV 0-1), or nearly so, are Ampullariidae, the Dyris Thiaridae, cator value (IV) >30) is the Tryonia group. This assem- group, Pachychilidae, perimarine taxa, pul- blage is furthermore characterised by the comparatively monates, Corbiculidae, Sphaeriidae, Hyriidae and My-

Thiaridae common occurrence of Neritina roxoi-group + cetopodidae.

+ Pachychilidae. (Almost) absent from this assemblage Ecology — Endemicity is 97.3%. The Pachydon obli-

is almost made of infaunal (IV 0-1) are Ampullariidae, perimarine taxa, pulmonates. quus assemblage entirely up

Corbicula and Sphaeriidae. taxa (89.4%) and contains few substrate-dependent

and few freshwater taxa Ecology — Endemicity is 86.0%. The Tryonia assem- (4.0%) very stenotypic (0.4%). blage is dominated by infaunal taxa (74%), contains few Environment — Lacustrine. The abundance of infaunal substrate-dependant (3.2%) and stenotypic freshwater bivalves and the relative paucity of mainly epifaunal hy-

to absence of and taxa (2.4%). drobiids may point the aquatic plants,

Environment — Shallow lacustrine. The comparatively thus deposition below penetration depth of sunlight. This

to is indication of since is surfi- common occurrence of non-endemic groups compared no depth, light penetration the other assemblages (with the exception of the Thiari- cial in modem Amazonian waters (Furch & Junk, 1997).

the environment low dae-Pulmonata assemblage described below), is inter- Also, was very energetic. Judging

fluvial from the sediments from which preted as a sign of proximity to coastal settings or mostly organic-rich sam- influence. Given the comparatively large amounts of ples assigned to the Pachydon obliquus assemblage have small and rather delicate shells of Tryonia, the deposi- been collected, the depositional environment was char-

from acterised The abundance of tional environment was possibly protected waves by widespread dysoxia. ar-

of ticulated P. to the most and currents. The common occurrence organic matter obliquus may point tranquil (and

environments in the Pebas (inclusive of wood fragments) may be indicative of dense deepest) depositional system,

estimated to have been in the order of 10-20 metres (aquatic) vegetation (reed swamps, floating meadows, etc.). (Rasanen et ah, in prep.).

4 - Small Dyris Assemblage (Figure 7)

2 - Tall Dyris Assemblage (Figure 5)

Composition — This assemblage is dominatedby species

— and and the small Composition This assemblage is dominated by species of Pachydon (P. obliquus others) Dy-

and both the small Indicator taxa > the small ofPachydon (P. obliquus and others) ris group. (IV 30) are Dyris

tall The indicator (IV is the other 32), and Dyris groups. only taxon 62) group (IV 32), Cochliopinae group (IV which D. ortoni Pebasia and the (IV 31). The IV the tall Dyris group, comprises (Gabb, (IV 35) Ostomya group

1869) and D. lintea (Conrad, 1871). Absent from this for perimarine taxa is low (11), but clearly higher than in

the Neritina the other Absent from this assemblage, or assemblage, or nearly so, are roxoi-group, assemblages. the Neritina roxoi Ampullariidae, Thiaridae, perimarine taxa, pulmonates, nearly so, (IV 0-1) are group, Ampul-

Corbiculidae, Sphaeriidae and Hyriidae. lariidae, Pachychilidae, Thiaridae, pulmonates, Sphaerii-

Ecology — Endemicity is 91.6%. The abundance of in- dae and Hyriidae.

of — Estimated is 93.3%. This faunal taxa (43.6%) is clearly below average, that epi- Ecology endemicity assem- contains few faunal gastropods well above. This assemblage contains blage is dominated by infaunal taxa (61%),

few substrate-dependant (3.4%) and stenotypic freshwa- hard-substrate epifaunal (4.0%) and stenotypic freshwa-

ter taxa (2.2%). ter taxa (0.9%).

— Lacustrine some marine Environment — Often samples assigned to this assem- Environment (with rarely

The of blage are foundin comparatively coarse-grained deposits influence). common occurrence well-preserved of and (often rich in organic matter) with abundant evidence for fragile shells, such as small species Dyris Osto-

to conditions, that physical disturbance (e.g., waves and currents). mya points very low-energy might below fair-weather base. Shoreface environments (including submerged sandbars) occur e.g. wave -49-

5 - Thiaridae-PulmonataAssemblage (Figure 8) nate amounts of endemic Pebasian taxa. The assemblage

includes groups that can be found in present-day Amazo-

Composition — This assemblage is dominatedby species nian aquatic environments, such as ampullariids, pul- of Tryonia, Thiaridae, pulmonates and Dreissenidae. monates, freshwater cerithioideans, pearly freshwater

Indicator taxa (IV > 30) are Thiaridae (IV 31), pul- mussels and sphaeriids. monates (IV 55) and Dreissenidae (IV 32). Absent from Four of the five assemblages are dominated by

the and bivalves and this assemblage, or nearly so, (IV 0-1) are Neritina cochliopine gastropods pachydontine

both small and tall characterised rates of The ortoni group, Dyris groups, peri- by very high endemicity. mean marine taxa, Pachydon obliquus, Pebasia and species of faunal composition of these four assemblages shows

Ostomya. The other Pachydon group is rare and has a great similarity (similarity coefficients of average com-

low indicator value other between .64 and and their limits very compared to assemblages. position .94), are not

Ecology — Estimated endemicity is with 14.4% very low well defined. The fifth assemblage (Thiaridae-Pulmonata compared to the other assemblages. The Thiaridae- assemblage) is very different from the other four (simi-

Pulmonata assemblage also contains few (17.4%) infau- larity with the other four assemblages between -.01 and - nal taxa (compared to 67% for the overall fauna) but is .20). The endemicity of the Pebas fauna, and the wide rich in substrate-dependent (37.6%) and predominantly range of morphological ‘oddities’ point to an isolated

freshwater lake inland composed of stenotypic taxa (56.6%). long-lived aquatic setting (either or sea; see

Environment — Fluviolacustrine-deltaic (river channels, discussion below). Boss (1978) distinguished two eco- floodplain lakes and swamps, seasonally flooded forests) logical groups which he referred to as ‘moeities’, in an-

This contains subordi- cient lake molluscan faunas. and/or swamp. assemblage only (= long-lived)

Figure 8. Example of a Thiaridae-Pulmonataassemblage residue. Scale bar represents 10 mm. Sample 702 (Porvenir: late Middle-

Late is of Molluscs to freshwater early Miocene) composed mostly lignite/wood fragments. belong mainly cosmopolitan groups,

browni such as the pearly freshwater mussel Castalia sp. (1) and the freshwater cerithioideans Aylacostoma (Etheridge, 1879)

indet. Toxosoma and (2) and Sheppardiconcha sp. (3). sp. indet. (4) Tryonia scalarioides (5) are considered to be Pebasian en-

demics, and make up only subordinate amounts in samples assigned to the Thiaridae-Pulmonataassemblage. -50-

consists of endemic elements is One mostly and re- 1998). Overlying lignites are sometimes separated from

stricted lakes the second contains various the interval thin to proper, underlying sandy by a (<0.5 m) organic-

amounts of molluscan species with a wider distribution rich clay interval, representing slack deposition in back-

and is found in peripheral habitats of the lake. The four barrier settings. Evidence for occasional emergent con-

lacustrine assemblages contain of endemics ditions (i.e rare palaeosols, mostly root is found percentages ., traces)

from 86.0 to 97.3% and in the of these Almost ranging (estimated abundance), top parts parasequences. every

therefore be Boss’s lake The is surface. This may assigned to moeity. parasequence capped by a Glossifungites

Thiaridae-Pulmonata assemblage, with its low percent- ichnologic feature is related to burrowing (in this case of

endemics of estimated age (14.4% abundance) typically thalassinidean shrimps) in compacted substrates in meso-

falls into Boss’s peripheral moeity. The lacustrine moeity polyhaline settings and is often associated with transgres-

(97% of the samples) predominates markedly within the sive erosion surfaces (Rasanen et al, 2000).

studied faunas. Less common in the Pebas Formation, but neverthe-

less occurring regularly, are channel-like structures.

These channels (up to c. 10 m in depth) are characterised

Molluscan distribution and depositional environ- by inclined heterolithic stratification, and commonly in-

ments clude structures that indicate tidally influenced

subaquatic deposition (Rasanen et al, 1995, 1998). Tidal

facies envi- have been documented from the Pebas Sedimentary (and interpreted depositional deposits upper characterised in order ronments) are shortly to provide a Formation {i.e., in intervals also containing marine paly-

framework for of molluscan distribution in exploration nofloras and faunas; see Rebata, 1997; Rasanen et al.,

the different depositional environments represented by 1998). Tidal deposits have now also been recognised in

the Pebas system. the middle and lower portions of the Pebas Formation

(Rasanen & Irion, unpubl. data).

As a whole the sedimentary facies occurring in the

and Sedimentary facies ichnofossils Pebas Formation point to shallow, open aquatic (brackish

to shallow marine) settings with intermittentshoals and

Pebas Formation deposits are dominated by small-scale occasionally beach/swamp/deltaic depositional environ-

(3-7m) generally coarsening-up parasequences (Rasanen ments. The difference between maximum flooding sur-

et al, 1998) reflecting drowning, maximum flooding, faces and the topmost part of the parasequences is usu-

and shoaling. An ideal parasequence overlies sand or ally between 3 and 7 m, which may give some indication

lignite strata. The basis of a parasequence is commonly a for the maximum depth of the system (although some

thin sandish interval, occasionally containing reworked metres should be added to take into account compac-

bioclasts such shells and wood maximum horizons as fragments, overlying a tion). Organic-rich flooding may

wave ravinement surface. Commonly, this zone is char- reflect periods of extensive drowning and larger maxi-

acterised by marine trace fossils, such as Asterosoma and mum depths (probably in the order of a few of tens of

(well-developed) Ophiomorpha (Rasanen et al, 2000). metres).

These ichnofossils in the Pebas Formation are large, and

the ichnodiversity is high compared to ichnodiversity in

non-marine settings. Upwards, the sediments turn rapidly Molluscan distribution

into clays that are interpreted as maximum flooding in-

tervals. These clays usually contain in situ (or nearly so) For seventeen outcrops sedimentological and palaeon-

molluscan faunas (bivalves found in life position, either tological data have been combined. Sixty molluscan

articulated or isolated valves found in close proximity). samples from these outcrops could be assigned to sedi-

Often these clays directly overlie the basal contact, re- mentary facies, enabling a comparison of molluscan dis-

sulting in completely CU (coarsening-up) parasequences. tribution and sedimentary facies (Table 5).

The clays grade upwards into lower shoreface clay- Molluscs are commonest at the base of parasequences

shoreface sand alternations and are topped by upper (intervals overlying ravinement surfaces, maximum

sands or lignites. Upwards-coarsening intervals of the flooding and lower shoreface intervals), but are quite rare

sedimentary succession are consistently burrowed by in upper shoreface intervals.

brackish/marine ichnofauna, to a de- the base of locally very high Samples directly overlying a parase-

gree. Rarely to moderately bioturbated mud intervals are quence (usually a wave ravinement) are dominated by

indicative of stressed (restricted) conditions (Rasanen et Pachydon obliquus (42.1%) and the other Pachydon

al, 2000). Shelly faunas are not very common in these group (28.2%). Pebasian endemics dominatethese sam-

of but coarsening-up portions the parasequences. ples (96.5%), a few terrestrial and strictly freshwater

Parasequences can be incompletely or imperfectly groups (3.0%) occur as well. Samples from the maximum

Most developed. of the sands in the upper portions ofthe flooding intervals (including ‘organic-rich’ maximum formed shorefaces parasequences were along and occa- flooding intervals, as well as the lowermost shoreface

sionally they represent beach settings (Rasanen et al. intervals) are very similar to the samples at the base of -51 -

intervals the parasequence. In the maximum flooding intervals, the maximum flooding (0.2% vs 3.0%). Sam-

from maximum intervals molluscan faunas contain more small Dyris (15.5% vs ples flooding are completely

6.5%), and fewer Tryonia (2.2% vs 6.2%) and tall spe- dominated by Pebasian endemics (99.0%), even more cies of Dyris (3.3% vs 9.4%, see Table 5). Terrestrial than in the samples at the base of the parasequences. and strictly freshwater taxa are almost entirely lacking in

maximum wave maximum upper revinement flooding shoreface

n 14 40 5

Neritina ortoni Neritina ortoni group 0,4 0,9 0,7

Neritina roxoi Neritina roxoi group 0,2 0,0 0,0

Ampulariidae 0,1 0,0 0,0

Tall Dyris group 9.49,4 3,3 21,2

Small Dyris group 6.56,5 15,5 16,3

Tryonia group 6,2 2,2 6,4

Sioliella group 2.02,0 3,4 2,3

Other Cochliopinae 0,7 1,5 1,9

Pachychilidae 0,9 0,1 0,1

Thiaridae 0,2 0,0 1,4

Pulmonata 0,1 0,0 0,0

Dreissenidae 0,5 0,8 0,7

Hyriidae 0,5 0,1 0,0

Mycetopodidae 1,01.0 0,0 0,1

Pachydon obliquus 42.142,1 43,8 19,9

other Pachydon group 28.228,2 27,0 28,0

Pebasia group 0,4 0,6 0,4

Ostomya group 0,5 0,8 0,6

Table 5. Distribution oftaxa in sedimentary facies. Numbers are estimated percentages.

Only five samples were available from upper shoreface upper portions of parasequences in Pebas Formation out-

intervals. the of of The of molluscs in These show occurrence some (1.4% crops. comparative rarity (upper)

estimated + Thiaridae shoreface that make amount of abundance) Neritina roxoi-group intervals, up a significant

+ Pachychilidae, are considerably enriched in the Pebas deposits, appears to be a feature characteristic

snails in the other of the entire Pebas Formation. Given the cochliopine (44% vs 25-26% assem- common pres-

blages, especially for the large Dyris group), and de- ervation of sedimentary structures in these intervals, with

prived in pachydentine bivalves (44% vs 70% for the only rare bioturbation traces, the lack of molluscs can be

other assemblages). The comparative low abundance of explained in terms of high sedimentation rates, physical

is due to the low abundance of disturbance of stress. The pachydontines only sediments, oxygen or salinity

Pachydon obliquus. ichnofauna indicateselevated salinities as a possible rea-

for The bases of parasequences contain mainly samples son the lack of molluscs. However, salinity stress

assigned to the Tryonia assemblage, but also samples does not seem to be a good enough explanation, given

wide assigned to the small Dyris and Pachydon obliquus as- the absence of molluscs with a range of salinity

semblages do occur. Maximum flooding intervals are tolerances (including some ‘marine’ molluscs) in samples

dominated by samples assigned to the small Dyris as- studied from other parts of the Pebas Formation. With

semblage and the Pachydon obliquus assemblage. Only common indications for physical reworking, neither does

five samples from upper shoreface intervals were avail- oxygen stress appear to be a reasonable explanation. Ei-

sedimentation able. Two ofthese belong to the small Dyris assemblage, ther high rates, or common physical sedi-

one to the Pachydon obliquus assemblage and two to the ment disturbance (or both) should explain the relative

of tall Dyris assemblage. However, samples dominated by paucity of molluscs in the upper parts parasequences.

tall species of Dyris were commonly encountered in the Figure 9 summarises the distribution of assemblages in -52 -

the different environments of the Pebas information the depositional yield on aquatic chemistry, provenance system. areas and climate during shell growth. The preservation

of Pebasian molluscs is extraordinary (Vonhof et al,

them suitable for this of Geochemistry 1998), rendering especially type

analysis. Sediment geochemistry provides additional

During their life, molluscs record environmental infor- clues for the interpretation of depositional settings, mation in the chemical oftheir shells. Shells water composition provenance areas and chemistry.

9. Inferred distributionof Note that near-coastal in channel scours, indicated a dashed Figure assemblages. dysoxic settings (e.g., by contain the that otherwise dominatesthe ofthe Pebas line) may Pachydon obliquus assemblage, deepest parts system. Shelfc. 5-

10 m deep, lake bottom c. 10-?30m deep.

