Int. J. Plant Sci. 171(4):392–408. 2010. Ó 2010 by The University of Chicago. All rights reserved. 1058-5893/2010/17104-0005$15.00 DOI: 10.1086/651229 PHYTOGEOGRAPHIC HISTORY AND PHYLOGENY OF THE HUMIRIACEAE Fabiany Herrera,1,*,y Steven R. Manchester,* Carlos Jaramillo,y Bruce MacFadden,*,z and Silane A. da Silva-Caminhay *Department of Biology, Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611, U.S.A.; ySmithsonian Tropical Research Institute, Apartado Postal 0843-03092, Balboa, Ancon, Panama´, Repu´blica de Panama´; and zDivision of Research on Learning (EHR/DRL), National Science Foundation, 4201 Wilson Boulevard, Arlington, Virginia 22031, U.S.A. To place a new fossil occurrence of Sacoglottis in a broader context, we surveyed the fruit morphology of all extant genera of the Humiriaceae, conducted a cladistic analysis, and critically reviewed the fossil record for this family. Living and fossil fruits of Humiriaceae are recognized by a woody endocarp, germination valves, and, in some genera, wall cavities. The phylogenetic analysis based on 40 morphological characters yielded two most parsimonious trees indicating Vantanea as sister taxon to all genera among Humiriaceae. Schistostemon is indistinguishable from Sacoglottis in fruit morphology and is recovered as sister to Sacoglottis in the topology; we recommend restoring Schistostemon to the rank of subgenus within Sacoglottis. A review of prior published reports of fossil fruits attributed to Humiriaceae led to the rejection and/or reattribution of some records but supports recognition of Vantanea, Humiria, Humiriastrum, and Sacoglottis. The available characters do not support recognition of multiple fossil species of Sacoglottis. We recognize the occurrence of Sacoglottis tertiaria Berry emend. Herrera from Peru, Ecuador, Colombia, and a newly collected Miocene site from Panama. The Cenozoic fossil record of Humiriaceae in South and Central America, together with discreditation of former reports from Europe, strongly supports a Neotropical origin for this family. Keywords: fossils, Humiriaceae, phylogeny, morphology, Miocene, Panama. Online enhancements: appendixes. Introduction hiemstra et al. 2006; Pons and Franceschi 2007). The first en- docarps were recognized in the early 1920s and ’30s (Berry The Humiriaceae is a relatively small flowering plant family 1922a, 1922b; Reid 1933). Unfortunately, little was known of of the Malpighiales, with eight genera and ;50 species (Cua- the family, and comparative collections of extant species were trecasas 1961). It is distributed primarily in lowland to mon- very limited at that time. Many of those pioneering fossil iden- tane rain forests of the Neotropics but has a single species in tifications have been assumed to be systematically accurate rain forests of western Africa (fig. 1). The family has distinc- and therefore cited as indicators of ancient biomes (Burnham tive drupaceous fruits with woody endocarps having longitu- and Johnson 2004), but they remain in need of careful reeval- dinal germination valves and, in some genera, apical foramina uation. and endocarp wall cavities. The fruits are known to be con- We studied the fruit morphology of all extant genera of Hu- sumed and/or dispersed by rodents, tapirs, primates, birds, miriaceae with three objectives: (1) to seek new characters and bats and sometimes inhabited by beetle larvae (Cuatreca- that could allow the recognition of genera, (2) to discuss the sas 1961; Macedo and Prance 1978; Henry et al. 2000; John- characters useful for the identification of fossil remains, and son et al. 2001; Sabatier 2002; Lopes and Faria 2004; Ridgely (3) to consider the utility of the fruits in phylogenetic analyses. et al. 2005). Additionally, the extinct megafauna and Paleo- Finally, we describe a new record of Sacoglottis based on per- Indians of the Amazon rain forest apparently interacted with mineralized endocarps from ;19.5–17-Myr-old deposits from Humiriaceae (Roosevelt et al. 1996; Guimara˜es et al. 2008). the Cucaracha Formation of the Panama Canal and critically Although the family is clearly monophyletic, the precise review previously reported fossil fruits and pollen of the family. phylogenetic position of Humiriaceae relative to other Mal- pighiales is not clear yet (Bove 1997; Wurdack and Davis Material and Methods 2009). An early to middle Cretaceous origin has been inferred from molecular divergence estimates (Davis et al. 2005). The We examined collections from the U.S. National Herbarium oldest fossil record of Humiriaceae, however, dates back to (US) at the Smithsonian Institution in Washington, DC; the the Eocene (Berry 1924a, 1929a; this study). The fossil record Smithsonian Tropical Research Institute (STRI) in Panama; of the family has been recognized on the basis of pollen, the University of Florida Herbarium (FLAS) in Gainesville, wood, and endocarps (Berry 1922a; Lorente 1986; Hoog- Florida; the Missouri Botanical Garden Herbarium (MO); the U.S. National Arboretum Herbarium (BARC); the Natural 1 Author for correspondence; e-mail: fherrera@flmnh.ufl.edu. History Museum, London; and the Universidad Nacional Manuscript received August 2009; revised manuscript received January 2010. (UN) in Bogota´, Colombia. Approximately 100 specimens of 392 HERRERA ET AL.