ORNITOLOGIA NEOTROPICAL 23: 13–27, 2012 © The Neotropical Ornithological Society

Diversification of the Neotropical Avifauna: disentangling the geographical patterns of persisting ancient taxa and phylogenetic expansions

Jon Fjeldså

Centre of Macroecology, Evolution and Climate, Zoological Museum, University of Copenhagen. Universitetsparken 15, DK-2100 Copenhagen, Denmark. E-mail: [email protected].

Resumen. – Diversificación de la avifauna neotropical: desentrañando los patrones geográficos de taxones persistentes antiguos y las expansiones filogenéticas – Con el desarrollo de bases de datos digitales de distribución de especies y de filogenias moleculares, ahora es posible analizar los procesos espacio-temporales de diversificación. Esto se puede hacer mediante el análisis de la distribución geográfica de especies evolutivamente próximas en los árboles filogenéticos, de los patrones de diversidad de clados antiguos y de especies de diferentes edades. La “conservación de nicho tropical” parece estar relacionado con el desequilibrio de la suma neta de los procesos de nacimiento y muerte, y con la mayor persistencia de clados con bajas tasas netas de especiación en las tierras bajas tropicales, en comparación con tasas más elevadas en otros lugares. Las distribuciones geográficas divergentes de los grupos que sufrieron ex- pansiones filogenéticas durante el Neógeno sugieren una fuerte capacidad de acceder a nuevas regiones ecológicas, en particular en el sur y en la región tropical de los Andes, donde los pulsos de especiación más intensos tienen lugar ahora en el límite de la vegetación arbórea. La especialización y densa compartimen- tación de los nichos podrían ser los procesos que explican las continuadas elevadas tasas de especiación en los Andes. ABSTRACT. – With the development of digital distribution databases and comprehensive molecular phy- logenies it is now possible to analyze diversification processes in time and space. This can be done by con- trasting geographical distributions of sister groups up through the phylogenetic hierarchy, patterns of ancient clade diversity and representing different ages. The “tropical niche conservatism” appears to be linked with the imbalance of the net sum of birth-death processes and the higher persistence of clades with low net speciation rates in the most productive tropical lowlands, compared with a higher turnover elsewhere. Divergent geographical distributions in groups that underwent phylogenetic expansion during the Neogene suggest a strong ability to access new ecoregions, notably in the south and in the tropical Andes region, where the most intensive speciation now takes place in the tree-line zone. It is suggested that specialization and dense niche packing forms the basis for continuous high speciation rates in the Andean “hotspot”. Keywords: Suboscine , species diversity, lineage diversity, root-path groups, tropical niche conserva- tism, speciation, endemism.

Biogeographic research has long been ham- cerned with the methods for tracing generalized pered by the lack of communication between patterns of vicariance and identifying underly- disciplines. While the macroecologists have set ing barriers. Although a significant proportion out to exlain the large-scale trends in species of the large-scale variation in species richness diversity from present-day ecological condi- is well explained from climate-based environ- tions, ignoring that every species has a history, mental models, this correlation is mainly caused the historical biogeographers were most con- by the many data points representing the most