Isotope geochemistry values in molluscan shells compared to the calculated

water values as evidence for the lacustrine character of

I8 Pebasian depositional environments. 6 0 values were Strontium and stable isotope data published by Vonhofet

found to be very negative (usually in the range of -6 to - al. (1998) indicate a freshwater depositional environment those of 10 %o), incursion levels were higher (in the for the upper Pebas Formation (Grimsdalea Interval of -2 to -5 %o). These negative values of the Zone of Hoorn, 1993), with the exception of incursion range very

Pebas shells agree well with a freshwater nature of the levels, where maximum salinities were calculated to Pebas concluded from strontium between 3 and 5 Strontium data from system as analyses. range psu. isotope An additional 109 molluscs from seven samples were the middle and lower Pebas Formation (Crassoretitriletes I3 ls <5 6 Acme and Psiladiporites/Crototricolpites Concurrent analysed for C and O. Data are shown in Appendix

4. As show Range zones of Hoom, 1993, respectively) confirm the a whole, measurements (Figure 10) some

n I8 2 correlation between 6 6 freshwater nature of the Pebas system. Pebas lake waters C and 0 (r 0.41; standard

86 87 / Sr admixture of of the estimate and the values were, according to ratios, an pre- error 1.55), isotope are dominantly Andean fluvial input (c. 70-80%) and shield- negative (typically between -4 and -10 %o). The isotopic derived fluvial input (c. 20%). Based on end-member signature of molluscs from the seven measured levels isotope water compositions inferred from strontium iso- broadly overlap. Specimens belonging to species that all

of Pebas Formation tope compositions upper molluscs, samples have in common (Pachydon obliquus and P.

Vonhof et al. (1998) calculated expected stable isotope tenuis) confirm the generalised isotopic signatures of

for the the Pebas compositions waters in system. They samples as a whole (Figure 10). interpreted a large positive excursion of observed 6' C -53-

Figure 11. Stable isotope profiles through four shells. On the horizontal axis the growth increment number is indicated, going from

juvenile to adult from left to right.

Species-specific fractionation appears therefore subordi-

of nate as a source isotopic variation between samples,

although more measurements are needed to confirmthis.

If the overall isotope signature of the Pebas shells

freshwater points to settings, is it possible that we might

have missed salinity changes on a seasonal or even finer

scale? In order to address this question we have analysed

isotope chemistry ofshells along growth increments.

O0 O Co O CO COc° CO co 18 13

QQ QD D Q

cC CD C CD

Anodontites trapesialis -4,2 8,3 -13,1 7,0 38

Diplodon sp. -7,0 4,9 -13,9 2,9 36

Diplodon longulus -8,6 2,1 -12,1 2,3 22

Pachydon erectus -6,0 2,5 -10,1 8,4 20

various species (fossil) -6,0 -7,2-7,2 110

Stable Figure 10. isotope signature ofspecies ofPachydon (P. all recent shells -6,4 -8,3-8,3 74

tenuis and P. filled vs those of other obliquus circles) all fossilfossil shells -5,6 -13,5 154 species (open circles). No species-specific isotope signa-

ture differences can be observed. Both species of Pachydon

occur in all assemblages measured. Table 6. Summary of stable isotope measurements. -54-

this For pilot study, we have measured two fossil (Peba- onstrated. Otherwise pyrite is very rare or lacking. Sid-

and modem shells. erite is abundant in the Pebas The sian) two (Amazonian) Data are Formation. scarcity or shown in Table 6, Appendix 4 and Figure 11. The mean complete lack of pyrite and the abundance of siderite in

18 of the fossil shells is of the Pebas Formation 6 0 two slightly more negative (- most samples provides an argu- brackish conditions 7.3 %o) than those of the modem shells (-5.6 %o). The ment against during deposition. Only

18 in specific circumstances can siderite be formed in seasonal S 0 amplitude of the fossils (2.0 - 2.5 %o) is very marine or brackish settings (Martin, 1999). Pyrite and smaller than in the extant shells (4.9 - 8.3 %o). Average

13 siderite are diagenetic minerals, so it cannot be excluded 6 C for the fossil shells is about -11 %o, and those of the that their formationreflects rather than I8 post-depositional modem shells -13.5%o. The lower seasonal 6 0 ampli- syndepositional chemical settings. tude suggests that the Pebas system was chemically more stable than modem aquatic environments in Amazonia.

Residence times of waters in the Pebas system were longer than those of the present-day Amazonian flood-

n plains, based on the enrichment of 3 C isotopes (Vonhof

,8 et al., 1998). The very negative 5 0 values clearly show that the modem climatic regime (the so-called Amazon hydrocell) existed during deposition of the Pebas Forma-

and evidence from tion, argue (added to Sr isotope analyses) against saline conditions. Combined with the

ls low seasonal amplitude of the d O for the fossil shells this excludes seasonal salinity changes due to excessive

(seasonal) evaporation in a supposed closed Pebas sys- tem.

Sedimentgeochemistry

Twenty-seven samples from the Pebas Formation have been analysed for clay mineralogy (Figure 12). Pebas

Formation clay mineral assemblages are dominated by low-charged smectite. Kaolinite is rare, and illite is very rarely encountered in Pebas Formation sediments. No differences in clay mineralogical composition have been found between samples from differentstratigraphic levels of the Pebas from different facies. of Formation, or Many the smectite clays of the Pebas Formation contain in situ articulated valves of Pachydon. The co-occurrence of the smectite with these obligate aquatic taxa indicates that smectite is allochthonous to the Pebas Formation. Smec- tite can form as a result of weathering of a variety of

rocks in drained such source (sediments) poorly settings, Figure 12. Clay mineralogical signature of selected Pebas as Rivers smectite-rich floodplains. draining Neogene samples and a modem floodplain sample. deposits in western Amazonia, such as the Jurua and Pu-

smectite- dominated mineral rus, yield a clay assemblage resembling that of the Pebas Formation (Figure 12). Discussion

These modem river samples contain both high- and low- charged type of smectite. The latter thus probably origi- The depositional environmentof the Pebas system nated in poorly drained Andean floodplains made up of comparatively immature sediments. Mineral assemblages To start with, we shall summarise and discuss the palae-

(Kronberg et al, 1989; Hoorn, 1994b) as well as the ontological, geochemical and sedimentological charac- strontium isotope signature (Vonhof et al, 1998) under- teristics of the Pebas Formation, as well as implications line the Andes as a predominant source of sediments and of these for the reconstruction of Miocene Amazonian water in the Pebas system. The fine-grained smectite palaeoenvironments. Secondly, we shall compare these clays are responsible for the unusual preservation of fos- data with several characteristics of modem environments sils in the Pebas Formation. to which the Pebas system has been compared in previ-

Pyrite does occur in some intervals in the upper Pe- ous works. Finally, we shall synthesise the depositional bas Formation where marine incursions have been dem- system of the Pebas Formation, and link Miocene land- -55-

evolution in Amazonia biotic evolution confined lake-bottom result of scape western to were to sediments, as a of the area. high organic input and decomposition, and affected the

water column only marginally. Cochliopine snails might

have avoided unfavourable conditions by clinging onto

characteristics the Pebas A summary of for system erect or floating vegetation.

The molluscan fauna of present-day Amazonian * Molluscs aquatic environments is notably poor in species. The

flooded seasonally parts of the floodplains are charac-

The and diver- terised abundance extremely high endemicity morphological by an of ampullariid and pulmonale

sity of the dominant hydrobiid snails and snails and bivalves pachydontine sphaeriid (Irmler, 1975; pers. obs.). bivalves is clear indication of the a evolutionary longev- Floodplain lakes that do not suffer periodic anoxia, as

of the Pebas The of the well smaller ity system. general signature as channels, harbour mainly unionoid bi- fauna is similar to that of the modem Caspian Sea. The valves. These faunas are markedly different from the

latter fauna is dominated diverse by a highly and en- Pebas fauna which is characterised by the predominance

demic stock of (otherwise marine) limnocardiidbivalves of pachydontines and cochliopines, and the (near-) ab- and hydrobiid gastropods, and contains common dreisse- sence of ampullariids, freshwater pulmonates and nids and neritids. The is Caspian Sea currently an inland sphaeriids. Based on the molluscan faunas alone it is

with salinities between and sea, ranging 0 c. 16 psu. possible to exclude widespread floodplain conditions in

There is a balance between inflowing rivers and evapo- the Pebas system, with the exception of depositional in- ration, and this maintains this salinity. The Caspian Sea tervals containing the Thiaridae-Pulmonata assemblage

is now not connected to the oceans, but in its Neogene- faunas.

it used be Quatemary history to at various times. In 52% A comparatively high diversity and endemicity ofthe of the from the Pebas Formation samples stenotypic Pebasian molluscan faunas is at odds with a saline inland

freshwater molluscs often in life This occur, position. sea, apart from an evolutionary long-lived inland sea, or contradicts a supposed brackish origin for much of the an estuarine origin. These two environments are charac-

Pebas Formation that be from might interpreted the Cas- terised by low to very low faunal diversity, and by the

pian analogue. Furthermore, the few extant relics of taxa dominanceof widely distributed taxa.

dominating the Pebas fauna (the pachydontine Anticor-

and the Sioliella and * bula, cochliopines Dyris) all are Ostracods and other invertebrate fossils

freshwater forms (see e.g., Wesselingh, 2000), illustrat-

the of uniformitarian-based infer- Munoz-Torres al. recorded fauna ing uncertainty salinity et (1998) a of thirty- for environments. ences long-lived one species of ostracods from the Pebas Formation, all of

There are no representatives of cosmopolitan marine them endemic to the Miocene of western Amazonia.

and marine molluscan in the Pebas Forma- is flock of marginal taxa Cyprideis represented by a seventeen species,

other than at incursion intervals in its tion, upper part, making up over 90% of the individuals. Species of Cyp- where nassariid and rideis melongenid, pyramidellid gastro- occur in athalassic, brackish and hyper-saline envi- well the bivalves pods occur, as as Macoma (Vonhof et ronments worldwide. According to Whatley et al. (1996,

al., 1998; van Aartsen & Wesselingh, 2000) and p. 233) the Pebasian ostracod fauna is comparable only

Panamicorbula. Even the incursion intervals wide- faunas at to known from ‘brackish marginal marine or

spread marginal marine taxa such as oysters, mussels, athalassic environments of low energy’. Strictly fresh- and littorinids arcids, are wanting. In modem settings, water ostracods are rare in the Pebas Formation, as well

such is in of a pattern explainable terms salinity regimes: as species reflecting marine settings. The high diversity the marine marginal taxa that do occur in the Pebas For- of the genus Cyprideis in the Pebas is attributedto ‘evo-

mation have been from salinities reported around 3 psu lution in a stable, isolated body of saline water through-

and This well the higher {e.g., Rodriguez, 1963). agrees very out Miocene’ (Whatley et al., 1996, p. 233). Both with inferred maximum salinities of 3-5 psu for the in- ostracods and molluscs show high levels of endemicity, cursion intervals Sr using isotopes (see below). and are dominatedby groups whose extant members live

The of corbulids and the in predominance pachydontine a range of salinities. The abundance of Cyprideidae

paucity of corbiculidbivalves, a widely distributed has been used to for brackish group argue palaeoenvironments. in modem and Cainozoic freshwater environments of However, comparison with the marine-like Pachydonti- South America, may be explained by widespread dysoxia nae which have a single descendant widespread in fresh-

in the Pebas corbulid system. Many extant taxa are well water ecosystems, serve as a clear warning against using

to et adapted low-oxygen conditions (Beesley al, 1998; the highly endemic and extinct ostracods in a uniformi-

& Samtleben, Intervals dominated in situ tarian fact that has been Lewy 1979). by way to assess palaeosalinities, a

Pachydon obliquus are often composed of black organic- acknowledged by Whatley et al. (2000). Sr isotope ratios rich The other dominant clays. group, cochliopine gas- measured on a few of the Pebasian cypridid ostracods

is known for its tolerance of tropods, not low-oxygen (H. Vonhof, unpubl. data) show a freshwater signature

conditions. It is possible that low-oxygen conditions for them. -56-

Foraminifera (mostly Ammonia are to the spp.) commonly point presence of swamp forests rather than Terra

found in incursion intervals, but not have been recog- Firme rainforest withinthe Pebas system.

nised in other far. Barnacles found in samples so were Charophyte oogenia are common in the Pebas For-

two samples. barnacles in the Rodriguez (1963) reported mation, indicating common presence of shallow, low-

Lake Maracaibo to in salinities low occur as as 2 psu. energy waters.

* Fishes

* Plants and pollen

The fish remains of the Pebas Formation show a mixed

Based on composition of the pollen flora as well as on ecological picture. Strictly freshwater inhabitants, such

sedimentary Hoom 1994a, con- as cichliids structures, (1993, b) (P. Gaemers, pers. comm.) and piranhas cluded that the Pebas Formation had been deposited in a (Monsch, 1998) occur. Extant relatives of the most

fluvio-lacustrine modem white- abundant system comparable to Pebas fish taxa ( Sciaenidae, catfish) occur e.g., Amazonian water floodplains, with some marine influ- both in modem Amazonian freshwater and coastal South

ence. Most of the pollen taxa from the Pebas are American The of both samples ecosystems. presence myliobatid also present in slightly older fluvial deposits to the north, and dasyatid teeth throughout Pebas Formation deposits

the Mariname Sand Unit. Both floras domi- pollen are (Monsch, 1998; J. Lundberg, pers. comm.) suggests ma-

nated the same but the Pebasian flora is rine influence. Freshwater by groups, pollen rays (potamotrygonoids) are

considerably enriched in to the Mari- spores compared abundant in present-day Amazonia, but only very few

name flora (e.g., Verrucatosporites usmensis (van der remains from the Pebas Formation could be attributed

Hammen, 1956) and Psilamonoletes tibui van der Ham- with to this certainty family (J. Lundberg, pers. comm.). men, 1956) make up an average of c. 40% in the Pebas The combination of obligate freshwater and marine flora compared to only c. 10% in the Mariname is known from Lake Maracaibo flora). rays e.g. (Venezuela).

Such a difference well result from prolonged trans- Marine are and found into the may rays abundant, are up del-

in rivers that into lake environment. tas port emptied a Ro- surrounding the lake. The occurrence of marine rays bust survive spores transport better than less robust pol- in the Pebas Formation therefore does not indicate len (G. Sarmiento, comm.). Hoorn found brackish but also pers. (1993) settings during deposition per se may indicative of in be result mangrove pollen, perimarine settings, explained as a of a restricted connection of the

intervals in the Pebas two Formation, viz. in the Early- Pebas system to the sea. early Middle Miocene Psiladiporites-Crototricolporites

Concurrent Range Zone (c. 20-17 Ma), and in the late * Sedimentology

Middle-early Late Miocene Grimsdalea Interval Zone

12-10 The latter interval contains incursion (c. Ma). lev- The predominance of coarsening-up parasequences in the els that have been corroborated data and the Pebas Formation by isotope is at odds with a supposed fluvial origin of occurrence perimarine molluscs, foraminifera and of these deposits as proposed by Hoom (1993, 1994a, b). barnacles (Hoom, 1994a; Vonhof el ah, Other Fluvial are 1998). cycles usually fining-up, see e.g. Rasanen et indicators of marine influence do not accompany the al. (1998). Tidal deposits have been recognised in vari- of occurrence mangrove pollen in the lower Pebas For- ous parts of the Pebas Formation (Rasanen et al., in mation. Mangroves occur along the southern shores of inland basins mi- prep.). Although large may experience

Lake Maracaibo (Wesselingh & Rasanen, pers. obs.), crotidal regimes, the widespread tidal deposits in the where salinities close 0 Connection marine Pebas Formation are to psu. to suggest a common connection to ma- habitats thus be for the may more important appearance rine environments. The marine signature of the ichnofa- of than cies mangrove taxa salinity. Furthermore, mangroves supports such an interpretation, but contradicts indi- have been reported along riverbanks far inland, out of cations excluding widespread brackish conditions in the reach of marine influence (Morley, 2000). Hoom (1993) Pebas Formation from molluscs and isotope geochemis- used the occurrence of the fem Acrostichum in the Mid- try. The blueish-turquoise clays of the Pebas Formation dle Miocene Crassoretitriletes of Acme Zone the Pebas with common organic material are indicative of wide-

Formation (c. 17-12 Ma) to infer coastal plain settings. spread bottom dysoxia.