—PHYLOGENY OF THE HUMIRIACEAE 393 Fig. 1 Distribution of modern genera and location of fossil endocarps of the Humiriaceae. Base map courtesy of National Aeronautics and Space Administration Jet Propulsion Laboratory, California Institute of Technology. 46 species were measured and photographed (fig. 2; table A1 ferring to specimens deposited at the National Museum of in the online edition of the International Journal of Plant Sci- Natural History (USNM) in Washington, DC; the Florida ences). Transverse and longitudinal sections were cut through Museum of Natural History (UF) in Gainesville, Florida; the the center of the fruits or endocarps. The characters inspected University of Amsterdam (UA); and the Philipps-University are summarized in table 1. Marburg (UMBG), Germany. Newly recovered specimens of A new matrix, modified and expanded from that of Bove Sacoglottis tertiaria from Panama were deposited at STRI. (1997), was prepared for the phylogenetic analysis (fig. 3). Morphological data on extant Humiria pollen were ob- The character descriptions and scoring criteria are available tained from the extensive study of pollen morphology of Hu- in appendix B in the online edition of the International Jour- miriaceae by Bove and Melhem (2000) and the observation nal of Plant Sciences and online at the MorphoBank Web site of two species from the Graham Pollen Collection hosted at (http://www.morphobank.org; project P277, Matrix of Mor- STRI (Humiria balsamifera, slide 17520, Costa Rica, and phological Characters of Humiriaceae). Cladistic analyses Humiria guianensis, slides 5723 and 3011, Brazil). We used were performed using PAUP*, version 4.0 (Swofford 2003), a Nikon I80 camera with Nomarsky microscopy and a Nikon and the examination of character state distributions was com- DXM 1200 camera for the pollen observations. pleted using Mesquite, version 2.6 (Maddison and Maddison The newly recovered fossil endocarps reported in this study 2009). Our analysis was rooted with four outgroups, including (fig. 5) were collected from the Gaillard Cut section (Lirio the three used by Bove (1997), plus Caryocaraceae, in accor- East outcrop) of the southeastern part of the Panama Canal dance with the most recent phylogeny of the Malpighiales (lat. 9°3920N, long. 79°399400W). The fossils were found in (Wurdack and Davis 2009). To avoid assumptions regarding March of 2007 while exploring new temporary exposures cre- character state transitions, the 40 characters were treated as ated during the expansion of the canal. The plant locality is unordered. An exhaustive search with multistate taxa was car- placed in the lowermost part of the Cucaracha Formation. ried out. Morphological characters were evaluated through Lithofacies, sedimentary structures, fossils, and ichnofossils a review of Bove (1997), Bove and Melhem (2000), and refer- suggest that the ;100–140-m-thick Cucaracha Formation ences therein (e.g., Cuatrecasas 1961; Narayana and Rao was deposited in a succession grading from nearshore shallow 1977) and through direct observation of the fruits. Characters marine environments at the base to terrestrial facies in upper were coded as polymorphic when more than one state was levels. Kirby et al. (2008) reconstructed the formation as present, and characters not applicable for a taxon or simply a coastal delta plain consisting of abundant paleosols, chan- not known were coded as missing. Support levels for tree nel, levee, floodplain, marsh, and volcanic deposits. The Cu- nodes were assessed with the bootstrap procedure provided in caracha Formation overlies the Culebra Formation, also PAUP*, version 4.0, using a branch-and-bound search and exposed in the canal, which includes a distinct marine succes- 10,000 bootstrap replicates. sion ranging from neritic environments at the base (coral reef, Eleven of 16 fossil endocarp taxa previously assigned to costal lagoon) to delta and prodelta fronts at the top (Wood- Humiriaceae were physically reexamined (table 2; fig. 4), re- ring and Thompson 1949; Kirby et al. 2008; Moro´ n et al. 394 INTERNATIONAL JOURNAL OF PLANT SCIENCES Fig. 2 Extant fruits of Humiriaceae. A, B, Endocarps of Sacoglottis guianensis Benth (US 3335606); note wide valves and five