The tribe Chlamydini was highly diversified in the marine Neogene of southern South America, reaching its maximum taxonomic diversity during the Miocene. However, the evolutionary relationships of South American taxa remain uncertain. This is the first phylogenetic analysis based on a large morphological matrix on Pectinidae, which is focusing on South American taxa and species related to Chlamys s. The phylogenetic analysis is based on a matrix composed of 145 shell characters scored for 48 species and multiple searches were conducted using equal and implied weighting. Two new monophyletic clades are defined, Multiplicata and Pauciplicata. The first includes Dietotenhosen, Ckaraosippur, Zygochlamys, Moirechlamys (South America), the Northwest Pacific Azumapecten, and the Northeast Pacific Chlamys hastata. Pauciplicata is represented by Chokekenia (Patagonia Argentina), Laevichlamys (tropical Atlantic and Indo-Pacific), Semipallium (Indo-Pacific), Swiftopecten (South America and North Pacific), and Jorgechlamys +Reticulochlamys (Patagonia, Argentina). All these genera are monophyletic except for the paraphyletic Jorgechlamys. The oldest documented occurrence of the tribe is Semipallium foulcheri from the early Oligocene, which is a derived taxon that pushes the divergence time of basal genera to the Eocene–Oligocene boundary, generating ghost lineages in several clades, except for Jorgechlamys+Reticulochlamys. Pauciplicata and Multiplicata diverge in the early history of the tribe, at the Eocene–Oligocene boundary. Future analyses are necessary to gain a better understanding of the taxonomic arrangement of this poorly understood tribe to discover the relationships on its deepest nodes, which probably allow to resolve many ghost lineages.

The present contribution is focused on 17 species of Chlamydini from the late Cenozoic of Argentina, Chile, and Peru and it comprises 48 species of pectinids in total. Two South American species of Psychrochlamys are also included. Most of those taxa were studied from specimens in hand and some through high-quality images after being compared with proper systematic descriptions.

Phylogenetic searches, branch supports, and tree calibration.

The software TNT v1.5 (Goloboff & Catalano 2016) was used to conduct heuristic tree searches because the high number of herein included taxa. Space was configured for 20000 trees in memory and the searches were conducted applying maximum parsimony criterion, starting from 100 replicates of Wagner trees with random addition sequence (RAS) of taxa, which is followed by Tree Bisection and Reconnection (TBR) branch-swapping algorithm saving 10 trees per replication. To minimize the effect of the homoplasy, characters were weighted (as suggested by Goloboff et al. 2003; Goloboff et al. 2008), which allows obtaining more reliable and stable results when a complex matrix that includes a high number of homoplastic characters is analysed.

Search strategies involve two weighting arrangements: equal weights and implied weighting (Goloboff 1993), the latter using concavity (k) values from 1 to100 by increments of 1, and both strategies under traditional searches.

Clades recovered in different search strategies were illustrated through sensitive grids (Wheeler 1995; Pérez, 2019). To measure branch support, the resampling method was used by means of frequency differences (GC), and the Jackknife (jk) index (Farris et al. 1996) was computed with a p=0.14 (equivalent to remove 10% of the characters) (Goloboff et al. 2003) for1000 pseudo-replicates. The consistency (CI) and retention (RI) indices were computed (Farris 1989). Common synapomorphies of the selected topology (k=48–68) are described and listed below in the results section.

Phylogenetic time-scaling

One of the most frequent topologies (implied weighting, k=48–68, recovered for twenty values of concavity) was selected to be temporally calibrated using R (R Core Team 2019 3.6.0). The packages ‘paleotree’ (Bapst 2012), ‘ape’ (Paradis et al. 2004; Popescu et al. 2012) and ‘strap’ (Bell & Lloyd 2014) were loaded to read the matrix and tree topology, and implement different functions in R. The functions ‘timePaleoPhy’ and ‘geoscalePhylo’ were applied by means of the minimum branch length (mbl) method of calibration with a mbl of 0.1 My, in which a minimum branch duration is setting a priori.