There marine molluscan are no species, isotope data or

* mangrove pollen to support such an interpretation. Ac- Sediment geochemistry rostichum have also been observed along freshwater lakes in Indonesia (B. Morley, pers. comm). Remains of Clay composition in the Pebas Formation is completely

and trees in the Pebas Forma- dominated plants are not uncommon by low-charged smectite that presumably re- tion. In few of sulted a cases shallow-rooting stumps trees were from weathering of volcanic source material in the found, indicating presence of (swamp-) forests. Pol- poorly drained floodplains of rivers draining the Andes. len of Mauritia-like palms are common in the Pebas A predominantly Andean sediment source in the Pebas flora. These palms are typical of parts of floodplains system is also shown by heavy mineral assemblages which All these data experience prolonged flooding. (Kronberg et al, 1989; Hoom, 1994b), as well as the -57-

strontium isotope signature (Vonhof et al, 1998). The lived lakes are located in (semi)enclosed drainage basins, almost lack of kaolinite in the Pebas indicates often complete in tectonically active areas. These lakes are usually

the absence of Terra Firme habitats. The of and stable: residence times scarcity py- deep chemically water are rite (apart from incursion levels) and the abundance of long. They usually contain freshwater, but the geochemi-

siderite in Pebas Formation freshwater deposits point to cal composition may vary depending on evaporation re-

settings, although it cannot be excluded that these reflect gimes in, and influx of dissolved salts from, the catch-

conditions. post-depositional ment area. Many of the long-lived lakes experience

stratification, commonly depriving deeper parts from

* Baikal notable Isotope geochemistry oxygen (Lake being a exception). No

visible tides have been reported from these lakes. Lon-

7 The strontium of Pebasian shells is 6 signature reflects, apart gevity typically in the order of magnitude of 10 -10

from times when marine incursions did reach the area, a years. The invertebrate fauna of long-lived lakes is

predominantly Andean freshwater component and a mi- largely endemic and diverse (both in numbers of species shield freshwater nor component (Vonhof et al, 1998). and adaptive traits) and has originated from freshwater

The incursion level in Vonhof et al. ancestors published (1998) (Michel, 1994, 2000). The presence of species

was deduced to be a mixture of shield freshwater and flocks is a typical featureof these lakes.

marine with maximum salinities waters, not exceeding 5 Long-lived lakes and the Pebas system share their psu. Biogeographic evidence (Nuttall, 1990a; Vonhof et (taxonomically and morphologically) diverse, endemic

al, 1998) points to a northerly (Caribbean) origin of flocks of hydrobiid species. The high diversity of en-

marine influence. Based the on Sr isotope signature of demic bivalves and ostracods of marine/brackish origin

the the incursions found in the in the Pebas shells, upper Pebas For- system (Pachydontinae and Cyprididae) is at

mation must have entered through shield areas. Vonhof odds with a long-lived lake environment, as are other

al. et (1998) concluded that the Apoporis region in Co- marine indicators such as marine rays, mangroves and

lombia (Hoorn, 1994b) was the most likely pathway of marine ichnofossils.

marine invasions in the Pebas system from the Llanos

Basin. * Caspian Sea

et Vonhof al. proposed that a marked enrichment of

13 5 C compared to expected values (those of modem The Caspian Sea is a unique system: it is evolutionarily Amazonian rivers) is an indicationof a lacustrine palaeo- long-lived and brackish. The Caspian Sea is located in an environment. This is furthermore the lower supported by isolated drainage system, but was occasionally connected 18 of 6 0 variation in the fossil shells amplitude (seasonal) in its to the Salinities from geological past oceans. range I8 to the modem shells. In Pebasian 3 0 in the northern compared general, 0 psu part and near river entrances to c.

values between -4 and -8 %o 16 in the southern of the are negative (typically psu parts sea. Hypersalinity is

I8 in the balance of PDB). These values resemble the low 3 0 values found common lagoons surrounding sea. A

in modem western Amazonian freshwater bodies as a influx of river waters with low concentrations of dis-

solved salts and overall result of extensive recycling of rainwater (‘the Amazon an evaporative climatic setting ls maintain the salinity. No stratification in hydrocell’). These very low 3 O values argue for similar oxygen appears the Caspian Sea. Microtides have been demonstrated. wet climatic regimes in the Pebas, and exclude an inland

The Caspian Sea has existed at least for some 10 million sea setting whose salinities were maintainedby evapora-

The invertebrate fauna of the Caspian Sea is com- tion (Kaandorp et al, 2000). years. pletely dominated by systematically and morphologically

diverse endemic clades that have originated from both

A comparison ofthe Pebas system with modern systems freshwater and marine ancestors (Dumont, 2000). The

latter group sets the Caspian Sea fauna apart from long- In the literature Nuttall, (eg., 1990a; Hoorn, 1993, lived lake faunas. Only in (the vicinity of) delta areas do

Rasanen 1994a, b; et al, 1998; Whatley et al, 1998; more widely distributed, stenotypic freshwater taxa

Vonhof the Pebas has been abound. The et al, 1998), system com- Caspian Sea and the Pebas system share of modem environments. pared to a variety depositional their overall molluscan compositions dominated by hy-

Here, we briefly characterise these depositional systems, drobiid snails and endemic bivalves of marine origin and list similarities and dissimilaritiesbetween them and (Limnocardiidae v.v Pachydontinae). Furthermore, both the Pebas system. systems commonly contain dreissenids and neritids. The

ostracod faunais similarly endemic and diverse.

* Long-lived lakes However, stenotypic freshwater molluscan species,

found commonly in situ in the Pebas Formation, are re-

ofmodem lakes include Examples long-lived giant lakes, stricted to marginal (deltaic) environments in the Caspian

18 such Lake Baikal and Lake The as (Russia) Tanganyika (Af- Sea. very negative 0 signature of many of the Pe- and small such Lake Ohrid rica), ones as (Balkan). Long- bas shells is incompatible with salinities maintained by -58 -

evaporation; this should have led to considerably higher tidal deposits. These two systems differ, however, in the

molluscan faunal and oxygen isotope ratios. signature (taxonomic morphologi- cal diversity, high endemicity in the Pebas). Furthermore,

* freshwater molluscs Saline inland lakes or playa lakes the in situ occurrence of stenotypic

is only known from the uppermost (freshwater) reaches

Examples of these are found in (semi-)desert areas of estuaries, but common in Pebas Formation deposits.

ls worldwide. Saline inland lakes are located in isolated The Sr signature and the commonly highly negative 5 O

basins. of these lakes condi- drainage Many dry up entirely signature in the Pebas are at odds with brackish

from freshwater I3 during the dry season. Salinity varies, (in tions. Enrichment of 6 C in Pebas molluscs indicates times of massive influx of runoff), to saline and hyper- residence times longer than those in estuaries. Estuarine saline. Tides are absent. Stratification leading to oxygen indicators such arcids and littorinids as oysters, mussels, in the lake does despite their shallow depletion occur, are lacking in the Pebas. Pyrite is commonly formed in nature. Sediments deposited in these systems often con- estuarine settings, but is very scarce in the Pebas Forma- tain abundant minerals. Saline inland lakes evaporitic are tion. 3 5 usually geologically short-lived (10 -10 years), and

exist seasonal scale. The invertebrate many only on a * Lake Maracaibo faunas of these lakes are typically very low in species numbers. These taxa are morphologically conservative, Lake Maracaibo in western Venezuela is a large lake

2 and often widespread in other, similar systems. Occa- and connected the (14,344 km ) located at sea level, to such sionally marine indicators, as foraminifera, occur sea through the Maracaibo straits (Rodriguez, 1963). inland in this ofenvironment even far type (Gasse et very Another, similar coastal system is the Cienega Grande of al., 1987; Patterson et ah, 1997). northern Colombia(von Cosel, 1986). Lake Maracaibo is

et al. that the delicate Whatley (1998) suggested located in a tectonically active (subsiding) area. The wa- of Pebasian ostracod shells could be attrib- preservation ters in the lake are mainly derived from the catchment uted to deposition in an inland saline lake. The foramini- area. Some direct precipitation and some influx of ma- fer Ammonia and ‘marine’ bivalves and gastropods have rine waters also occur, resulting in a freshwater to oligo- been reported from Saharan salt lakes (Gasse et al, haline lake (Holmden et al., 1997). The lake is not very

them to the Pebas 1987), making a potential analogue to well deep (only some 35 m) and appears be rather the of the fauna of system. However, general signature With 10 et water oxygenated. years (Holmden al., 1997), saline lakes and the Pebas system is different. No very residence times are higher than in fluvial or estuarine

highly diversified endemic faunas are known from saline environments, but much lower than in long-lived lake inland lakes (apart from the Caspian Sea, discussed Microtides in the which environments. are present lake, saline conditions in the Pebas above). Furthermore, sys- into existence with the came only some 10,000 years ago, I8 are defined the 6 0 of of The molluscan tem by highly negative signature rising sea level the last deglaciation. the shells and the of in common presence situ stenotypic fauna of Lake Maracaibo is dominated by a few widely

freshwater taxa. from secondary Apart possibly gypsum distributed species of generalist groups, occurring found in Pebas Forma- at a few levels, evaporites are not worldwide in tropical marginal marine and freshwater

tion and the tidal known from the Pe- deposits, deposits environments, viz. Polymesoda spp., Neritina spp. and

bas Formation are unknown from saline lake environ- the gastropod Melanoides tuberculatus (Muller, 1774)

ments. docu- (Rodriguez, 1963; pers. obs.). Rodriguez (1963)

mented faunal dines in the Maracaibo straits related to

* Estuaries salinity regimes. For instance, he found that common

marine indicators, such as thaiid and littorinid snails as

Estuaries of of river- and are zones constant mixing sea- well as mussels and oysters, occurred in salinities typi-

water. Often, occurs predictable gradients Maracaibo Lake and the Pebas mixing along cally exceeding 5 psu.

riverwater seawater et The from to (Eisma al, 1976). system share the presence of mangroves, the combination size and water residence time varies, depending on con- of freshwater and marine stingrays and their tidal signa-

from to weeks. estuaries Both of figuration, days Many experi- ture. systems are predominantly composed wa-

tidal ence strong to very strong regimes. They are usually ters received from their catchment areas but Lake Mara-

the last 2 estuaries are short- well oxygenated. For Myr, caibo experiences some marine influence as shown by

3 4 lived geological phenomena (10 -10 years), linked to low salinities {e.g., Rodriguez, 1963). of level rise. The invertebrate faunas of short periods sea The species-poor, cosmopolitan nature of the Mara-

estuaries are characterised as specimen rich, but species caibo molluscan fauna is in sharp contrast to the diverse

poor. Typically, non-endemic taxa of widespread genera endemic Pebas fauna. No strictly freshwater taxa, such as

dominate. unionoids, have been recognised during brief visits by

The Pebas system and (tropical) estuaries have abun- the senior author to the lake in 1996 and 1997. Sulphur is

dant in well the of in Mara- hydrobiids common, as as presence very common in reductive deltaic sediments the

euryhaline fish taxa, mangroves, marine ichnofossils and caibo system, which should translate in the formationof -59-

pyrite. Pyrite is very rare in the Pebas Formation, where tonically induced accommodation and rapid sediment

siderite dominates. infill drove the depositional system. The aquatic system

was driven by precipitation in the catchment area and the * White-water floodplains of the Amazon system (‘var- lake itself, though there must have been connection to

zea’) marine The settings. Pebas system contained shoals and

The of the swamps but lacked extensive emerged environments. At floodplains main Amazon River (known as times, marine incursions invaded this lake. varzea) are mainly aquatic depositional environments.

The active Andean loaded with sediments from the floodplains are usually tens of kilometres rivers, emerging

hinterland, formed an extensive wide, and experience almost total inundation during the likely swamp-delta

on the western and southern of the lake wet season. During the dry season (difference between fringe margins

From the east, rivers draining shield areas and highest and lowest water levels are in the order of 10 m system. low-lying forests entered Lake Pebas. This in many places) the rivers become well defined by their tropical type

levees. The of the in the is of rivers carry little sediment. Nevertheless, Rasanen major part varzea dry season et al. record deltaic occupied by floodplain lakes and by seasonally flooded (in prep.) some structures of an easterly origin in the Pebas Outflow of freshwater forests. Varzeas have been extensively documented by system. and marine incursions towards and from north of Sioli (1984) and Junk (1997). Residence times of waters were areas the short. Rivers and Pebas system (Hoorn et al, 1995; Vonhof et al, are inundatedfloodplains are well oxy-

1998; et al, It cannot be excluded that genated. In floodplain lakes seasonal anoxia is common Lundberg 1998). marine connection with the Pacific (Junk, 1997). The chemical composition of the water is some existed at that time entirely dominatedby the Amazon River. Like Lake Ma- through south-central Ecuador (Steinmann et al,

1999). Very little is known about the of the racaibo, the present-day varzeas originated at the begin- stability Pe- bas of the Holocene system in terms of water-level fluctuations. Water- ning some 10,000 yrs ago (Irion et al, depth fluctuations recorded in the in- 1995). The molluscan fauna of the varzea is dominated parasequences are reflect local freshwater and terrestrial such terpreted to as a result of tectonic by obligate taxa, as pearly drowning, and result of freshwater mussels, sphaeriids, ampullariids and pul- subsidence, shoaling as a progradation

monates (Rasanen et al, 1998), lake-level variation or even (Irmler, 1975; Wesselingh, pers. obs.). The

fauna is not and of eustacy (Hoorn, 1993;Rasanen et al, 2000). diverse, composed widespread spe-

cies. Lake Maracaibo is modem a good example of a pre- The Pebas and the share system varzea their highly cipitation-driven system connected to the sea. Lake Pe-

ls 3 O the Andean-dominated bas was far larger than Lake and have negative signature, geo- Maracaibo, may

chemical been from the sea two sills instead signature ofthe waters, and the common occur- separated by of one; between rence of Mauritia-like palms. The molluscan faunas of one Lake Pebas and the Llanos area, and a tem-

these two one between the Llanos and the Maracaibo systems are very different. Morphologically porary area

eastern Venezuela Basin. Tectonics rather and taxonomically diverse endemic groups dominate the or than eus-

Pebas tatic level controlled the influx of saline fauna, whereas the varzea fauna is species-poor, sea change wa-

ters into the Llanos Basin and entirely dominated by obligate freshwater taxa. No (Villamil, 1999), and also may

marine marine ichnofauna tidal de- have controlledoverflow of the marine Llanos mangroves, rays, or (episodic)

known from into the Pebas Marshall & posits are the central and upper Amazon system (L.G. Lundberg,

varzeas. 1996).

Various terms have been applied to describe the set-

None of the above-mentioned modem environments is in which the ting Pebas Formation was deposited. The similar to the Pebas system. The Pebas system contains use of poorly defined terms has caused considerable con- evidence of marine influence (marine ichnofauna, tidal fusion amongst geologists and biogeographers, and marine deposits, mangrove pollen, while on the therefore discuss of rays), we some the names applied to the other hand there is evidence of the (ecologically) pro- Pebas system here. isolated longed nature of the Pebas system: the highly Floodplain environments (Room, 1993, 1994a; diverse and endemic nature of molluscan and ostracod Room et al, 1995) hardly seem an appropriate term for faunas. Strontium and oxygen isotopes and the common the Pebas system which was permanently aquatic with of freshwater occurrence obligate taxa point to predomi- minor and fluvial swamps influence, and was connected nantly freshwater settings. These conflicting indications to marine environments. To illustrate this: only 8 of 285 in the occur together same beds, so they are not ex- studied samples contained fauna comparable to (but not plained in terms ofsuccessive facies changes. identical with) modem floodplain faunas.

