Palaeogeography, Palaeoclimatology, Palaeoecology 214 (2004) 195–223 www.elsevier.com/locate/palaeo Extinction and recovery patterns of scleractinian corals at the Cretaceous-Tertiary boundary Wolfgang Kiesslinga,*, Rosemarie C. Baron-Szabob aInstitute of Paleontology, Museum of Natural History, Humboldt-University Berlin, Invalidenstr. 43, 10115 Berlin, Germany bSmithsonian Institution, Department of Zoology, National Museum of Natural History, W-329, MRC-163, Washington, DC 20560, USA Received 1 July 2003; received in revised form 20 April 2004; accepted 20 May 2004 Abstract The extinction and recovery of scleractinian corals at the Cretaceous–Tertiary (K–T) boundary was analyzed based on a global database of taxonomically revised late Campanian to Paleocene coral collections. In contrast to earlier statements, our results indicate that extinction rates of corals were only moderate in comparison to other marine invertebrates. We have calculated a 30% extinction rate for Maastrichtian coral genera occurring in more than one stratigraphic stage and more than one geographic region. Reverse rarefaction suggests that some 45% of all coral species became extinct. Photosymbiotic (zooxanthellate) corals were significantly more affected by the extinction than azooxanthellate corals; colonial forms were hit harder than solitary forms, and among colonial forms an elevated integration of corallites raised extinction risk. Abundance, as measured by the number of taxonomic occurrences, had apparently no influence on survivorship, but a wide geographic distribution significantly reduced extinction risk. As in bivalves and echinoids neither species richness within genera nor larval type had an effect on survivorship. An indistinct latitudinal gradient is visible in the extinction, but this is exclusively due to a higher proportion of extinction-resistant azooxanthellate corals in higher-latitude assemblages. No significant geographic hotspot could be recognized, neither in overall extinction rates nor in the extinction of endemic clades. More clades than previously recognized passed through the K–T boundary only to become extinct within the Danian. These failed survivors were apparently limited to regions outside the Americas. Recovery as defined by the proportional increase of newly evolved genera, was more rapid for zooxanthellate corals than previously assumed and less uniform geographically than the extinction. Although newly evolved Danian azooxanthellate genera were significantly more common than new zooxanthellate genera, the difference nearly disappeared by the late Paleocene suggesting a more rapid recovery of zooxanthellate corals in comparison to previous analyses. New Paleocene genera were apparently concentrated in low latitudes, suggesting that the tropics formed a source of evolutionary novelty in the recovery phase. D 2004 Elsevier B.V. All rights reserved. Keywords: Cretaceous–Tertiary boundary; Mass extinction; Corals; Evolution; Reefs * Corresponding author. E-mail address: [email protected] (W. Kiessling). 0031-0182/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.palaeo.2004.05.025 196 W. Kiessling, R.C. Baron-Szabo / Palaeogeography, Palaeoclimatology, Palaeoecology 214 (2004) 195–223 1. Introduction compatible with the impact scenario at the K–T boundary. The mass extinction at the Cretaceous–Tertiary boundary (K–T, officially Cretaceous–Paleogene) still provokes many questions considering the rea- 2. Database and methods sons, magnitude, speed, selectivity and geographic patterns of extinction and subsequent recovery. The 2.1. Database Chicxulub impact is currently considered the most likely culprit for the mass extinction, but apart from Data in this paper are part of a comprehensive other potential causes, such as Deccan Trap volcan- geographic database on the K–T boundary (KTbase), ism (McLean, 1985; Courtillot, 1999), the impact- which currently comprises geological, mineralogical kill effect remains ambiguous in many instances. and paleontological information from 490 K–T Extinctions in some clades were severe, while others boundary sites (Kiessling and Claeys, 2001; Claeys show almost no taxonomic change at the K–T et al., 2002). A K–T boundary site is defined as an boundary. The most dramatic examples were area (100 km2 or less) where at least Maastrichtian reported from the marine plankton, where calcareous and/or Danian sediments are preserved. Closely nannoplankton and planktonic foraminifers faced spaced sections in the same depositional environment extremely high extinction (Gartner, 1996; Arenillas are combined for the summary data (e.g., sedimentol- et al., 2002), while radiolarians and dinoflagellates ogy and geochemistry) but paleontological data are only