Synchronized Geologic and Biotic Change

During the Late Ordovician, the Taconian orogeny developed along the eastern margin of Laurentia as a result of diachronous collisions with an island arc and/or microplates [3–5] (Figure 1). The Taconian orogeny began with the Blountian tectophase along the extreme southeast margin of Laurentia during the Sandbian Stage [4,5]. During the Blountian tectophase, tropical carbonate deposition was widespread throughout Laurentia, and biogeographic and faunal divisions were relatively stable [5,6]. Following a brief quiescent interval, the locus of tectonic activity shifted northward to the New York promontory during the Taconic tectophase of the late Sandbian to Katian Stages [3,4]. The onset of the Taconic tectophase occurred during the middle Mohawkian (M5 sequence, late Sandbian, Figure 2) [4,5]. The associated crustal flexure, regional transgression, and siliciclastic influx coeval with an episode of global cooling introduced major lithologic and faunal changes throughout Laurentia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Late Ordovician paleogeography of Laurentia.

Biogeographic areas analyzed are: (1), Anticosti Island (2);, Appalachian Basin (3);, Cincinnati Basin (4);, Central Basin (5);, Southern Midcontinent; and (6), Northern Midcontinent. Paleogeographic reconstructions (e.g. [45]) support the separation of these areas by physical barriers, such as structural platforms, trenches, and intracratonic arches, indicated with grey lines. Primary locus of tectonic activity during the Blountian (B) and Taconic (T) tectophases are indicated along the subduction zone.

https://doi.org/10.1371/journal.pone.0068353.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Stratocladograms.

Species-level evolutionary and biogeographic patterns for Glyptorthis Foerste, 1914 [28], Plaesiomys Hall and Clarke, 1892 [29], and Hebertella Hall and Clarke 1892 [29] mapped onto a stratigraphic column at the level of depositional sequence. Ghost lineages reconstructed via equal phylogenetic split. Abbreviations: Ro, Rocklandian; Ki, Kirkfieldian; G, Gamachian; H, Hirnantian. Timescale modified from Holland & Patzkowsky [6].

https://doi.org/10.1371/journal.pone.0068353.g002

Beginning in the middle Mohawkian (M5 sequence), carbonate deposits of eastern Laurentia reflect a heterozoan assemblage typified by cool water and/or non-photosynthetic organisms, whereas tropical-style carbonates became restricted to the continental interior and marginal basins [6,7]. This shift in carbonate deposition exhibits both regional and global geological signatures. Tectonic restructuring associated with the Taconic tectophase may have facilitated increased subsidence, thermohaline circulation, upwelling, and nutrient flux [5,8,9]. In addition, the mid-M5 sequence Guttenberg carbon isotope excursion (GICE) records a significant global cooling event corresponding to an episode of pre-Hirnantian Late Ordovician glaciation [10]. The synergistic combination of flexural downwarping and nutrient flux associated with the Taconic tectophase and coeval global cooling facilitated the development of an upwelling zone originating from the Sebree Trough which introduced cool, nutrient rich water into eastern Laurentia [5–7].

Also during the M5 sequence, faunas of the Appalachian Basin transitioned from predominantly warm to cool water adapted. The biotic overturn was accomplished primarily through immigration of extra-basinal taxa into a region following regional generic extinction [2]. For example, trilobites and some brachiopods interpreted as adapted to cooler water conditions participated in northward biogeographic expansion in associated with the spread of temperate carbonates [11,12].

The Richmondian Stage (late Katian [C4-C6 sequences], Figure 2) marked a return to tropical-style carbonate deposition across Laurentia which continued until the Hirnantian Stage. Whether the return to tropical conditions was due to a brief interval of global warming related to the Boda Event [13] or termination of upwelling due to basinal infilling [6] remains uncertain. Nonetheless, the C3/C4 sequence boundary corresponds to the return of tropical-style carbonate deposition in eastern Laurentia and marks the initiation of a large scale cross-faunal immigration event into the Cincinnati region known as the Richmondian Invasion [1,14,15].