The of use the term seaway (Rasanen et al, 1995; Synthesis: landscapes and landscape evolution in Webb, 1995), despite the occurrence of marine influ- Miocene western Amazonia ences, also appears inappropriate to describe the Pebas

The Pebas and has caused lot of confusion system may be characterised as a geologically system, a (Hoom, 1996; L.G. Marshall long-lived aquatic system of huge size. Continuous, tec- & Lundberg, 1996; Paxton et al, 1996). -60-

13. A Lake The Andes to the shield in Figure palaeoenvironmental impression of Pebas. left, the Guyana the upper right-hand cor-

ner. The width of the block diagram is c. 1,500km. Abbreviations:FBZ - Foreland Basin Zone, PCZ - Pericratonic Zone.

The notion proclaimed by Webb (1995) that Miocene Whatley et al. (1996) referred to an ‘inland sea’ to western Amazonia was a marine area, with sea arms suggest a hydrologically closed saline lake, whose salin-

not stretching south, east, west and north, is supported by ity was maintained by inflow of salts with rivers and data at all. evaporation in the lake itself. This was not the case in the -61 -

Pebas evidenced system as by e.g. the occurrence of somewhat; on the other hand, the possibility that the

18 extended mangrovesand the highly negative 6 0 signature. system during the Middle Miocene more east-

Neither is ‘coastal environment’ suitable than at the plain a term erly present acknowledged may put figure

2 somewhat A size of million km is to use for the entire Pebas system. The Pebas system was higher. 1.1 roughly

of with three times the size of the Sea. It is that merely a large body water islands, swamps and Caspian likely

shores. These be classified Lake Pebas much earlier than the late areas might as (non-marine!) originated Early

coastal environments. Estuaries orders of Miocene. Rather low-diverse, typical pachydon- plain are mag- yet nitudes smaller faunas known from the than the Pebas system. They are geologi- tine/cochliopine are Late Oligo- cene-earliest Miocene cally short-lived and do not have large bodies of water La Cira fauna of the Magdalena and Llanos basins in Colombia with comparatively long residence times as in the Pebas. (Pilsbry & Olsson, 1935;

They also lack highly endemic faunas. We have chosen Nuttall, 1990a; Guerrero, 1997). The Early Miocene La

the Lake Pebas. fauna of southern Colombia is to name system However, use of this Tagua a stratigraphically

is free from either. A lake intermediatefauna between the La Cira and Pebas faunas term not problems suggests no

connection stricto. The La Cira and La faunas to the sea (but see e.g., Lake Maracaibo), and sensu Tagua are char- acterised the does not describe perfectly the episodic brackish settings by occurrence of Pachydon hettneri (An-

in the Pebas There is modem the Pebas fauna the of system. no depositional derson, 1928), by occurrence

similar The fauna setting to the Pebas system, so arguably a new Pachydon obliquus (Nuttall, 1990a). La Tagua

environmental term might be considered. An appropriate was also found in cores from the Pastaza-Maranon fore-

term for the Pebas would be land basin below intervals Pebasian mollus- system ‘para-marine mega- containing

lake’ (‘marine-like megalake’). This lake happened to be can fauna s. str. (Wesselingh, unpubl. data). This implies

located to that the La and La Cira faunas adjacent the sea, and was thus perimarine as Pebas, Tagua occurred different well. during time intervals in the same system that

The must have at least 17 million Pebas system was located in a series of active spanned years. The system

foreland basins and adjacent pericratonic regions. This was probably continuously (to some extent) lacustrine:

of the (intermittent) establishment of fluvial marine type huge geographic systems, occupied by long-lived or pa-

water bodies of ‘abnormal’ salinities is known from the laeoenvironments would have eradicated the endemic

dominatedfaunas. geological record. e.g., the Neogene Paratethys of cen- Pachydon-

tral/eastem which the The earliest lake the La Cira Europe (of Caspian Sea is a relic) (containing fauna) occu-

and the Late Cretaceous Clearwater Seaway of northern pied a narrow foreland basin zone east of the North An- dean thrust front the latest America. Thus, on a large (subcontinental) tectonic scale during Oligocene, the deposits

the Pebas system is not unique. of which became later incorporated into intramontane

How then is it that basins. This lake the possible a system with some con- gradually expanded (south)east onto of the South nection to marine environments is not invaded by cos- edge American craton (e.g., Cooper et al,

mopolitan, marginally marine taxa? The combination of 1995; Villamil, 1999). It split into the Llanos

the freshwater nature of the system with widespread Lake/embayment in the north and Lake Pebas in the

dysoxia would have induced ‘ecological’ isolation. This south during the Early Miocene (Figure 14). During this

isolation insufficient to the at least two of was prevent some marine organ- time, region experienced periods ex- tensive marine isms (rays, ichnofauna, mangroves) to invade and estab- ingressions, one during the late Middle Miocene lish in the area, but apparently prevented the successful (c. 12 Ma: Vonhof et al., 1998) and the other establishment of the late Miocene with widespread marginally marine molluscs during Early (contemporary ma- rine and ostracods that should be competitively superior to incursions in the Mariname Sand Unit of Colombian

the endemic of the Pebas in More of species any ordinary, mar- Amazonia; see Hoom, 1994b). periods exten-

marine environment. sive marine influence could have ginal This issue deserves further occurred in this system, but study, and Miocene molluscan, ostracod and fish faunas have not been demonstrated so far.

The Pebas-Llanos from the Llanos Basin should provide excellent material system may well have played a

to elucidate this. role in the adaptation ofmarine biota to freshwater Ama-

The Pebas attained zonian al. calculated system a huge size (Figure 1). ecosystems. Lovejoy et (1998) that

Pebasian molluscs assigned to the Middle Miocene Cras- potamotrygonoid rays evolved from marine ancestors in

soretitriletes the Amazon the Acme Zone have been found in boreholes system during Early-Middle Miocene, as

based as far west as the Pastaza region in Peru (c. 75°30’W) on molecular-genetic divergence rates calibrated and far east with vicariance event of known as as Fonte Boa, Brazil (c. 66°°W). An even a geological age. This would more easterly extension cannot be ruled out. Faunas from comply with the age of the Pebas system. How-

this fossil finds interval are known from outcrops as far north as the ever, new of potamotrygonoid rays in the

lower in Colombia Pebas Formation cast doubts the estimates Caqueta region (c. 1°S), and as far on age of al. south as the lower Urubamba valley in Peru (c. 11°S). Lovejoy et (J. Lundberg, pers. comm.). The oldest The maximum size of the fossils of ‘marine-like’ Amazonian dol- system during the Middle Mio- some taxa (iniid

2 cene is in the order of magnitude of 1.1 millionkm . An phins, manatees, potamotrygonoid rays and the pachy-

form of dontine be of irregular the system may downsize this estimate Anticorbula) appear to Miocene age -62-

(Lovejoy et al, 1998; Lundberg, pers. comm.). Rainfor- to Lake Pebas, and should not solely be attributed to the est biota, apart from those of swamp forests, cannot have Andean montane or Panamanian seaway dispersal barri- flourished in western Amazonia during the existence of ers. Lake Pebas is thought not to have continued into the

Lake Pebas. The large size of the system, as well as its Parana-Plate river basins (Lundberg et al, 1998), allow- longevity, must have created a formidable barrier to dis- ing interchange of terrestrial biota along its southern persal of terrestrial biota, and have promoted separate shores. A possible southerly connection, however, is still evolution on the west and east sides ofthe lake. Possibly, subject of debate. The modem rainforest biota could only divergence patterns within terrestrial groups, such as e.g. start to develop in western Amazonia after fluvial Ama- snakes (Zamudio & Greene, 1997) and poison frogs zonian ecosystems had replaced Lake Pebas, somewhere

(Clough & Summers, 2000), may (in part) be attributable between 9 and 8 Myr ago (Lundberg et al, 1998).

Figure 14. Evolution ofLake Pebas.

Conclusions Five assemblages are described from the Pebas fauna.

One of these resembles modem Amazonian floodplain

Molluscan faunas from the Miocene Pebas Formation of but it makes 3% of the assemblages, up only samples

Peruvian Amazonia are characterised as almost entirely studied. Endemic pachydontine bivalves and cochliopine aquatic, endemic and extinct. The fauna is dominated by gastropods dominate the other four assemblages, which

in to pachydontine bivalves and cochliopine gastropods are assumed represent evolutionarily long-lived en- numbers of species and specimens. demic molluscan communities. -63-

The taxonomic composition of the molluscan fauna Acknowledgements and shell isotope geochemistry indicate that (open)

dominated the Pebas aquatic settings system. Sedimen- For all their help we thank Jose Arimuya (Bellavista

tological and isotope data, as well as specific fossil plant Nanay, Peru), Charles Barnard (Nationaal Natuurhis- and animal groups, point to widespread shallow condi- torisch Museum, Leiden, the Netherlands); Rafael Buni-

tions. not Swamps were uncommon in Lake Pebas. Due muya (INRENA, San Roque, Peru), Gerhard Cadee

the endemic and extinct character of the mol- to largely (Nederlands Instituut voor Onderzoek der Zee, Den luscan well the (as as ostracod) fauna it is difficult to Burg, the Netherlands), Alejandro Chaleo (formerly in the Pebas based assess palaeosalinities system on the OXI, Lima, Peru), Pieter Gaemers (Leiden, the Nether- uniformitarian approach. Isotope data from the shells lands), Gerald Ganssen (Vrije Universiteit, Amsterdam, show from that, apart brief periods when marine incur- the Netherlands), Edi Gittenberger (Nationaal Natuur- sions reached Amazonia and western conditions were historisch Museum, Leiden, the Netherlands), Javier

oligo-mesohaline, Lake Pebas was composed of (mainly Guerrero (UN, Bogota, Colombia), Henry Hooghiemstra freshwater. The Andean-derived) occurrence of (rare) (Universiteit van Amsterdam, the Netherlands), Hugo the freshwater in than half of the boatsman strictly taxa more samples (Iquitos, Peru), John W.M. Jagt (Natuur-

supports a predominantly freshwater signature. Bottom historisch Museum Maastricht, the Netherlands), Arie W.

dysoxia was widespread, explaining the abundance of Janssen (Nationaal Natuurhistorisch Museum, Leiden,

pachydontine corbulids and the near-absence of corbicu- the Netherlands), Salomon Kroonenberg (Tech-nische lid bivalves. Universiteit Delft, the Netherlands); Henri Laurent (for- The molluscan and ostracod faunas and isotope data merly OXI, Lima, Peru), John Lundberg (Academy of both indicate lake environment a long-lived depositional Natural Sciences of Philadelphia, USA), Kenny Monsch for the Pebas Formation. The term lake does fully cover (Bristol University, UK), Bob Morley (Cambridge, UK), the is than system, but more appropriate other terms. Fernando Munoz-Torres (Ecopetrol, Bucaramanga, Co- Lake Pebas neither in interior was deposited an seaway lombia), Luisa Rebata (UTU, Turku, Finland), Kalle in nor floodplain settings nor in an inland (hydrologically Ruokolainen (UTU, Turku, Finland), Juan Saldarriaga lake Based closed) setting. on the presence of marine (Tropenbos-Colombia, Bogota, Colombia), Jukka Salo

organisms (rays, mangroves) and the occurrence of tidal (UTU, Turku, Finland), Gustavo Sarmiento (UN, Bo- the Pebas deposits throughout Formation, a connection to gota, Colombia), Pepe Torres (UNAP, Peru), Hanna marine environments is Lake Pebas proposed. was lo- Tuomisto (UTU, Turku, Finland), Geerat Vermeij (Uni- cated in freshwater a tidal basin with only occasional versity of California, Davis, USA), Antonio Villa (for- marine incursions. is best described It as a paramarine merly INDERENA, Leticia, Colombia), Jaana Vormisto

which be megalake, happened to perimarine as well. (UTU, Turku, Finland), John de Vos (Nationaal Natuur- The ichnofauna includes common brackish assemblages, historisch Museum, Leiden, the Netherlands), Kitty Vi- also in intervals where and faunal data indicate Universiteit isotope jverberg (formerly van Amsterdam, the

freshwater The of documented salin- strictly settings. use Netherlands), Rob van Weezel (formerly Nationaal of marine ity preferences extant marginal taxa in recon- Natuurhistorisch Museum, Leiden, the Netherlands), Cor in structing palaeosalinities evolutionary long-lived lake Winkler Prins (Nationaal Natuurhistorisch Museum, Lei-

is not is shown for the the systems straightforward, as e.g. den, Netherlands) and Francisco Zavaleta (formerly Pebasian Corbulidaeand the Caspian Limnocardiidae. OXI, Lima, Peru). Maaike Wickardt performed the tedi- The maximum size of Lake Pebas is estimated to ous task of putting in order stratigraphic data in Appen- 2 . have been c. 1.1 million km Tectonic accommodation dix 1. Patrick Nuttall (formerly The Natural History Mu- and in its hinterland drove its precipitation mainly per- seum, London, UK) not only provided an absolutely sistence. The initial lake formed presumably during the brilliant starting point for this taxonomy-based research latest in Oligocene the Colombian east Andean foreland (i.e. his 1990 paper), but also greatly contributedto clari- basins the (including Magdalena Basin at that time), and fying systematic matters through discussions during the

eastward, into Llanos- of the work. In wish expanded gradually separating a early stages present particular, we to Sea/Lake in the north and Lake Pebas in the south. The thank Carina Hoom (formerly UVA, Amsterdam, the for 17 million system persisted c. years, although few Netherlands), who initiated the palaeontological research data on the early and the latest parts of its history are on the Pebas Formation, introduced the first author to the available. The Pebas-Llanos have system may played Pebas faunas and facilitated initial fieldwork in Colom-

some role in the evolution of marine taxa in Amazonian bia and Peru, and, last but not least, contributed enor-

freshwater It also have of ecosystems. may played a signifi- mously to our knowledge the geological history of

cant, hitherto unrecognised, role as a dispersal barrier of northwest Amazonia.

terrestrialbiota on the east and west sides of the lake. -64-

Die References Cosel, R. von 1978. Gattung Polymesoda Rafinesque,

1820 an der Nordkiiste Sudamerikas (Bivalvia: Corbiculi-

dae). Archiv fur Molluskenkunde 108, 201-213. Aartsen, J.J. van & Wesselingh, F.P, 2000. New Odostomia R. 1986. Moluscos de la de la species (Gastropoda, Heterobranchia, Pyramidellidae) from Cosel, von region Cienaga Grande de Santa Marta del Caribe de the Miocene Pebas Formation of Western Amazonia (Peru, (Costa Colombia). Anales Instituto del Mar, Punta de Betin Colombia). Basteria 64, 163-168. Investigaciones

H. 15/16, 79-370. Adams, I860. Description of a new genus of freshwater Costa, E.V. 1981. Revisao fosseis da localidade do bivalve Mollusca, belongingto the family Corbulidae, from gastropodes Tres Formacao Pliocene do Alto Amazo- the collection of Hugh Cuming, Esq. Proceedings of the Unidos, Pebas, Brasil. Anais II Latino-Americano Palae- Zoological Society ofLondon 28, 203. nas, Congresso ontologia, Porto Alegre, Abril 1981,635-649. Anderson, F.M. 1928. Notes on Lower Tertiary deposits of H. Novae Caledoniae Colombia and their molluscan and foraminiferal fauna. Crosse, 1872. Diagnosis molluscorum

incolarum. Journal de 69-75. Proceedings of the California Academy of Sciences 17, 1- Conchyliogie 20, G.A. 1992. 28. Darrigan, Variation temporal y espacial de la dis-

tribucion de las especies de Corbicula Megerle, 1811 (Bi- Anderson, L. 1996. Neogene paleontology in the northern Do- el estuario del Rio de La minican Republic, 16. The family Corbulidae (Mollusca, valvia, Corbiculidae), en Plata, 38, 59-63. Bivalvia). Bulletins ofAmerican Paleontology 110, 6-34. Republica Argentina. Neotropica

M. & P. 530 Dufrene, Legendre, 1997. Species assemblages and Baker, F.C. 1945. The molluscan family Planorbidae, pp.

indicator species: the need for a flexible asymmetrical Urbana (Illinois Press). ap- 345-366. Dumont, Baldock, J.W. 1982. Geology of Ecuador; explanatory bulletin proach. Ecological Monographs 67, HJ. 2000. Endemism in the Ponto-Caspian fauna, with of the National geological map of the Republic of Ecuador

on the Ad- scale 1:1,000,000.MRNE, Quito. special emphasis Onychopoda (Crustacea).

