Computational protocol: An analysis of pterosaurian biogeography: implications for the evolutionary history and fossil record quality of the first flying vertebrates

Similar protocols

Protocol publication

[…] The cladistic biogeographical analyses described below require information on pterosaurian phylogeny and the geographical and stratigraphical ranges of taxa. The reference phylogeny used in this analysis is that of Andres et al. () which includes 108 terminal pterosaurian taxa ranging from the Late Triassic to the Late Cretaceous in age (Figures and ). This phylogeny is fully resolved apart from a single trichotomy linking the three terminals comprising Nyctosauridae. Cladistic biogeographical methods, such as Treefitter (see below), cannot deal with polytomies, so this trichotomy must be removed prior to analysis. All three of the nyctosaurid taxa occur in the Late Cretaceous, two in North America and one in South America (Figure ). As this trichotomy involves just two geographical areas, we can simply resolve it into a set of bifurcating relationships without significant alteration of the biogeographical ‘signal’ in the data (i.e. all possible resolutions support a close relationship between North and South America in the Late Cretaceous). Here, therefore, we have arbitrarily resolved this trichotomy as [Muzquizopteryx coahuilensis (Nyctosaurus gracilis, Nyctosaurus lamegoi)] in the three versions of the data-set (i.e. ‘all taxa’, ‘Cretaceous taxa’ and ‘Late Cretaceous taxa’ – see below for discussion of data-set partitioning).The geographical and stratigraphical ranges of pterosaurian taxa were obtained from Barrett et al. () with some modifications based on The Paleobiology Database and Fossilworks. We have assigned each pterosaur to one or more of five areas: EA, East Asia (e.g. China and Mongolia); CA, Central Asia (e.g. Kazakhstan); EU, Europe; NA, North America; SA, South America (note that the Late Triassic Eudimorphodon cromptonellus from Greenland is here assigned to ‘Europe’ because these two areas were in close contact at this time; see Brusatte et al. , Figure ). These areas could be divided more finely: for example, we could assign South American taxa to Patagonia, Brazil and Chile. However, the current pterosaurian data-set is relatively small compared with those for other groups, such as dinosaurs, and further subdivision of areas would decrease the ability of cladistic biogeographical analyses to recover a distribution pattern common to several clades. As with the time-slicing of data (see below), the selection of areas used in a biogeographical analysis is based on the judgement of the investigators: too few areas or too many areas can render the results both biogeographically meaningless and statistically non-significant. We would need a somewhat larger data-set before further subdivision of areas could be attempted.The rhamphorhynchids Nesodactylus and Cacibupteryx from the Oxfordian Jagua Formation of Cuba (Gasparini et al. ) are potentially problematic because Cuba lies in the Caribbean corridor between North and South America. Thus, these pterosaurian taxa could potentially be assigned to area NA, SA or NA+SA. Gasparini and Itorralde-Vinent (, p. 354) suggest that an emergent ridge stretched from Florida to the Yucatan during the Oxfordian and probably represents the source for the terrestrial fauna present in the Jagua Formation. This means that the Jagua Formation deposits were laid down on the continental margin of Laurasia and probably received North American terrestrial taxa. Therefore, in our analyses, we have provisionally assigned Nesodactylus and Cacibupteryx to area NA.Certain terminal taxa have been pruned from some or all time-sliced data-sets because their geographical area occurs only once in that data-set. For example, Arambourgiania philadelphiae occurs in the Late Cretaceous of Jordan. The latter lies on the Arabian plate that, in the Late Cretaceous, was still connected to the rest of the African landmass (e.g. Smith et al. ; Scotese ). Although Africa has produced some fragmentary pterosaurian remains from other time periods (see Table ), Arambourgiania is the only ‘African’ pterosaur in the phylogeny. Cladistic biogeographical analyses cannot produce meaningful reconstructions of the relationships of areas that occur only once in a data-set. This is because such analyses typically work by determining the level of congruence between area relationships in two or more clades (Nelson and Platnick ): such congruence cannot be assessed when an area occurs only once. Therefore, Arambourgiania has been