The paleoequatorial region of the western United States and Canada has long been considered the biogeographic source of the Richmondian invaders [14,16]. However, this generalization has not previously been tested by detailed quantitative analysis and therefore requires further investigation because there were many possible source regions with different pathways of invasion available during the Late Ordovician (e.g., [17]). Invader taxa were geographically widespread throughout the Late Ordovician. Some invader genera simultaneously occupied multiple basins during the Richmondian Stage, such as Wyoming [18], Canada [19], and Tennessee [20]. The spatiotemporal pervasiveness of invasive genera limits the biogeographic precision of studies limited to genus-level patterns because a single invading species may belong to a genus present in multiple biogeographic areas. Consequently, a species-level phylogenetic framework, such as the one provided herein, is required to rigorously test competing hypotheses regarding the biogeographic sources of the Richmondian invaders and elucidate the geologic controls on Late Ordovician paleobiogeographic patterns.

Macroevolutionary Impact of Biogeographic Change

Results of these analyses indicate fundamental differences in biogeographic and macroevolutionary patterns between the T1 and T2 intervals (Table 1). Analysis of speciation mode at cladogenetic events (Figure 3) identified vicariance as the dominant mode of speciation during the T1 interval, operating in 77% of the 26 discernible speciation events; whereas only 15% of 13 discernible speciation events during the T2 interval were attributable to vicariance (Figure 2). This pattern arises from differences in evolutionary differentiation of the genera analyzed through time (Figure 3). The taxon-area cladogram for species of Glyptorthis indicates that the genus originated in the Southern Midcontinent region and subsequently underwent a major dispersal event into the Appalachian and Central Basins (Figure 3). Accordingly, most of the ancestral lineages within Glyptorthis are reconstructed to occupy multiple tectonic basins (Figure 3). Because the early evolutionary history of the clade is characterized by geographically widespread species, most speciation events during T1 were accomplished via range contractions that led to speciation by vicariance. In contrast, ancestral species of Plaesiomys and Hebertella occurred predominantly in a single tectonic basin (the Northern Midcontinent and Cincinnati Basin, respectively) with frequent episodes of speciation via dispersal into neighboring tectonic basins during T2 (Figure 3). Documentation of this pattern in additional taxa would be desirable, but no additional species-level phylogenetic hypotheses exist for contemporaneous Laurentian endemics. The recovered pattern of high vicariance during T1 followed by high dispersal during the T2 is congruent with patterns documented by other researchers in genus-level biogeographic analyses among a diverse suite of Laurentian clades including corals, graptolites, gastropods, bivalves, monoplacophorans, ‘inarticulate’ brachiopods, and trilobites [2,33–36]. Although these patterns were documented at a higher taxonomic level, the consistent signal across clades and taxa supports the interpretation that our recovered pattern cannot be attributed to sampling bias alone.

The diametrically opposed patterns of speciation and biogeographic differentiation between T1 and T2 reflect different biogeographic controls operating before and after the M4/M5 sequence boundary. This discrepancy is clearly illustrated by the divergent topology of vicariance trees between times slices and the dissimilarity of the geodispersal and vicariance trees within each time slice (Figure 4). The T1 vicariance tree is well resolved and indicates a strong biogeographic relationship between the Appalachian Basin and the Southern Midcontinent region. The numerous T1 vicariance events produce a topology supported by strong tree statistics (Figure 4). Conversely, the strict consensus of the T1 geodispersal tree is unresolved, which reflects conflicting biogeographic patterns among the few speciation events occurring via dispersal in T1 (compare Figures 2, 3). The T2 vicariance tree is also well resolved and supported despite the limited number of vicariance events during the T2 interval, which indicates little conflict amongst recovered patterns. Notably, the geodispersal tree in T2 is even more unresolved than in T1 although speciation occurred predominantly by dispersal during this interval. The lack of resolution arises from conflicting data regarding dispersal pathways during T2 (compare Figures 2, 3). When the two most parsimonious geodispersal trees are examined individually, one is consistent with a close relationship between the Cincinnati Basin and the Midcontinent regions (supported by biogeographic affinities of G. insculpta), whereas the other tree is consistent with a more recent exchange of taxa between Anticosti Island and the Midcontinent region (supported by biogeographic affinities of P. subquadratus). These data indicate that both dispersal pathways were operational during the Richmondian Invasion.

The level of congruence between vicariance and geodispersal trees is indicative of the style of geologic processes impacting macroevolutionary patterns [22]. Congruent trees support the influence cyclical geologic processes, such as cyclical sea level or climatic changes, in which barriers arose (causing vicariance) and fell (facilitating dispersal) in the same relative order. Incongruous trees support the primacy of singular geologic forcers, such as tectonics or linear climate change [22]. In this case, the dissimilar vicariance and geodispersal tree topologies indicate that cyclical processes were less important than singular processes in facilitating Late Ordovician faunal differentiation.