show some ecological response but almost no partitioned as finely as possible from published extinction (Hollis, 1997; Brinkhuis et al., 1998). sources. For the particular purpose of this study, the Among invertebrate macrofossils, detailed studies concept for K–T boundary sites has been widened to have revealed a similar though less extreme dispar- include also sites with only late Campanian or late ity: bivalves including the rudists exhibit a 63% Paleocene sediments preserved. This was done to generic extinction (Raup and Jablonski, 1993), compare the extinction and recovery at different whereas the extinctions in sea urchins was shown temporal scales. to have been not more than 36% (Smith and Jeffery, Data in KTbase were largely extracted from the 1998). Clearly, additional datasets have to be published literature but personal observations and analyzed at global scales to extract general patterns unpublished data are also included. Paleontological of the end-Cretaceous mass extinction and to get a data are stored in faunal lists representing individual better idea of the underlying cause(s). fossil collections. Each taxon reported in a collection This study documents global geographic patterns constitutes a taxonomic occurrence in the database. of extinction and recovery in scleractinian corals in Abundance data are available for less than one third of the Maastrichtian and Paleocene. We emphasize five the taxonomic occurrences. We thus used the number key points: of occurrences as an approximation of recorded abundances and applied this criterion for resampling. 1. the intensity of the end-Cretaceous coral extinc- All taxonomic data and some 80% of the occur- tions and the pace of recovery; rence data have been taxonomically revised at the 2. the ecological selectivity of coral extinctions; species-level (Baron-Szabo, 2002;B.-S.workin 3. the impact of coral extinctions on the global reef progress) and stratigraphic assignments were updated ecosystem; for the purpose of this study. The occurrence table 4. the geographic patterns of coral extinction and with updated taxonomic information is available as an recovery; electronic appendix (electronic appendix see Back- 5. a comparison of the results with other benthic ground Data Set). Only taxonomic data resolved at invertebrate groups. least to the genus level are represented in KTbase. Species-level taxonomic information is stored in the Although we do not focus on potential causes of the database whenever possible. Coral data from 228 late coral extinctions, we also discuss if our results are Campanian to Paleocene collections (Fig. 1) comprise W. Kiessling, R.C. Baron-Szabo / Palaeogeography, Palaeoclimatology, Palaeoecology 214 (2004) 195–223 197 Fig. 1. Global distribution of coral-bearing K–T-boundary sites (with at least genus-level taxonomic information) (triangles) and number of occurrences per 308 grids (pies—only shown for grids with more then five occurrences) plotted on a 65 million years paleogeographic reconstruction (modified from Golonka, 2002). Size of pies is proportional to the number of occurrences in each grid. Brick pattern:z-like (=taxa that are likely to have hosted zooxanthellae) coral occurrences; black fill:az-like (=taxa that are likely to not have hosted zooxanthellae) coral occurrences. 1235 taxonomic occurrences, 187 genera and 460 460 valid species). The main reason for the low species. The low ratio of occurrences to species occurrence/diversity ratio is probably the way coral richness indicates that (1) corals in this time interval data are usually reported in the literature. Most coral are underexplored and the true biodiversity is far from collections stem from whole outcrops or local areas being completely known (e.g., very high), (2) coral and usually combine several beds or even lithostrati- occurrences are usually reported on rather large spatial graphic members. This limits the applicability of and temporal scales, and/or (3) in spite of our resampling techniques such as rarefaction. taxonomic revisions the corals may still suffer over- Sepkoski’s compendium of the stratigraphic ranges splitting. Our revisions have reduced the taxonomic of marine animal genera (Sepkoski, 2002) was used to noise considerably by removing many subjective determine long-term diversity dynamics of scleracti- synonyms (632 nominal species were parsed into nian corals and to compare two independent global 198 W. Kiessling, R.C. Baron-Szabo / Palaeogeography, Palaeoclimatology, Palaeoecology 214 (2004) 195–223 datasets at the K–T boundary. Sepkoski’s data were a homogeneous taxonomic concept and specimen- largely derived from the secondary literature and give based revisions more important than phylogenetic a comprehensive