13 Fjeldså widespread species (Rahbek et al. 2006). On the veloped at the Centre for Macroecology, Evo- other hand, the very aggregated distributions of lution and Climate (version July 2010). The range-restricted species represent high residual geographical range of each species has been values that are hard to explain from present- mapped at a resolution of 1º x 1º lat-long day environmental parameters and therefore quadrates, following the approach outlined require historical explanations (Jetz et al. 2004, by Rahbek & Graves (2001). Maps represent Rahbek et al. 2006, Hawkins et al. 2006). Over a conservative extent-of-occurrence for the the last two decades, the rapid development of breeding range based on museum specimens, digital distribution data and molecular phylo- published sight records and spatial distribu- genetic data has opened new perspectives for tion of habitats between documented records teasing apart the environmental and evolution- (>1000 references used for South America), ary causes of biogeographic patterns (Jetz et al. and has been validated by many experts. The 2004; Hawkins et al., 2005; Weir 2006). species level is updated according Interpretations of large-scale patterns of di- to Remsen et al. (2011). The software used versification have until recently been hampered (Worldmap) calculates various diversity mea- by lack of phylogenetic data. Thus, biogeographic sures, such as species richness in grid-cells, interpretations were often based on simple as- range centers, mean range-size and endemism sumptions, such as centers of species richness (either number of species representing the 1st representing centers of origin (dispersal centers; range-size quartile, viz., the 25% of all species Müller 1976). With the emergence of modern with smallest distributions, or the sum of in- phylogenies, this was replaced by area cladograms verse range-size values). The species distribu- and recently by more sophisticated ancestral area tion data can then be linked with tree-spanning reconstructions that take into account dispersal measures such as mean root path (sometimes and uncertainties in the phylogenetic reconstruc- used as evidence of tropical niche conserva- tions (e.g., Burns & Haricot 2009; Santos et al. tism, although biased towards species poor 2011). Such methods cannot remove the uncer- areas), based on a supertree for all birds (com- tainty about how much of the ancestral distribu- piled through review of most of the published tions that have been erased by extinctions and phylogenetic studies), and densely sampled range dynamics caused by past climatic perturba- molecular phylogenies for some groups. Spe- tions, such as the disappearance of the species-rich cial attention will given here to the endemic tropical-like forest of the Argentinean Patagonia suboscine radiations, because of the well re- until the end of the Eocene (Wilf et al. 2003). Be- solved and densely sampled phylogenies (no- cause of this uncertainty, I will illustrate here more tably Irestedt et al. 2009; Ohlson et al. 2009, general tendencies based on comparison of spatial in review; Moyle et al. 2008; Tello et al. 2008; diversity patterns for ancient and terminal lineages. Derryberry et al. 2011), and because this is This will first of all be done for the endemic radia- the largest endemic radiation that evolved in tion of suboscine birds. On this background I will splendid isolation through most of the Tertia- address more general questions about the relative ry, until the establishment of the Panama Isth- roles of extinction, speciation, niche conservatism mus allowed a limited amount of phylogenetic and phylogenetic expansion. expansion to the north. In addition, dated phylogenies can be used Methods to reconstruct patterns of clade diversity at particular points in time. We cannot know the Species distribution data are part of a com- past distribution of a clade, but assuming that prehensive global distributional database de- it diversified in situ, by vicariance, we may esti-

14 Diversification of the Neotropical Avifauna mate the ancestral distribution by merging the son with the total diversity and with range size ranges of all constituent species and remov- groups, from the 1st to 4th quartiles [25%] of ing only the peripheral areas (notably outside least to most widely distributed). This provides South America) that evidently represent dis- an indirect assessment of the ecological deter- persal in a terminal subclade. Time windows minants of the illustrated pattern, which is well for reconstructions of clade diversity maps documented for all endemic South American were defined from earth history data: Assum- birds and analyzed Rahbek et al. [2006] with ing that modern groups flourished during respect to environmental models that best ex- the Eocene climatic maximum (when species- plain the geographical patterns. rich tropical-like forests existed south to Pata- Recent diversification can be illustrated for gonia; Wilf et al. 2003) but suffered a serious groups with densely sampled phylogenies. Fjeldså set-back at the Eocene/Oligocene transition & Irestedt (2009) illustrated the spatial diversifi- (34 Mya) during the first Antarctic chill. New cation of Furnariidae using root-path quartiles as crown group radiations started again late Oli- a surrogate measure of age, but this can now be gocene up until the mid-Miocene optimum (15 supplemented with molecular age estimates of all Mya) (see Zachos et al. 2001). Past diversity terminal species taxa thanks to the very densely patterns were therefore assessed by mapping: sampled phylogeny by Derryberry et al. (2011). During the preparation of this manuscript, most 1 the diversity of small clades (of all birds; other phylogenies of South American birds have defined as lineages with 1-3 allospecies) been examined for general trends. that date back to the warm Eocene and 2 the clade diversity for suboscine groups in Results the late Oligocene (26 Mya), after the Oli- gocene chill and just before the onset of in- Patterns of large scale variation in avian di- tensive diversification in crown groups, and versity in South America is illustrated in Fig. 3 the clade diversity of suboscines in the 1. The phylogenetic relict species (Fig. 1a) are mid-Miocene (15 Mya). most strongly represented in the most exten- These geographical patterns can be com- sive areas of lowland rainforests and flood- pared with diversity patterns for all South plains, corresponding well to the diversity of American birds, as a general measure of repre- the most widespread quartile of South Ameri- sentativeness (Tab. 1; r values refer to compari- can bird species (Tab.1). The other maps are

Table 1. Comparison of geographical diversity patterns (r values) of the illustrated diversity patterns (Figs 1, 2 and 3) with patterns for all endemic South American birds, as presented in Rahbek et al. (2006) (1st quar- tile is the 25% of species with the smallest geographical ranges; 4th quartile is the 25% of most widespread species). Because of the high autocorrelation in such datasets, it was not considered relevant to present p values).