vances in Ecological Research 31, 181-196. Beesley, P.L., Ross, G.J.B. & Wells, A. (eds), 1998. Mollusca: H.A. Fauna Eisma, D., Mook, W.G. & Das, 1976. Shell characteris- the Southern Synthesis. of Australia 5, 1234 pp. tics, and trace-element contents of Melbourne (CSIRO Publishing). isotopic composition molluscs indicators of Palaeo- Berggren, W.A., Kent, D.V., Aubrey, M.-P. & Hardenbol, J. some euryhaline as salinity. 39-62. (eds), 1995. Geochronology, time scales and global strati- geography, Palaeoclimatology, Palaeoecology 19, R. 1879. Notes Mollusca collected C. Bar- graphic correlation. Society of Economic and Petroleum Etheridge, on by from the Mineralogists, Special Publication54. rington Brown, Esq., A.R.S.M., Tertiary deposits of Solimoes and Brazil. Quarterly Journal Bernard!, A.C. 1856. Description des coquilles nouvelles. Javary Rivers, the London 82-88. JournaldeConchiology 5, 82-84. of Geological Society of 35, Figueiredo L.C. de & Ricci, C.N. 1989. Boettger, O. 1878. Die Tertiarfauna von Pebas am oberen Ma- Alvarenga, Especie

nova de Mytilopsis Conrad, 1857, do Rio Tocantins, Tucu- ranon. Jahrbuch der Koniglichen-Kaiserlichen Geologische Brasil Memo- Reichsanstalt Wien28, 485-540. rui, Para, (Mollusca, Bivalvia, Dreissenidae). rias do Instituto Oswaldo Cruz 84, 27-33. Boss, K.J. 1978. On the evolution of gastropods in ancient Furch, K. & Junk, W.J. 1997. Physicochemical conditions in lakes. In: Fretter, V. & Peake, J. (eds). Pulmonates, 2a: the In: W.J. The central Amazon Systematics, evolution and ecology, 385-428. London floodplains. Junk, (ed.). floodplain: Ecology of a pulsing system. Ecological Stud- (Academic Press). ies 126, 69-108. Cadee, G.C. 1984. ‘Opportunistic feeding’, a serious pitfall in Gabb, W.M. 1869. of fossils from the trophic structure analysis of (palaeo)faunas. Lethaia 17, Descriptions clay depos-

Amazon. American Journal 289-292. its of the Upper ofConchology

4, 197-200. Cemohorsky, W.O. 1984. Systematics of the family Nassarii- Gasse, Fontes, J.C., P., Kaczmarska, dae (Mollusca: Gastropoda). Bulletin of the Auckland In- F., Plaziat, J.C., Carbonel, de P., I., Y. & stitute andMuseum 14, 1-357. I., Deckker, Soulie-Marsche, Callot, Dupeable, P.A. 1987. Biological remains, Clough, M. & Summers, K. 2000. Phylogenetic systematics geochemistry and stable for the reconstruction of environmental and biogeography of the poison frogs; evidence from mito- isotopes

in chondrial DNA Lin- and hydrological changes the Holocene lakes from North sequences. Biological Journal of the Sahara. Palaeogeography, Palaeoclimatology, Palaeoeco- nean Society 70, 515-540. logy 60, 1-46. Conrad, T.A. 1871. Descriptions of new fossil shells of the E. & A.W. 1998. De Amazon. American Journal Gittenberger, Janssen, (eds), Nederlandse upper ofConchology 6, 192-

zoetwatermollusken. Recente en fossiele weekdieren uit 198.

zoet en brak water. NederlandseFauna 2, 1-288. Conrad, T.A. 1874. Remarks on the Tertiary clay of the Upper

Glaubrecht, M. 1996. Evolutionsokologie und Systematik am Amazon with descriptions of new shells. Proceedings of

Beispiel von SUss- und Brackwasserschnecken (Mollusca: the Academy of natural Sciences Philadelphia 1874, 25-

32. Caenogastropoda; Cerithioidea); Ontogenese-Strategien,

Befunde und historische Cooper, M.A., Addison, F.T., Alvarez, R., Coral, M., Graham, palaontologische Zoogeographie, 499 Leiden (Backhuys Publishers). R.H., Hayward, A.B., Howe, S., Martinez, J., Naar, J., pp. A. 1992. Peftas, R., Pulhma, A.J. & Taborda, A. 1995. Basin devel- Gorthner, Bau, Funktion und Evolution komplexer in Langzeit-Seen mit einem Beitrag opment and tectonic history of the Llanos Basin, Eastern Gastropodenschalen

zur Palaobiologie von Gyraulus im Stein- Cordillera, and Middle Magdalena Valley, Colombia. ‘multiformis’

heimer Becken. Stuttgarter Beitrdge zur Naturkunde American Association ofPetroleum Geologists Bulletin 79, (B)I90, 1-173. 1421-1443. -65 -

Greve, L. de 1938. Eine Molluskenfauna aus dem Neogenvon Kelleher, B., van der Velde, G. & bij de Vaate, A. 1999. Nu

Iquitos am Oberen Amazonas in Peru. Abhandlungen der ook levende Mytilopsis leucophaeata (Dreissenidae) in de

Schweizerischen paldontologischen Gesellschaft 61, 1-133. Waal, Correspondentieblad van de Nederlandse Mala-

Guerrero, J. 1997. Stratigraphy, sedimentary environments, and cologische Vereniging307, 26-28.

the Miocene uplift of the Colombian Andes. In: Kay, R.F., Kronberg, B.I., Franco, J.R., Benchimol, R.E., Hazenberg, G.,

Madden, R.H., Cifelli, R.L. & Flynn, J.J. (eds). Vertebrate Doherty, W. & VanderVoet, A. 1989. Geochemical varia-

paleontology in the Neotropics, the Miocene fauna ofLa tions in Solimoes Formation sediments (Acre Basin, west-

Venta, Colombia, 15-43. Washington D.C. (Smithsonian ern Amazonia). ActaAmazonica 19, 319-333.

Press). Kuiper, W. 2000. De weekdierenvan de Nederlandse brakwa-

Haas, F. 1949. On fresh water mollusks from the Amazonian tergebieden (Mollusca). Faunistische Mededelingen 12,

region. Anales Instituto Biologico del Universidad de 41-120.

Mexico 20, 301-314. Lamarck, J.B.A.P. M, de, 1819. Histoire naturelle des animaux

Radiation of crenobiontic vertebres. Haase, M. & Bouchet, P. 1998. gas- sans Bivalvia 6, 1-233. Paris (Deterville & Ver-

tropods on an ancient continental island: the Hemistomia- diere).

clade in New Caledonia (Gastropoda: Hydrobiidae). Hy- Lange de Morretes, F. 1949. Ensaio de Catalogo dos Moluscos

drobiologia 367, 43-129. do Brasil. Arquivos do Museu Paranaense 7, 1-194.

Hammen, T. van der 1956. A palynological systematic nomen- Leistikow, A. & Janssen, R. 1997. A record of the bivalve

clature.Boletino Geologico 4, 63-101. Guianadesma sinuosum Morrison from the central Amazon

Hedberg, H.D. 1936. Stratigraphy of the Rio Querecual section basin (Bivalvia: Corbulidae). Basteria 61, 17-22.

ofnortheastern Venezuela. Bulletin ofthe Geological Soci- Lewy, Z. & Samtleben, C. 1979. Functional morphology and

ety ofAmerica 48, 1971 -2024. palaeontological significance of the conchiolin layers in

Hershler, R. & Thompson, F.G. 1992. A review ofthe aquatic corbulid pelecypods. Lethaia 12, 341-351.

subfamily Cochliopinae (Prosobranchia; Hydrobiidae). Lovejoy, N.R., Bermingham, E. & Martin, A.P. 1998. Marine

Malacological Review, Suppl. 5, 1-140. incursion into South America. Nature 396, 421, 422.

Holmden, C., Greaser, R.A. & Muehlenbachs, K. 1997. Paleo- Lundberg, J.G., Marshall, L.G., Guerrero, J., Horton, B., Mala-

salinities in ancient brackish water systems determined by barba, M.C.S.L. & Wesselingh, F. 1998. The stage for 87 86 Sr/ Sr ratios in carbonatefossils: A case study from the Neotropical fish diversification: A history of tropical South

Western Canada sedimentary basin. Geochimica Cosmo- American rivers, In: Malabarba, L.R. et al. (eds). Phytog-

chimica Acta 2105-2118. 13-48. Porto 61, eny and classification of neotropical fishes,

Hoorn, C. 1993. Marine incursions and the influence of An- Alegre (Edipurcs).

dean tectonics on the Miocene depositional history of Maia, R.G., Godoy, H.K., Yamaguti, H.S., de Moura,P.A., Da

northwestern Amazonia: Results of a palynostratigraphic Costa, F.S., de Holanda, A.M. & Costa, J. 1977. Projeto de

Relatorio study. Palaeogeography, Palaeoclimatology, Palaeoecol- Carvao no Alto Solimoes. Final, 137 pp. Rio de

ogy 105, 267-309. Janeiro(CPRM-DNPM).

An L.G. & J.G. 1996. Technical Hoorn, C. 1994a. environmental reconstruction of the pa- Marshall, Lundberg, comments:

leo-Amazon River system (Middle to Late Miocene, NW Miocene deposits in the Amazonian Foreland Basin. Sci-

Amazonia). Palaeogeography, Palaeoclimatology, Pa- ence 273, 123, 124.

New laeoecology 112, 187-238. Marshall, W.B. 1927. Uruguayan mollusks of the genus

Hoom, C. 1994b. Miocene palynostratigraphy and paleo- Corbicula. Proceedings of the United States National Mu-

environments ofnorthwesternAmazonia: evidence for ma- seum 66, 1-12.

rine incursions and the Andean 156 R.E. 1999. 508 influence of tectonics, Martin, Taphonomy: a process approach, pp.

PhD pp. Amsterdam (University of Amsterdam) (unpubl. Cambridge (Cambridge University Press).

thesis). McCune, B. & Mefford, M.J. 1999. PC-ORD. Multivariate

Hoorn, C. 1996. Technical comments: Miocene deposits in the analysis of ecological data, version 4. Gleneden Beach Amazonian Foreland Basin. Science 273, 122, 123. (MjM Software Design).

Hoom, C., Guerrero, J., Sarmiento, G.A. & Lorente, M.A. Michel, E. 1994. Why snails radiate: a review of gastropod

1995. Andean tectonics as a cause for changing drainage evolution in long-lived lakes, both recent and fossil. Archiv

patterns in Miocene northern South America. Geology 23, fur Hydrobiologie (Beiheft Ergebnisse der Limnologie) 44,

237-240. 285-317.

1984. de de IFG, Mapa planimetrico imagenes satelite, Michel, E. 2000. Phytogeny of a gastropod species flock: ex-

1:250.000. Neu Isenburg (IFG). ploring speciation in Lake Tanganyika in a molecular

Irion, G., Muller, J., Nunes de Mello, J. & Junk, W.J. 1995. framework. Advances in Ecological Research 31, 275-302.

Quaternary geology of the Amazonian lowland. Geo- Monsch, K.A. 1998. Miocene fish faunas from the northwest-

Marine Letters 15, 172-178. ern Amazonia basin (Colombia, Peru, Brazil) with evi-

Irmler, U. 1975. Ecological studies of the aquatic soil inverte- dence of marine incursions. Palaeogeography, Palaeocli-

brates in three inundation forests of Central Amazonia. matology, Palaeoecology 143, 31-50.

Amazoniana 3, 337-409. Morley, R.J. 2000. Origin and evolution of tropical rain for-

WJ. 1997. The Central Amazon Junk, (ed.) floodplain. Ecol- ests, 362 pp. Chichester (John Wiley & Sons).

O.F. 1774. Vermium et ogy of a pulsing system. Ecological Studies 126, 1-525. Muller, terretrium fluviatilium, seu

Kaandorp, R.J.G., Vonhof, H.B., Wesselingh, F.P. & Ganssen, animalium Infusoriorum, Helminthicorum,et Testaceorum,

G.M. 2000. Stable isotope records of climate variation in nonmarinorum succincta historia2, 1-214.

growth increments of Amazonian fresh water bivalves. Muller, P., Geary, D.H. & Magyar, I. 1999. The endemic mol-

European Union of Geological Sciences, 25th Annual luscs of the late Miocene Lake Pannon: their origin, evolu-

1 and 47-60. Meeting, Nice, Abstracts, pp. tion, family-level taxonomy.Lethaia 32, -66-

R. & van D. M.G. 1924. Munoz-Torres, F., Whatley, Harten, 1998. The Roxo, de Oliveira Breve noticea sobre os fosseis

endemic non-marine Miocene ostracod fauna of the Upper terciario do Alto Amazonas. Boletim do Servicio

Amazon Basin. Revista espahola de Micropaleontologia geologico-mineralogico Brasil 11, 41-52. 89-105. 30, Russell, H.D. 1941. The recent mollusks of the family Neriti-

Nuttall, C.P. 1990a. A review ofthe Tertiary non-marine mol- dae of the Western Atlantic. Bulletin of the Museum of

luscan faunas of the Pebasian and other inland basins of Comparative Zoology 88, 347-404.

north-western South America. Bulletin the British Mu- Savazzi, of E. & Yao Peiyi 1992. Some morphological adapta-

seum Natural 45, of History (Geology) 165-371. tions in freshwater bivalves. Lethaia25, 195-209.

Nuttall, C.P. 1990b. Review of the Caenozoic heterodont bi- Simone, L.R.L. 1999. Anatomy and systematics of Anticorbula valve Dreissenacea. superfamily Palaeontology 33, 707- fluviatilis (H. Adams, 1860) (Bivalvia: Lyonsiidae) from

737. the Amazon Basin, Brazil and Peru. The Nautilus 113, 48-

Olsson, A.A. 1961. 574 55. Panamic-Pacific Pelecypoda, pp

Ithaca (Paleontological Research Institution). Sioli, H. (ed.) 1984. The Amazon. Limnology and landscape

Patterson, R.T., W.B., Kroker, E. & a river and its 763 McKillop, S., Nielsen, ecology of mighty tropical basin, pp.

Reinhardt, E.G. 1997. Evidence for rapid avian-mediated Dordrecht (Junk). foraminiferal colonization of Lake Winnepegosis, Mani- Smith, J. 1999. Multivariate techniques in palynofacies analy- the Holocene toba, during Hypsithermal. Journal of Pa- sis. In: Harper, D.A.T. (ed.). Numerical palaeobiology, leolimnology 18, 131-143. 361-393. London (John Wiley and Sons Ltd.). W.G.R. Paxton, C.G.M., Crampton, & Burgess, P. 1996. Steinmann, M., Hungerbiihler, D., Seward, D. & Winkler, W.

in the Technical comments: Miocene deposits Amazonian 1999. Neogene tectonic evolution and exhumation of the Foreland Basin. Science 123. 273, southern Andes: a combined stratigraphy and fission-track S. & Petri, Fulfaro, V.J. 1983. In: Queiroz, T.A. (ed.). Geolo- approach. Tectonophysics 307, 255-276.

gia do Brasil (Fanerozoico), 408-412. Sao Paulo (Univer- Tschopp, H.J. 1953. Oil explorations in the Oriente of Ecua- sidade de Sao Paulo). dor, 1938-1950. Bulletin of the American Association of

Pilsbry, H.A. & Olsson, A.A. 1935. Tertiary fresh-water mol- Petroleum Geologists 37, 2303-2347. lusks of the Magdalena embayment, Colombia. Proceed- Vermeij, G.J. & Wesselingh, F.P. 2002. Neogastropod mol- the Natural Sciences luscs ings of Academy of ofPhiladelphia from the Miocene of Western Amazonia, with com-

7-39. 87, ments on marine to freshwater transitions in molluscs.

Rasanen, M., Gingras, M.K., Wesselingh, F. & Pemberton, G. Journal ofPaleontology 76, 265-270.