pruned from all data-sets. Other areas, such as Central Asia (CA), occur several times in the ‘all taxa’ data-set, but occur only once in some of the time-sliced data-sets (e.g. ‘Late Cretaceous’). When time-slicing produces such ‘singleton’ areas, the relevant taxa have been deleted.The pterosaurian biogeographical data-set has been analysed as a whole (‘all taxa’ data-set) and in a variety of different time slices (e.g. ‘Late Jurassic’ and ‘Early Cretaceous’; see Tables and ). This time-slicing approach is based on the principle that biogeographical patterns change through time in a network-like (reticulate) rather than hierarchical way (Grande ; Lieberman ; Hunn and Upchurch ; Upchurch and Hunn ; Upchurch et al. ; Halas et al. ). This means that a single biogeographical data-set might contain two or more temporally distinct, but incongruent, distribution patterns that obscure each other. Time-slicing is therefore an exploratory technique designed to search data-sets at various temporal scales to elucidate how many separate patterns exist and how these are distributed. Turner () proposed a refinement to the application of time-slicing in cladistic biogeographical analyses. He noted that time-sliced cladograms may include divergence events that actually occurred prior to the time slice in question. For example, the cladogram (W (X (Y, Z))) includes three nodes. Suppose taxa W, Y and Z occur in time slice t2, but taxon X occurs in the earlier time slice t1. Time-slicing this cladogram so that it contains only taxa from t2 gives (W (Y, Z)), but the node representing the most recent common ancestor of W and (Y, Z) must also lie in time slice t1 because of the age of X. Under such circumstances, Turner's logic argues that W should be pruned from the t2 data-set so that the biogeographical analysis only considers divergence events that occurred in t2. Upchurch et al. () and Turner () applied Component version 2.0 (Page ) and Treemap (Page ) in order to search for biogeographical signals in their time-sliced data-sets. These methods require a single cladogram topology. This requirement means that the only way to remove divergence events that lie outside of the time slice under investigation is to prune terminal taxa from the cladogram. Fortunately, the analytical method applied here (i.e. Treefitter, see below) can search for biogeographical signals simultaneously across two or more tree topologies, which means that at least some of the phylogenetic events that lie outside of a time slice can be removed without loss of terminal taxa. For example, suppose we have a clade represented by seven taxa A–G, with the relationships ((A, (B, C)), (D, (E, (F, G)))), with D occurring in time slice t1 and ABCEFG occurring in t2. The t2 time-sliced cladogram contains the relationships ((A (B, C)), (E, (F, G))). The t1 age of taxon D means that the node representing the most recent common ancestor of (A (B, C)) and (E, (F, G)) is dated at t1 and should be removed from the t2 time slice. This can be done in Treefitter without any further deletions of terminal taxa simply by treating (A (B, C)) and (E, (F, G)) as two separate clades (rather than two sister clades in a single cladogram) in the data-set. This protocol of terminal taxon pruning and clade separation has been applied here in order to derive the time-sliced data-sets for pterosaurs.We have not analysed the ‘Late Triassic’, ‘Early Jurassic’ or ‘Late Triassic+Early Jurassic’ time-sliced data because these contain pterosaurs from just one area (i.e. all but one of the taxa come from Europe and the singleton is from Greenland, which is treated here as part of Europe; see above). Similarly, we have not analysed the ‘Late Cretaceous’ data-set because this has multiple representations of only two areas (North America and Europe), whereas all other areas (Central Asia, East Asia and South America) occur only once each. Application of a cladistic biogeographical analysis to a data-set containing taxa from just one or two areas is not meaningful: a minimum of three areas is required in a manner analogous to the way phylogenetic analysis is only meaningful when applied to three or more taxa. Here, therefore, information on the Late Triassic, Early Jurassic and Late Cretaceous biogeographical histories of pterosaurs is derived from the analyses of the ‘all taxa’, ‘Late Triassic–Late Jurassic’ and ‘Cretaceous’ data-sets (see Tables and ).The formatted Treefitter data files are presented in the online electronic supplement. […]

Pipeline specifications

Software tools TreeFitter, TreeMap
Databases PBDB
Application Phylogenetics