Fig1a b c d Fig.2a Fig.3ª b c d e All 0.82 0.91 0.98 0.24 0.94 0.87 0.82 0.84 0.15 0.28 1st 0.12 0.30 0.29 0.82 0.33 0.11 0.09 0.14 0.12 0.01 2nd 0.21 0.27 0.23 0.71 0.30 0.15 0.11 0.16 0.08 0.05 3rd 0.62 0.47 0.45 0.43 0.51 0.74 0.28 0.43 0.49 0.20 4th 0.88 0.90 0.98 0.31 0.94 0.74 0.79 0.82 0.33 0.29

15 Fjeldså

FIG. 1. General characteristics of diversity of South American birds. A, variation in distribution of 65 phy- logenetic relict avian species of Eocene age; b, diversity of 19 suboscine clades as of 26 million years in the late Oligocene; c, diversity of extant species; and d, of the 25% species with the most restricted distributions. Proportional colour scale, red colour representing the highest number of taxa (which is 180-285 species in map c). based on suboscine birds only. The highest lin- basin. However, some clades of Andean ori- eage diversity from the early phase of the di- gin appear from the Oligocene and early Mio- versification of suboscine crown groups (late cene (Rhinocryptidae, Chamaeza, Geositta; and Oligocene; Fig. 1b) is rather concentrated in Grallaria /Grallaricula), and several other ap- the southern Atlantic forests of Brazil, with pear towards the mid-Miocene. Some of these many small clades, and in the western Ama- are rather small clades (Ampelion group, Ampe- zon area up to the sub-Andean zone; many of lioides, Teledromas-Rhinocrypta-Acropternis, Pygar- the old lineages being disjunctly distributed in rhicas/Ochetorhynchus, Pseudocolaptes-Premnornis- both areas (Batalha-Filho et al. 2013). These Tarphonomus), suggesting a down-scaling of ancestral patterns are also evident in the map diversification rates, other clades maintained of terminal species taxa (Fig. 1c), with the high speciation rates throughout the upper quartile of most range-restricted species in Tertiary. the tropical Andes region (Fig. 1d). Very simi- Interestingly, by the mid-Miocene (15 Mya), lar species richness patterns are found when most suboscine clades can be classified as either comparing the richness patterns of Tyrannides based in the Andes region (or Andean-South versus Furnariides r 5 0.94), Suboscines ver- Brazilian uplands) or Amazonian/Guianan. sus Oscines (r 5 0.85), and Suboscines versus Fig. 2 presents diversity of clades at 15 Mya, hummingbirds, Trochilidae (r 5 0.89; Rahbek with separate colours for the two geographical & Graves 2001). The geographical patterns for origins in Fig. 2b (this graph excludes, though, range-restricted suboscines (Fig. 1d) is closely some clades of other, or uncertain, geographi- similar to that for all range-restricted birds cal origin; see legend). It is evident from this (Tab. 1). illustration that the maximum diversity (white Basal suboscine clades are widely distribut- cells, or red cells in Figs 1 b and c) are in grid- ed in the mesic tropics, notably in the ancient cells located along the borderline between the cratonic areas north and south of the Amazon Andean and Amazonas faunas, and that there