2000. Sequence stratigraphy of the marginal marine Mio- Villamil, T. 1999. Campanian-Miocene tectonostratigraphy, Pebas/Solimoes cene Formation in Amazonian Foreland depocenter evolution and basin development of Colombia

II Basin, Peru. Latin-American Sedimentological Con- and western Venezuela. Palaeogeography, Palaeo- VIII gress/ Argentinian MeetingofSedimentology, Mar del climatology, Palaeoecology 153, 239-275. March 2000, Abstracts, Plata, I p. Vonhof, H.B. 1998. The strontium isotope stratigraphic record

Rasanen, M.E., Linna, A.M., Santos, J.C.R. & F.R. selected 138 Negri, of geologic events, pp. Amsterdam (Vrije

1995. Miocene deposits in the Amazonian Foreland Basin. Universiteit) (unpubl. PhD thesis). Science 386-389. 269, Vonhof, H.B., Wesselingh, F.P, & Ganssen, G.M. 1998. Re- M., Rebata L., Rasanen, Linna, A., Irion, G., Hemani, Vargas construction of the Miocene western Amazonian aquatic

Huaman, R. & Wesselingh, F. 1998, Geologia y geoformas system using molluscan isotopic signatures. Palae-

In: R. & de la zona de Iquitos. Kalliola, Flores Paitan, S. ogeography. Palaeoclimatology, Palaeoecology 141, 85-

(eds). Geoecologia y desarollo Amazonico: estudio inte- 93.

Peru. grado en la zona de Iquito, Annales Turkuensis Way, C.M., Hombach, D.J., Miller-Way, C.A., Payne, B.S. &

59-137. (A1I)114, Miller, A.C. 1990. Dynamics of filter feeding in Corbicula

Rasanen, M.E., Gingras, M.K., Wesselingh, F.P., Vonhof, fluminea (Bivalvia: Corbiculidae). Canadian Journal of

H.B., Romero Pittman, L., Rasanen, P. & Siiro, P. in Zoology68, 115-120.

preparation. Seasonal and tidal brackish and freshwater Webb, S.D. 1995. Biological implications of the Middle Mio-

sedimentationof of 362. transgressive-regressive parasequences cene Amazon seaway. Science 269, 361, the lower Pebas formation (Middle Miocene) Amazonian Wesselingh, F.P. 1993. On the systematics of Mioceneaquatic

foreland basin, Peru. mollusks from Los Chorros (dept. Amazonas, Colombia) Rebata, L. Hemani 1997. Description ofNeogene-Quaternary and Nuevo Horizonte (dept. Loreto, Peru), with comments

tide and Nauta and 215 wave-influenced estuary-sediments along on palaeoenvironments palaeogeography, pp.

km 0-15, Loreto-Peru, NW 93 Amsterdam Iquitos road, Amazonia, pp. (Vrije Universiteit) (unpubl. MSc thesis).

Turku (University ofTurku) (unpubl. MSc thesis). Wesselingh, F.P. 1998. Over het onderscheid tussen Dreissena

C.A. 1849. de nouvelles Recluz, Description quelques especes van Beneden, 1835 en Mytilopsis Conrad, 1858. Corre-

de Revue et de coquilles. Magasin Zoologie Paris 2, 64-71. spondentieblad van de Nederlandse Malacologische Ver- Rodriguez, G. 1963. The intertidal estuarine communities of eniging 302, 52-54.

Lake Maracaibo, Venezuela. Bulletin ofMarine Science of Wesselingh, F.P. & Gittenberger, E. 1999. The Giant Amazo- the and Caribbean Gulf 13, 194-218. nian Snail (Pulmonata: Acavidae) beats them all. Veliger

L. Romero-Pittman, 1997. Relaciones bioestratigraficas y pa- 42, 67-71. - leoambientales de la fauna de la formacionPebas Area de Wesselingh, F.P. 2000. On relict hydrobiid species in Brazilian

Iquitos (Dpto. De Loreto, Peru). IX Congreso Peruano de Amazonia (Gastropoda, Prosobranchia, Hydrobiidae). Geologia. Resiimenes extendidos. Sociedad Geologico del Basteria 64, 129-136. Peru, Volumen Especial 1, 609-614. Whatley, R., Munoz-Torres, F. & vanHarten, D. 1996. Des -67-

ostracodes d’un lac sale Neogene a 1’ouest dubassin Ama-

zonien. the Third Proceedings of European Symposium on

25 Ostracoda (Bierville), pp.

Whatley, R., Munoz-Torres, F. & van Harten, D. 2000.

Skopaeocythere: a minute new limnocytherid (Crustacea,

Ostracoda) from the Neogene of the Amazon Basin.

Ameghiniana 37, 163-167.

Zamudio, K.R. & Greene, H.W. 1997. Phylogeography of the

bushmaster (Lachesis muta: Viperidae): implications for

neotropical biogeography, systematics, and conservation.

Biological Journalofthe Linnean Society 62, 421-442. -68 -

Appendix 1. Locality data

Figure 15. Fieldwork area. -69-

Figure 16. Fieldwork area.

The first author collected all from situated samples outcrops mainly along riverbanks in the Peruvian department of Loreto. River in the courses accompanying maps are redrawn from the Mapa Planimetrico (IFG, 1984). Co-ordinates estimates from the Planimetrico found are Mapa and to be slightly inaccurate when compared with GPS co-ordinates few in later (for a localities) years. LB- left bank, RB- right bank.

L Porvenir: LB (W-bank) Amazon River, exact location unknown (73°23’W, 4°14’S), sample 688 (2-9-96). 2. Porvenir II: LB Amazon ( W-bank) River, 50 m N of Porvenir IV, sample 702 (5-9-96). 3. Porvenir IV: LB (W-bank) Amazon River, 250 m N ofPorvenir II, samples 703-705,707(6-9-96). 4. Porvenir VI: LB Amazon (W-bank) River, 50 m N ofPorvenir IV, sample 715 (5-9-96). 5. Porvenir VIII: LB (W-bank) Amazon River, 200 m N of Porvenir VI, samples 716, 206 (5-9-96). 6. Porvenir IX: LB (W-bank) Amazon River, 100 m N ofPorvenir VIII, samples 207, 718-722 (5-9-96). 7 Porvenir X: LB (W-bank) Amazon River, 50 m N of Porvenir IX, samples 724-726 (6-9-96).

Porvenir XIV: LB 8. (W-bank) Amazon River, m N Porvenir 120 of IX, northern tip of village (73°23’W, 4°14’S), sample 687 (2- 9-96).

9. Barradero de I: RB Omagua (E-bank) Itaya River, below village (73°23’W, 4°10’S), sample 800 (13-9-96).

10. Barradero de II: stream 50 m above confluence with Omagua bank, Itaya River, in village (73°23’W, 4°10’S), sample 801 (13-9- 96).

11. I: RB River Itaya (E-bank) Itaya (73°23’W, 4°08’S), samples 802-803 (13-9-96). 12. Itaya II: RB (E-bank) Itaya River (73°23’W, 4°07’S), sample 804 (13-9-96). 13. III; RB Itaya (E-bank) Itaya River (73°23’W, 4°05’S), samples 805-806 (13-9-96). 14. Nuevo Horizonte II: road cutting (E- side) Iquitos-Nauta road, km 40, 200 m S of village (73°25’W, 4°05’S), samples 365-366 836(16-9-96).

15. Nuevo Horizonte III; road cutting (W- side) Iquitos-Nauta road, km 38, c. 2 km N of village (73°26’W, 4°02’S), samples 202- 203, 837(16-9-96).

16. Nuevo Horizonte IV: road road, ca. cutting (W- side) Iquitos-Nauta 400 m N ofNuevo Horizonte III, sample 368 (16-9-96). 17. Paraiso: LB (W-bank) Itaya River, below village (73°23’W, 4°03’S), samples 807-808 (13-9-96). 18. San Antonio I: LB (W-bank) Itaya River, at S side of village (73°23’W, 4°01’S), samples 809-810 (13-9-96).

19. San Antonio II: LB at N side ’ (W-bank) Itaya River, village, c. 200 m N of San Antonio I (73°23 W, 4°01’S), samples 811-815 (13-9-96).

20. San Antonio III: LB below 250 of (W-bank) Itaya River, school, c. m N San Antonio/ Itaya I (73°23’W, 4°01 ’S), samples 8lb- 819 (13-9-96). -70-

Figure 17. Iquitos area. -71 -

Figure 18. Lower Napo River and Amazon River, north of Iquitos.

21. Palo Seco: LB (W-bank) Itaya River (73°22’W, 4°00’S), sample 820 (13-9-96).

22. Soledad: LB (W-bank) Itaya River at S-side village (73°2rW, 3°59’S), sample 821(13-9-96). 23. Santo Tomas: rill at W-side ofroad Iquitos-Nauta, km 1 (73°18’W, 3°48’S), sample 360 (13-8-96). 24. San artificial INIA-research Roque; station, c. 1 km SE ofairport Iquitos (73°16’W, 3°47’S), sample 913 (3-10-96). 25. Iquitos-Itaya: LB (W-bank) Itaya, 500 m S ofport of Belem, at factory (73°I5’W, 3°46’S), samples 361-364 (13-8-96).

Puerto in 26. Iquitos Ganso-Azul: outcrop harbour, below market, at grifos (73°14’W, 3°45’S). 27. Puerto Almendras: RB (E-bank) Nanay River, 25 m N (downstream) of port (73°22’W, 3°49’S), sample 835 (15-9-96).

28. Santo Tomas: LB c. 1km W/NW of (W-bank) Nanay River, port of Bellavista (73°15’W, 3°41’S), samples 302-304 (3-8-96). 29. Santa Maria: LB (W-bank) Nanay River, 300 m S of naval base (73°15’W, 3°40’S), samples 200, 350, 353 (12-8-96). 30. Barrio Florido: W-bank of confluence outcrop on Nanay and Amazon rivers, below refinery, 200 m N of village (73°12’W, 3°37’S), samples 305-306, 308-310 (3-8-96).

31. Boca Momon: LB (N-bank) Momon River, 300 m upstream from confluence with Nanay River (73°15’W, 3°41’S), samples 356-358 (12-8-96).

32. Momon V: RB Momon km (S-bank) River, c. 1.5 W of confluence with Nanay River, below Fundo de Dueno Donaire

(73°16’W, 3°41’S), sample 355 (12-8-96).

33. Momon IV: LB Momon (N-bank) River, c. 4 km NW of confluencewith Nanay River, below house (73°16’W, 3°40’S), sample 354 (12-8-96).

34. Porvenir-Momon; LB (E-bank) Momon River (73°20’W, 3°36’S), samples 832-834 (14-9-96).

35. Santo Tomas-Momon: LB (E-bank) Momon River (73°21’W, 3°33’S), samples 370 (date unknown), 827-828 (14-9-96).

36. Momon III: LB (E-bank) Momon River (73°20’W, 3°36’S), samples 829-831 (14-9-96).

37. Momon II: RB (W-bank) Momon River (73°23’W, 3°32’S), sample 826 (14-9-96). 38. Momon 1; RB and LB (W and E-bank) Momon River (73°22’W, 3°31’S), samples 823-825 (14-9-96). 39. RB Amazon Tamshiyacu: (E-bank) River, 500 m E ofport (73°09’W, 4°01’S), samples 684-686 (2-9-96), samples 754-758 (8- 9-96)

40. LB of Quebrada Tamshiyacu: (N-bank) Quebrada Tamshiyacu, c. 50 m above confluence with Amazon River (73°08’W, 4°0rS), sample 851 (22-9-96).

41. Nuevo Tarapaca; LB (N-bank) ofQuebrada Tamshiyacu, under tourist lodge (73°07’W, 4°02’S), sample 850 (22-9-96). 42. Santa RB of Elena-Tamshiyacu: (S-bank) Quebrada Tamshiyacu (73°05’W, 4°04’S), samples 848-849 (2-9-96). 43. Boca I: RB of 2 km E of confluence with Napo (N-bank) Napo River, c. Socosani River, c. 100 m W of church on hill (72°53’W, 3°17’S), sample 855 (23-9-96). -72-

Figure 19. NapoRiver.

44. BocaNapoII: LB (N-bank) Napo River, c. 5 km W of Socosani (73°00’W, 3°18’S), samples 906, 210 (1-10-96).

45. Boca Napo IV: LB (N-bank) Napo River, c. 7 km W of Socosani, sample 905 (1-10-96).

46. Socosani II: LB (N-bank) Napo River, 500 m E of confluence with Socosani River (72°55’W, 3°16’S), sample 911 (1-10-96).

47. Mazan 1: LB (E-bank) Napo River, 900 m S of port (73°06’W, 3°30’S), sample 856 (23-9-96), sample 902 (1-10-96).

48. Mazan IV: RB (E-bank) Napo River, 300 m N ofport (73°06’W, 3°30’S), sample 903 (1-10-96).

49. Santa Marta: LB N-end 900-901 (E-bank) NapoRiver, of exposure (73°13’W, 3°20’S), samples (30-9-96).

50. Buen Pasa I: LB (E-bank) Napo River, c. 1 km E of village, near confluence with brook (73°11’W, 3°15’S), sample 898 (30-9-

96).

51. Buen Pasa II: LB (E-bank) Napo River, 800 m E of village (73°11’W, 3°15’S), sample 899 (30-9-96).

52. Oran: outcrop LB (N-bank) Amazon River, 1 km W ofvillage, 400 m NW ofcape (72°31’W, 3°28’S), sample 640 (1-9-96).

53. Yanamono III: outcrop LB (W-bank) Amazon River, 4 km N of S-tip Isla Yanamono (72°52’W, 3°27’S), samples 204-205 (1-9-

96).

54. Yanamono I: brook, 20 m above confluence with Amazon River (W-bank), c. 3 km N of S-tip Isla 13. Yanamono (72°54’W,

3°28’S), sample 311 (6-8-96).

55. Yanamono II: LB (W-bank) Amazon River, c. 1 km S of Yanamono I, c. 2 km N of S-tip Isla Yanamono (72°55’W, 3°28’S),

samples 312-313 (6-8-96).

56. Santa Teresa III: LB 500 (W-bank) Amazon River, m N of village, oncape (72°59’W, 3°29’S), sample 314 (7-8-96).

57. Santa Teresa II a: LB (W-bank) Amazon River, 200 m N of village, at brook (73°00’W, 3°29’S), samples 318, 320-322 (7-8- 96).

58. Santa Teresa II b: LB (W-bank) Amazon River, 230 m N ofvillage (73°00’W, 3°29’S), samples 323-324 (7-8-96).

59. Santa Teresa I: outcrop LB (W bank) Amazon River below village (73°00’W, 3°29’S), samples 649, 653, 659, 668-669 (1-9-

96), samples 325-328 (7-8-96). -73 -

Figure 20. Pebas-Chimbote.

60 Santa Teresa IV: LB (W-bank) AmazonRiver, 1 km S of village (73°01’W, 3°29’S), sample 329 (7-8-96).

o 61. Santa Teresa V; LB (W-bank) Amazon River, 3 km S of village (73 01’W, 3°29’S), samples 382, 383 (17-8-96). Indiana 62. I: LB (W-bank) Amazon River, 200 m N of port (73°02’W, 3°29’S), samples 330-332 (7-8-96).

63. Indiana II: LB Amazon River, 3.1 km S of of Barradero (W-bank) village, 150 m N de Mazan (73°05’W, 3°31’S), samples 333,

335-336, 353 (8-8-96), samples 371, 383 (15-8-96).

64. Indiana III: LB (W-bank) Amazon River, 1.2 km S ofport (73°03’W, 3°30’S), sample 338 (8-8-96).

65. Indiana IV: LB (W-bank) Amazon River, 1.4 km S of port, 200 m S of Indiana III, samples 340 (8-8-96), 370 (16-8-96), sam-

ples 681-682(1-9-96).

66. Indiana V; LB (W-bank) Amazon River, 1.65 km S of port, 250 m S of IndianaIV, at stair to tourist lodge, samples 342-343,

345. Level 345 was taken 50 m S of outcrop (8-8-96).

67. Indiana VI: LB (W-bank) Amazon River, 1.9 km S of port, 250 m S of Indiana V, samples 346-350 (8-8-96).

68. Indiana VII: LB Amazon 2.3 km S of 400 m S of Indiana (W-bank) River, port, VI, at brook, samples 351, 201 (100 m N of

large channel incision) (8-8-96). Sample 379 was taken 150 m N of Indiana VII (16-8-96).

69. IndianaVIII; LB (W-bank) Amazon River, 2.95 km S ofport, 650 m S ofIndiana VII, 300 rn N ofBarradero de Mazan, samples

373-377(16-8-96).

70. Indiana X; LB (W-bank) Amazon River, 2.15 km S ofvillage, sample 372 (16-8-96).

71. Indiana, unknown section: LB (W-bank) Amazon River S of village, samples 213-214 (unknown date).

72 Napo I: RB (S-bank) Napo River (74°17’W, 2°07’S), samples 208-209 (25-9-96).

73. Tarapoto I: LB (E-bank) Rio Tarapoto, under the village (74°08’W, 2°07’S), sample 859 (25-9-96).

74. Copal Urco I: LB (E-bank) Napo River, 500 m S ofconfluence with Rio Urco (23.47’W, 2°20’S), samples 869-871 (26-9-96). Clotilde 75. Santa I: RB (S-bank) Napo River, c. 1 km E ofvillage (73°38’W, 2°30’S), sample 882 (29-9-96).

76. Santa Clotilde II: RB (S-bank) Napo River, c. 1.5 km E ofvillage (73°37’W, 2°30’S), sample 881 (29-9-96). Clotilde 77. Santa III: RB (S-bank) Napo River, c. 2 km E of village (73°37’W, 2°30’S), sample 880 (29-9-96).

RB W-end of 78. Fortaleza: (S-bank) Napo River, exposure (73°36’W, 2°31’S), sample 885 (29-6-96).

79. San Lorenzo: RB at 100 m W of 1 E of (S-bank) Napo River, cape village at E-end of exposure, c. km Fortaleza (73°35’W,

2°32’S), sample 886 (29-9-96).

80. San Luis Tacsha: RB (W-bank) Napo River, at confluence with Tacsha Curaray River (73°33’W, 2°48’S), sample 888 (29-9- 96).

81. Santa Maria Tacsha: RB (W-bank) Napo River, c. 1 km S (downstream) from confluencewith Tacsha Curaray River (73°33’W, 2°49’S), sample 889 (29-9-96). -74-

82. Santa Teresa Tacsha: RB (W-bank) Napo River, c. 2 km below confluence with Tacsha Curaray River (73°33’W, 2°50’S), sample 890 (29-9-96).

83. Caseria Bellavista 1: RB (S-bank) Napo River at caseria Bellavista (73°30’W, 2°54’S), sample 891 (30-9-96). 84. Caseria Bellavista II: RB (S-bank) Napo River, c. 500 m E of Caseria Bellavista I (73°30’W, 2°54’S), sample 892 (30-9-96).

85. San RB Felipe: (S-bank) Napo River, c. 800 m E of caseria Bellavista (73°29’W, 2°55’S), sample 893 (30-9-96).

86. Nueva Paleta I: RB River, of the (S-bank) Napo westernmost outcrop Negro Urco area (73°26’W, 2°59’S), sample 894 (30-9- 96).

87. Nueva Paleta II: RB (S-bank) Napo River, 1 km E of Nueva Paleta I, 2 km W of Negro Urco (73°25’W, 3°00’S), sample 895 (30-9-96).

88. Negro Urco: RB (S-bank) Napo River, 50 m E ofport (73°22’W, 3°02’S), sample 896 (30-9-96).

89. Bellavista-Napo: RB (S-bank) Napo River, few hundred meters W of school (73°20’W, 3°01’S), sample 857 (24-9-96). 90. Chimbote 1: of RB (S-bank Amazon River), c. 500 m west of village (70°45’W, 3°55’S), sample 470 (26-8-96).

91. Chimbote 111: RB (S-bank ofAmazon River), c. 1.5 km west of village (70°46’W, 3°55’S), samples 473-474 (26-8-96). 92. Timareo II: RB of (S-bank) Amazon River, c. 500 ra W of W tip of Isla Timareo (71°01’W, 4°00’S), samples 477-478 (27-8- 96).

93. Timareo 1: RB of (S-bank) Amazon River, c. 1 km W of W tip of Isla Timareo(71°02’W, 4°00’S), samples 475-476 (27-8-96).

94. RB Amazon o San Pablo de Loreto 1: (S-bank) of River, 200 m E ofport at water level (71°06’W, 4 01’S), sample 479 (27-8- 96).

95. San Pablo de Loreto II: RB ofAmazon c. 4 km W of (S-bank) River, San Pablo, at sport playground (71°09’W, 4°00’S), sam- ple 480 (27-8-96).

96. Mayoruna I: RB (S-bank) of Amazon River, 400 m S of confluencewith Nuevo Octubre River (71°12’W, 3°58’S), sample 481 (27-8-96).

RB 200 S of 97. Mayoruna II: (S-bank) Amazon River m confluence with Nuevo Octubre River (71°12’W, 3°58’S), sample 482 (27-8-96).

98. Santo Tomas-Amazon: RB ofAmazon W ofIsla (S-bank) River, at tip San Isidro (71°22’W, 3°52’S), samples 483-484 (27-8- 96).

99. Santa Elena I: RB (S-bank) Amazon River, 1 km W of Santo Tomas Amazonas (71°23’W, 3°52’S), samples 485-489 (27-8- 96).

100. Santa Elena II: RB (S-bank) Amazon River, 500 m W of Santa Elena I (71°23’W, 3°52’S), samples 496-499 (28-8-96).

101. Santa Elena III: 150 m long outcrop. Outcrop A located 300 m W of woodmill, and 600 m W of outcrop 102. Santa Elena II

(71°24’W, 3°53’S), samples 500-509 (28-8-96).

103. RB Beiruth: (S-bank) Amazon River (71°28’W, 3°53’S), samples 490-492 from outcrop at river bank ('isleta') (27-8-96). 104. San Antonio de Mateo I: RB (W-bank) Amazon River, c. 1 km N ofS-tip Isla (71°34’W 3°49’S), sample 493 (27-8-96). 105. San de Miguel Cochiquinas; RB (S-bank) Amazon River, c. 1.5 km E of confluence with Cochiquinas River (71°36’W,

3°47’S), samples 494, 495 (28-8-96).

106. Condor: RB Amazon lower 300 outcrop (S-bank) River, part m W of village, upper part 150 m W of village (71°41’W, 3°44’S), samples 510-515 (29-8-96).

107. San Francisco; RB Amazon lower SE outcrop (S-bank) River, part at tip of Isla Pichana, upper part 50 m to the E (71°43’W, 3°43’S), samples 530-531 (29-8-96).

108. Santa Rosa de Pichana: RB outcrop (W bank) Amazon River, c. 200 m S of confluence with Pichana River (71°46’W, 3°40’S), samples 211-212, 532-537, 539-542 (30-8-96).

109. Pebas I LB 150 of (Pijoyal IV): (N-bank) Ampiyacu River, m W naval base Pijoyal (71°50’W, 3°20’S), samples 435-437 (25- 8-96).

110. Pebas 11 (Pijoyal I): LB (N-bank) Amazon River, 150 m E of naval base Pijoyal, sample 433 (24-8-96). 111. Pebas III (Quebrada Pijoyal): LB (N-bank) Amazon River, 300 E of naval base Pijoyal, at brook (waterfall), samples 560-561 (31-8-96), sample 429 (20-8-96).

112. Pebas IV (Pijoyal II): LB (N-bank) Amazon River, 350 m E of naval base Pijoyal, samples 427-428 (24-8-96). Pebas 113. V (Pijoyal III): LB (N-bank) Amazon River, 450 m E of naval base Pijoyal, samples 430-431 (24-8-96). 114. Pebas VI Maria (Ave III): LB (N-bank) Amazon River, 900 m E ofnaval base Pijoyal, samples 576-578 (31-8-96). 115. Pebas Maria VII (Ave IV): LB (N-bank) Amazon River, 1050 m E ofnaval base Pijoyal, sample 404 (22-8-96). 116. Pebas VIII Maria LB (Ave I): (N-bank) Amazon River, 1150 m E ofnaval base Pijoyal, samples 401-402 (21-8-96). 117. Pebas IX Maria (Ave II): LB (N-bank) Amazon River, 1350 m E of naval base Pijoyal, sample 400 (21-8-96).

118. Pebas XI (Santa Julia II): LB (N-bank) Amazon River, 1650 m E ofnaval base Pijoyal, sample 588 (31-8-96), samples 405- 408 (22-8-96).

119. Pebas XIII (Santa Julia IV): LB (N-bank) Amazon River, 2250 m E of naval base Pijoyal, samples 415, 417-420 (23-8-96). 120. Santa Julia: LB Amazon River, 2650 of (N-bank) m E naval base Pijoyal, published in Hoorn, 1993; samples 625-626, 628- 629, 631,635-636 (31-8-96).

121. Pebas XV (Santa Julia V): LB (N-bank) Amazon River, 3000 m E ofnaval base Pijoyal, sample 424 (23-8-96).

122. Pebas XVI Julia (Santa VI): LB (N-bank) Amazon River, 3200 m E ofnaval base Pijoyal, samples 421-423 (23-8-96). 123. Pebas XVII (Tarma); LB (N-bank) Amazon River, 6500 m E of naval base Pijoyal (71°47’W, 3°22’S), samples 425-426 (23- 8-96). -75 -

Appendix 2. Sampling categories and faunal indices

Classification

The following taxonomic groups have been used (with references to illustrated specimens):

Neritina ortoni group — Nuttall (1990a) recognised one species of Neritina in the Pebas Formation {Neritina ortoni

The abundant material shows that least four Conrad, 1871). very new at species are represented. Two of these, viz.

Neritina 1990a, + Neritina and in the ortoni (Nuttall, figs 9-22) sp.l (Nuttall, 1990a, fig. 16a, b, Figure 7.6 present pa-

ortoni per) make up the Neritina group.

Neritina roxoi group - This group is composed of Neritina roxoi de Greve, 1938 (Nuttall, 1990a, figs 23, 24) +

Neritina 1879 puncta Etheridge, (Nuttall, 1990a, figs 7, 8)

Ampullariidae - Ampullarius s.l. sp. indet.

- This contains lintea Large Dyris group category Dyris ortoni Gabb, 1869 (Nuttall, 1990a, figs 104-108), Dyris

(Conrad, 1871) (Nuttall, 1990a, figs 49-53, not figs 54-58) and Dyris sp. Wesselmgh (1993, fig. 37). -All other Small Dyris group Dyris species (e.g. Dyris gracilis Conrad, 1871, D. tricarinatus (Boettger, 1878), Dyris hauxwelli various undescribed in Nuttall, 1990, Dyris spp. figured Wesselingh (1993) and in Figure 3 of the present paper).

- Includes all referred in Nuttall Tryonia group species to (1990a) as Liris. Tentatively attributed are the species listed by Nuttall as Dyris tuberculatus (de Greve, 1938) and D. semituberculatus Nuttall, 1990 (whose generic assignment needs clarification) as well as a species of a presumably new genus, a shell of which Nuttall (1990a, fig. 43) classified

of as ajuvenile Dyris gracilis Conrad, 1871.

Sioliella - All referred to Ebora Eubora here to Sioliella group species formerly as or are assigned following Wessel-

ingh (2000). Furthermore, species belonging to the genera Tropidobora, Toxosoma and Littoridina listed by Nuttall

(1990a) are included in this group. Also Liosoma curta Conrad, 1874 was recognised valid (contra Nuttall, 1990a) and

is included in this group, as well as the species listed by Nuttall as Vitrinellahauxwelli Nuttall, 1990 and V. degrevei

Nuttall, 1990 (which both should be transferredto Nanivitrea). In addition, species tentatively attributed to Lithococcus

are included in this group.

Other Cochliopinae - Include undescribed species assigned by Wesselingh (1993) to Heleobia, Lyrodes (now tenta-

tively transferredto Onobops, see Figure 3), Pyrgophorus and Cochliopina. Apart from Pyrgophorus, generic assign-

ment is very uncertainin this group.

- All of and included in the Pachychilidae species Sheppardiconcha Doryssa are Pachychilidae. Sheppardiconcha is

assigned to the Pachychilidae because of the lack of embryonal shells that are well known from Aylacostoma and

and the of subsutural micro-omamentalso Hemisinus, occurrence a known in the pachychilid genus Paleoanculosa.

Thiaridae - Include of listed Nuttall two species Aylacostoma by (1990a) as Verena browni (Etheridge, 1879) and V.

lataguensis Nuttall, 1990, as well as Hemisinus kochi (Bernard!, 1856). Charadreon species (including C. eucosmius

(Pilsbry & Olsson, 1935) were tentatively placed in this family (Wesselingh, 1996), but should possibly be included in

the Pachychilidae.

Perimarine snails - Include Melongena woodwardi (Roxo, 1924), Nassarius sp. and three species of Odostomia (for the

latter, see van Aartsen & Wesselingh, 2000).

Pulmonata - Include the terrestrial Pebasiconcha immanis Wesselingh & Gittenberger, 1999, Orthalicus linteus

(Conrad, 1871) and an indeterminatespecies, as well as three species of the freshwater Planorbidae and a species of

Hebetancylus.

Corbiculidae - Corbicula cf. cojambitoensis Palmer, 1941.

Sphaeriidae - Contains two, possibly unnamed, species, one of which has been assigned to Eupera, the other to Pisid-

ium.

- Dreissenidae Includes Mytilopsis sallei (Recluz, 1849) and M. scripta (Conrad, 1874). Assignment to Mytilopsis fol-

lowing Kelleher et al. (1999) and not Wesselingh (1998).

- Includes undescribed Hyriidae Diplodon longulus Conrad, 1874, a possible Diplodon species and a Castalia species

(Figure 8).

- Includes Anadontites batesi Mycetopodidae Woodward, 1871 and A. capax (Conrad, 1874).

Pachydon obliquus Gabb - includes only this species.

Other Pachydon group- This group includes Pachydon tenuis Gabb, 1869, P. carinatus Conrad, 1871, P. amazonensis

P. 1990, P. Conrad, 1871, P. and (Gabb, 1869), trigonalis Nuttall, erectus ledaeformis (Dali, 1871) Pachydon sp. Wes- selingh (1993, figs 142, 143).

Pebasia group - This group includes Pebasia dispar (Comad, 1874) and Pebasia indet. -76-

- Ostomya group Ostomya papyria Conrad, 1874, Ostomya sp. indet.(listed in Wesselingh (1993, figs 158, 159) as

et indet.l in et Cryptomya sp., Gen. sp. (listed Wesselingh (1993, figs 152-154) as aff. Raetomya, Gen. sp. indet.2

(listed in Wesselingh 1993 as aff. Bushia: fig. 163) and Anticorbula sp. (listed in Wesselingh (1993, figs 155, 156) as

Guianadesma sp.). Anticorbula is considered a corbulid (following Nuttall, 1990) and not a lyonsiid (Simone, 1999).

Endemicity

The endemic/non-endemicstatus ofPebasian taxa is inferred on the basis of a number of criteria. Pebasian taxa known from deposits from intramontane basins in Ecuador, or from Venezuela, are not considered endemics (Mytilopsis

Corbicula cf. scripta (Conrad, 1874), cojambitoensis Palmer, 1941, Panamicorbula sp., Sheppardiconcha tuberculif- era (Conrad, 1874), and Neritina roxoi de Greve, 1938). The two extant members of the Pebas fauna (Mytilopsis sallei

(Recluz, 1849) and Hemisinus kochi (Bernard!, 1856)) are not endemic either. Furthermore, groups that are character- istic of ‘normal’ fluvial (Ampullariidae, Mycetopodidae, Hyriidae, Pachychilidae, Thiaridae, Sphaeriidae) or peri- marine settings (Melongena, Nassarius and Odostomia spp.) that occur in the Pebas Formation are assumed to be non- endemics. Within some of the other groups listed as endemics below taxa possibly occur that were presumably not en- demic (Lithococcus sp. in the Sioliella group and Pyrgophorus spp. in the ‘Other Cochliopinae’ group), but their num- bers are subordinate. The Tryonia groupposes a problem, since several Tryonia finds are known from Miocene depos-

Since Pebasian has outside its of Andean basins. none of the species been found western Amazonia so far, Tryonia is classified as endemic.