16 Diversification of the Neotropical Avifauna

FIG. 2. Diversity of suboscine clades at 15 Mya. To the right, these have been split up into 35 clades rooted in the Andes region (green), and 36 clades rooted in tropical lowlands (purple) (ρ² 0.416). Note that mixing of these groups produce grey hues, and overall lineage diversity is expressed as brightness, with the most diverse areas appearing whie. (This comparison excludes the south Brazilean Calyptura and Carpornis and five larger groups whose biogeographic origin is difficult to place although it is most probably rooted in a broadly defined southern savanna region, namely the crown group of spinetails, thornbirds and elaenine, tyrannine and fluvicoline flycatchers). has been very limited geographical mixing of increasingly prominent towards the Pliocene. these faunas. Among the Amazonian clades, An apparently high diversity along the lower some species have apparently dispersed north Amazonian channel east of the Purus Arc (Fig. of the Colombian Andes or crossed the Andes 3 b-d) marks the in-grid-cell overlap of species to the Chocó region, but only very few have de- inhabiting the northern and southern banks veloped a decided tolerance for montane condi- (thus this does not reflect local endemism, as tions. Similarly, only few terminal species in An- no range-restricted species are centered in this dean clades colonized the Amazonian lowlands. area). Also the local peaks at Río Sao Francisco The densely sampled and time calibrated in eastern Brazil and along the western edge of phylogeny for Furnariidae (Derryberry et al. the Brazilian shield (Fig. 3d) represents points 2011) is used in Fig. 2 to illustrate the historical of marginal overlap between regional faunas. progress of speciation events. The oldest spe- Recent diversification took place throughout cies, representing speciation events around the the southern savannas (Fig. 3d) and in the An- mid-Miocene, are mainly found in the Amazon des (Fig. 3e). The high levels of Pleistocene basin and along its ancient drainage channel to speciation along the humid eastern edge of the the north, and in the Chocó and to some ex- highest parts of the Andes (Fig. 3 d and e). tent in the Guianas and southern Atlantic for- Areas with no new species originating during est (Fig. 3a). The diversification then continued the upper Pleistocene (white in Fig. 3e) cor- in the Guiana highland and in the upper (So- respond to areas of highest median range-size. limões/Pebas) Amazon basin, with a concen- The Tyrannida follow a similar pattern, tration in the sub-Andean forelands becoming with basal radiations (Cotingidae, Pipridae,

17 Fjeldså

FIG. 3. Diversification of South American furnariids, illustrated as spatial variation in richness of species in groups based on time to the most recent common ancestor, following molecular phylogenies of Derryberry et al. (2011) and Ohlson et al. (2013).A, 45 species lineages with origin 20-7.5 million years ago; B, 34 species age 7.5-5 my; C, 79 species age 2.5-5 my; D, 86 species age 2.5-1.5 my; E, 48 species age ,1.5 my.

Tityridae, Rhynchocyclidae and various small a fairly general pattern of variation in terrestrial old clades) mainly in the mesic tropical low- biodiversity in South America, at this scale of lands, and elaenines, tyrannines and fluvico- spatial resolution. The diversity pattern for the lines expanding out in the open, first in the most ancient avian species (Fig. 1a) resembles southern tropical woodland savannas, but con- the pattern for the most widespread birds (Tab. tinuing in the southern cone of the continent 1), corresponding to extensive mesic lowland (Xolmiini) and in the Andes (Fluvicolini and habitats and well explained from net primary Xolmiini), and with phylogenetic expansions productivity (Rahbek et al. 2006 fig. 3). Most of out of South America in Tyrannini and Conto- the ancient, small clades are relatively large birds pini (Tello et al. 2009; Ohlson et al. 2008 and that may have required extensive areas to persist in review). The tapaculos diversified along the over such a long time, which may explain their entire Andes (with gaps in the arid highlands), general absence inside the montane regions. with three colonizations to S Brazil and one The mid-Tertiary lineage diversity pattern old colonization to the Amazon lowlands. (Figs 1b and 2a) appears to be reflected in the current species diversity pattern (r 5 0.91 and Discussion 0.94; Tab. 1). Using the same distributional and phylogenetic data as this study, Sanín et al. Imbalanced distribution of old and young taxa. The (in review) found that 81% of the variation in illustrated species richness pattern for the en- species species diversity at a spatial resolution demic suboscine radiation (Fig. 1c) resembles of 10x10º could be explained from the lineage that for groups that colonized in recent geo- diversity as far back as 40 mya. This would logical times (tanagers; Fjeldså & Rahbek 2006), seem to suggest a strong role of regional varia- and for other taxonomic groups, such as mam- tion in speciation rates. This may notably be mals (Tognelli & Kelt 2004) and plants (Barth- the case near the geologically dynamic western lott et al. 2005). Thus, this appears to represent edge of the continent. However, part of the