The endemicity ratio is the estimatedabundance of endemics in samples.

Endemics: Neritina ortoni group, Tryonia group, small Dyris group, large Dyris group, Sioliella group, other

other Pebasia and Cochliopinae, Pachydon obliquus, Pachydon group, group Ostomya group.

Non-endemics: Neritina roxoi group, Ampullariidae, Pachychilidae, Thiaridae, perimarine taxa, Pulmonata, Corbicula,

Sphaeriidae, Dreissenidae and Tellinidae.

Freshwater indicators

The following groups are considered as freshwater inhabitants (see also remarks in text): Neritina roxoi group, Ampul- lariidae, Pachychilidae, Thiaridae, Pulmonata (although this group includes both freshwater and terrestrial snails), Cor- bicula, Sphaeriidae, Flyriidae, Mycetopodidiae.

Marine indicators

The following taxa are considered as indicators of marine or perimarine conditions (salinity >2 psu): the molluscs Me-

woodwardi Odostomia nuttalli cotuhensis longena (Roxo, 1924), Nassarius sp., van Aartsen & Wesselingh, 2000, O.

Odostomia Panamicorbula indeterminate barnacles and the van Aartsen & Wesselingh, 2000, sp., Macoma sp., sp.,

Foraminifera Ammonia sp. and Haplophragmoides sp. (but see for the occurrence of foraminifers in non-marine envi- ronments Patterson et al, 1997). -77-

3. Appendix Taxonomic composition and ecological characteristics of the Pebas fauna..

SP AB FE LSC T F FB B HS

- 1 Neritidae 0.9%0,9% brow (scav) epf vag X X X X I Neritina 4(3)

~22 ArripullariidaeAmpullariidae 0.0% leafs/lich epf vag *~7~X X \AmpullariusAmpullarius s.l.s.l. 1 (0) ~

~33 Hydrobiidae 28,2% depf epf/infepf/inf vag Cochliopinae K~Y~'X X TryoniaTryonia 8(4) T~? (jen.Gen. nov. 2(0) X DyrisDyris 25 (6) ? X PyrgophorusPyrqophorus 2(0)2 (0) X Lyrodes7Lyrodes 2(0)2 (0) X X Onobops70nobops 4(0) X X NanivltreaNanivitrea 4(3) X X ? Sioliella I 9(5) *121 X Lithococcus 3(0) r? Tropidobora 3(1)3d) Littoridina __iiU2(1) JZ1Z? ? ? Toxosoma 7(1) 9 Liosoma -Jill2(1) £1 X Heleobops7Heteobops 2(0) yz ~44 Pachychilidae 0.6%0,6% brow epfepf vagvag

Sheppardiconcha 5(4) X X boryssaDoryssa 2(0) ZZ ~~56 Thiaridae 0,7% brow epfepf vagvag rX Aylacostoma 2(2) ~ Hemisinus X Hemisinus in)1 (1) y-jz)w Charadreon X Charadreon 2(1)2d) El Pe Perimarinerimarine taxa (6-8) 0J%0,1% 6 ~5]MelongenidaeMelongenidae carn/scav epfepf vagvag X || Melongena 1 (1)(D X ~

7 NassarlidaeNassariidae scav epf vagvag X Nassarius\7Nassaiius IJJS)1 (0) El „ 8 Pyramidellidae ' carn/scavcam/scav epf vagvag X \70dostomiaOdostomia 3(0) ZZ' PuPulmonata monata (9-12) 0,7%0 7% J 99|PlanorbidaePlanorbidae depf (brow)(brow) epf/inf vagvag~ Helisoma 2(0) y~X X Drepanotrema I...11 (0) TO10 FerrissidaeFerrissidae depf (brow)(brow) epf vagvag |I Hebetancylus 1 (0) ElX 11 „ TT Acavidae leafs/lich ' epf vag

Amazoniconcha 1 (D(1) X T?12 Bulimidae Bu Imidae leafs/lich ' epf vag X BulimusBullmus 1 (1)(D *1 pulmonale indet. 1 (0) X

~ TT13 Sphaeriidaetipnaeriidae 0,1%0,1% susfsusf inf/epfsessei~ Eupera 1 (0) *1X Sphaerium 1 (0) ElX TT14 „ Corbiculidae 0,1%0.1% susf (depf) inf ses | Corbicula 1 (1) ZZZZX X

~ TT 16 Hyriidae 0,4%0.4% susfsusf inf ses

Castalia 1 (0)(Q) y~X DiplodonDiplodon 2(1)2d) ElX T516 Mycetopodidae 0.3%0.3% susfsusf inf ses | Anodontites 2(2)2 (2) ZZX

~ 17 17 Dreissenidae 1.6% susfsusf epf ses mMytilopsis 2(2)2 (2) EZEZX X EZX T516 le Tellinidaelimdae 0.0%0.0% susf (depf) inf ses

| Macoma K0)1 (0) I I II |x~[x~'X X -78-

19[19 Corbulidae 166,5%66,5% I susf infinfsesses I II Corbulinae

*X | Panamicorbula ~nor1 (0) Pachydontinae

' Pachydon 12(10) ? Pebasia ? Pebasia 2(1)2(D Ostomya _2ilL2(1) ? 7?en. indet.i *> Gen. indet.1 2(0)2 (0) (Jen. Gen. indetindet.22 _Li2I.1 (0) Anticorbula X Anticorbula I 1 (0) I |x I I I 1 totaltotal 1136(55)136 (55)

SP estimated number of species (with described species in brackets)

AB estimated abundance corrected for 100% sum

FE feeding ecology

brow browsers

earn carnivores

depf depositfeeders

leafs

lich lichens

scav scavengers

susf suspension feeders

LSC life site characteristics

T terrestrial

FE obligate freshwater

FB fresh and/or brackish water

B obligate brackish/marine

HS hypersaline -79-

Appendix 4. Stable isotope data

S.D. of replicate analyses < 0,05, S.D. > 0,05 in bold(max. 0,22)

level species d13Cd13C Q.d18000O

686 Pachydon obliquus -9,8 -6,8

686 Pachydon obliquus -8,1-8.1 -6,3

686 Pachydon tenuis -9,7 -6,9

686 Pachydon carinatus -9,9 -6,8

686 Pachydon obliquus -9,6-9.6 -7,5

686 Pachydon cuneatus -6,1 -4,9

686 Pachydon erectus -7,7 -5,0

686 Pachydon obliquus -10,3 -7,8

686 Mytilopsis cf. sallei -9,2 -6,7

686 Pebasia dispar -9,4 -6,2

686 Diplodonlonguluslongulus -10,5 -5,9

686 Tryonia tubercutatatuberculata -6,4 -4,5

686 SioliellaSiollella cf. crassilabra -7,9 -5,7

686 Dyris sp.sp.11 -10,3 -5,4

686 Littoridina cf. crassa -6,2 -4,7

686 Dyris ortoniortoni -6,0 -4,5

686 Neritina roxoi -7,3 -7,3

686 NehtinaNeritlna roxoi -5,8 -5,0

686 Neritina ortoni -8,3 -5,4

686 Shepp. cf. coronatum -7,1 -6,3

685 Pachydon tenuis -7,1 -5,4

685 Pachydon erectus -7,1 -4,6

685 Neritina ortoni -7,5 -7,3

685 Toxosoma eboreum -7,9 -7,4

685 Dyris tricarinatus -4,9 -4,8

685 Dyris sp.1 -10,3 -5,9

685 Neritina ortoni juv. -6,5 -5,3

685 ToxosomaToxosoma eboreumeboreum -7,1 -6,5

685 Mytilopsis cf. sallei -10,4 -7,5

685 Pebasia dispar -8,0 -6,7

685 Dyris ortoni -6,9 -4,4

685 Pachydon obliquus -6,7 -4,1

685 Pachydon obliquus -5,8 -3,7

685 SioliellaSiollella crassilabra -10,5 -9,2

685 Pachydon amazonensis -7,2 -4,1

685 Pachydon erectus -8,9 -5,2

533 Pachydon obliquus -8,3 -6,8

533 Pachydon obliquus -7,5 -8,0

533 Pachydon obliquus -8,9 -6,1-6.1

533 Pachydon obliquus -7,7-7,7 -5,7

533 Pachydon obliquus -8,0-8,0 -6,3

533 Pachydon obliquus -6,8 -5,5

533 Pachydon obliquus -9,5-9,5 -8,1

533 Mytilopsis scripta -10,3 -7,2

533 Pachydon carinatus -8,5-8,5 -5,7-5,7

533 Dyris tricarinatus -10,2 -7,0-7,0

533 Toxosoma eboreum -9,3-9,3 -7,4-7,4

533 Dyris hauxwelli -9,7-9,7 -6,5

533 Dyris tricarinatus -8,5-8,5 -6,5-6,5

533 Onobops sp.1 -8,6-8,6 -6,9-6,9

533 Mytilopsis scripta -10,5 -7,3-7,3

536 Pachydon obliquus -7,3 -6,5-6,5

536 PachydonPachydon obliquus -6,8-6,8 -7,2-7,2 -80-

level species d13C Q.d18000 O

536 Pachydon obliquus -8,6-8,6 -7,9-7,9

536 Pachydon obliquus -6,0-6,0 -7,1

536 Pachydon obliquus -9,9-9,9 -10,1

536 Pachydon obliquus -8,4-8,4 -8,5

536 Pachydon tenuis -8,0-8,0 -6,7

536 Pachydon tenuis -8,0-8,0 -6,4

536 Toxosoma eboreum -6,2-6,2 -6,3

536 Toxosoma eboreum -6,6-6,6 -7,2

536 Tryonia minuscula -7,2-7,2 -8,8

536 Tryonia minuscula -6,3-6,3 -6,7

536 Dyris ortoni -7,1-7,1 -7,2

536 DyrisDyhs hauxwelli -7,2-7,2 -7,0

538 Pachydon tenuis -5,0-5,0 -4,9

538 Pachydon tenuis -6,6-6,6 -4,8

538 Pachydon tenuis -4,8 -5,0

538 Pachydon tenuis -9,0 -4,8

538 Pachydon tenuis -4,6-4.6 -4,6

538 Mytilopsis cf. sallei -4,9-4,9 -6,1

538 Pachydon tenuis -1,6-1,6 -4,1

538 indet. Pulmonata -4,8 -4,6

538 Mytilopsis cf. sallei -5,8 -2,8

538 Pachydon obliquus -4,0 -4,0

538 DyrisDyhs lintealintea -5,2 -1,8

538 DyrisDyhs lintealintea -3,5 -3,2

538 DyhsDyris lintealintea -3,2 -3,1

538 DyrisDyhs lintea -4,8 -4,0

538 DyrisDyhs lintealintea -5,0 -5,2

538 DyrisDyhs lintealintea -2,2 -3,3

538 DyrisDyhs lintealintea -3,0 -2,8

538 Tryonia minuscula -3,4 -4,2

538 Tryonia minuscula -4,7 -4,9

539 PachydonPachydon obliquus -8,1 -6,5

539 PachydonPachydon obliquus -5,4 -5,0

539 PachydonPachydon obliquus -5,9 -5,0

539 PachydonPachydon obliquus -5,7 -3,5

539 PachydonPachydon tenuis -5,5 -4,2

539 PachydonPachydon obliquus -7,1 -4,7

539 PachydonPachydon obliquus -7,1 -5,3

539 PachydonPachydon tenuis -7,4 -4,3

539 DyrisDyhs tricarinatusthcahnatus -7,8 -4,8

539 DyrisDyhs tricarinatusthcahnatus -5,7 -6,3

539 DyrisDyhs tricarinatusthcahnatus -8,9 -6,8

539 Tryonia minuscula -8,4 -6,7

539 Tryonia minuscula -5,8 -3,1

539 Tryonia minuscula -5,2 -2,7

542 PachydonPachydon obliquus -7,9 -8,0

542 PachydonPachydon obliquus -7,9 -7,9

542 PachydonPachydon obliquus -9,5 -9,1

542542 PachydonPachydon obliquus -7,5 -8,0

542 PachydonPachydon obliquus -8,1 -9,0

542 Toxosoma eboreumeboreum -6,9 -7,6

542542 Toxosoma eboreumeboreum -6,8 -7,9

542542 PachydonPachydon erectus -7,9 -8,8

542 Tryonia minuscula -7,5 -8,0

542 DyrisDyhs hauxwelli -6,8 -8,7

Sioliella indet. 542 Sioliella spec, -7,7 -7,2 -81 -

Stable isotope data of growth increments standard deviation of 5 replicate measurements < 0,05

Diplodon sp. Anadontites Diplodon Pachydon

(Itaya River) trapezialus longulus erectus

(Amazon (Pebas(Pebas (Pebas(Pebas

River) Formation) Formation) growth-growth- increment DI3CD13C DIBOD180 DI3CD13C DIBOD180 DI3CD13C DIBOD180 DI3CDISC DIBOD180

1 -12,7-12,7 -8,7 -14,1 -7,0 -11,0 -8,9 -11,8 -6,9

2 -14,6-14,6 -9,3 -12,1 -9,2 -11,0 -8,0 -12,2 -6,1

3 -14,3 -7,1 -10,5 -6,3 -11,9 -8,8 -11,6 -6,0

4 -13,0 -6,1 -9,0-9,0 -6,4 -13,1 -8,9 -11,7 -5,8

5 -12,4-12,4 -6,2 -11,6 -7,3 -13,0 -7,8 -13,1 -5,5

6 -13,3-13,3 -5,4 -14,3 -4,6 -11,0 -8,7 -11,3 -6,1

7 -13,4 -6,4 -15,5 -5,0 -11,5 -7,7 -6,3 -5,4

8 -13,9 -5,9 -15,7 -4,5 -12,1 -8,6 -7,2 -4,6

9 -12,4 -8,8 -16,0 -4,0 -12,6 -8,5 -9,4 -6,0

10 -14,8 -10,3 -15,5 -4,0 -11,7 -8,2 -10,3 -5,9

11 -14,3 -7,3 -14,2 -0,9 -12,3 -9,5 -8,2 -6,2

12 -13,1 -6,4 -14,3 -1,6 -13,1 -9,5 -9,7 -6,3

13 -12,6 -5,7 -14,4 -2,0 -11,5 -7,8 -13,7 -6,4

14 -14,2 -7,0 -12,6 -1,2 -12,7 -8,4 -10,5 -7,2

15 -13,1 -6,2 -13,2 -1,5 -12,2 -8,2 -8,9 -6,1

16 -12,8 -5,8 -12,9 -1,7 -12,4 -9,8 -10,2 -6,3

17 -14,6 -9,3 -12,1 -1,4 -13,4 -9,7 -13,0 -7,2

18 -13,8 -7,1 -12,8 -1,9 -13,5 -9,7 -10,8 -6,1

19 -14,4 -6,5 -12,8 -4,0 -12,3 -8,3 -5,3 -5,0

20 -13,4 -6,2 -14,0 -6,7 -11,2 -7,8 -7,4 -5,5

21 -15,2 -7,5 -14,1-14,1 -6,9 -11,5 -8,3

22 -14,5 -6,5 -15,2 -6,9 -11,5 -8,6

23 -15,2 -6,8 -12,2-12,2 -6,8

24 -14,8 -8,2 -13,8-13,8 -6,2

25 -14,9 -8,3 -12,2-12,2 -6,0

26 -13,9 -7,9 -10,9-10,9 -5,0

27 -14,1 -6,7 -11,4-11,4 -3,1

28 -14,5 -6,8 -12,2-12,2 -3,1

29 -14,2 -6,6 -11,8-11,8 -2,9

30 -14,3 -8,1 -13,0-13,0 -2,6

31 -14,4 -7,9 -12,7-12,7 -3,1

32 -13,9 -7,3 -12,9-12,9 -3,3

33 -14,1 -8,4 -12,5-12,5 -3,6

34 -13,5 -7,9 -12,7-12,7 -3,7

35 -13,4 -7,0 -12,8-12,8 -3,8

36 -13,8 -6,5 -13,0-13,0 -3,8

37 -13,9 -7,3-7.3

38 -13,4 -6,8