18 Diversification of the Neotropical Avifauna explanation may also be range-dynamics: A situ, mainly in the lower montane forest (e.g., high loss of diversity in the south as a con- Laniisoma, Snowornis, Thamnistes, Premnoplex, sequence of major climate perturbations, and Margarornis, Hellmayrea and Pseudocolaptes, and a higher persistence of lineages in the areas Aphrasture in the south), with little or no net that remained warm and humid. The disap- diversification, at least not during the Pleisto- pearance of the diverse tropical-like forests cene. of Patagonia during the abrupt global cooling As illustrated in Fig. 3, the geographical (Terminal Eocene Event, 34 Ma; Wilff et al. patterns of diversification changed in time. 2004; Zachos et al. 2001) must have put an end The lowland rainforest clades diversified to many warm-adapted groups. The Tyran- mainly during the Miocene, with some shifts nides phylogeny (Ohlson et al. 2009, 2013) has over time, but apparently there was a down- a long basal branch, a single species surviving shift in the radiation of rain-forest groups, and up until 32 Ma. This ancestor was probably a these groups were only moderately elevation rainforest frugivore (judging from the ecol- tolerant and successful in colonizing the rising ogy of the basal tyrannid groups of manakins, tropical Andes region. The philydorine cotingas and some small clades). Frugivory is Thripadectes, which radiated in the Andes since also widespread (as a diet supplement) in other 8-9 Mya (Derryberry et al. 2011), is the most tyrannid groups. Such frugivorous birds may noteworthy exception. World wide, it seems have become brutally restricted by the global that speciation in tropical lowland biomes cooling at the Terminal Eocene Event, and the ceased during the last million years of great early radiation of the Tyrannides crown group climatic instability (Päckert et al. 2011 for Asia; therefore took place near the equator. As the Fjeldså et al.2007 for Africa). old groups were thus shuffled together in the In contrast, the Andean clades continued remaining tropical forest regions, the continu- to diversify, and various synallaxine groups ing diversification in the south was constrained (as well as the Tyrannidae flycatchers, sensu by the development of a large rain shadow Tello et al. 2009 and Ohlson et al. 2013) di- area that developed as a consequence of the versified strongly during the upper Miocene Andean uplift. in the mosaic habitats that developed with the The asymmetries of many phylogeneties spread of savannah vegetation and grasslands suggests that early bursts of diversification in the southern subtropical zone (Zachos et al. were followed by apparent down-shifts, lead- 2001). This shift of the frontier of diversifica- ing to the emergence of many small clades tion to the south is reflected in the low r values (Ricklefs 2005), which mainly survived in the in the right column in Tab. 1. Apparently, the southern part of the Brazilian Atlantic forest southern savanna region was a center diversi- and in the sub-Andean Amazon Basin (Fig. fication and expansion to riparian habitats and 1b). The high persistence of old lineages in edge habitats in the Amazon (Cranioleuca, Syn- this region has given rise to the idea of tropi- allaxis) as well as to the central parts of the cal niche conservatism as a historical narrative Andes. There has been a significant exchange for explaining such broad species richness gra- between Andean-Pampean region and the dients (Fig. 1 a, b; Hawkins et al. 2005; Weir South Brazilian highland forest. Interestingly, 2006). However, the distinct pattern in Fig. 2b as the involved groups are strongly associated suggests that also the Andean clades may also with rugged terrain (or well-drained soils), this be constrained by niche conservatism. The Andean-Brazilian track was broken by the de- Andes contains an interesting mix of radiating velopment of hydrologically unstable plains in groups and small/old clades that persisted in the Beni-Pantanal region (Fig. 2b). During the

19 Fjeldså climatically most unstable upper Pleistocene groups may have their origin in the southern the diversification continued only in some Atlantic forest of Brazil but expanded along the clades and mostly at 13-18º south in Peru and western edge of the Brazilian cratonic shield Bolivia, and mainly at the upper tree-line and into the western Amazon basin (Batalho-Filho upper montane valleys (Fjeldså & Irestedt et al. 2013). The tropical lowland groups invad- 2009). ed the Andean habitats only marginally (Thri- In the New World flycatcher group, phy- padectes); yet some clades of the mesic tropics logenetic expansions outside the topical low- colonized the Chocó region by circumventing lands, and in open habitats, happened only the Colombian cordilleras, or by crossing the from the mid-Miocene, and only in one ma- low passes of southern Colombia or northern jor clade (Tyrannidae sensu stricto; Ohlson et Peru (Haffer 1974). al. 2009), with different subclades moving The comparison of Andean and Amazo- into distinct areas, thus the Xolmini radiat- nian clades that diverged in the mid-Miocene, ing mainly in southern open habitats and the well before the final uplift of the northern An- barren Andean puna, while the Ochthoecini des, is a novel demonstration of long histori- radiated in the montane cloud-forest zone cal isolation of clades in the Andean orocline (García-Moreno et al. 1998) and the Contopini and specialization for this misty environment. dispersed to North America. The extraordinary biodiversity at the Andean- Oscine groups arriving from the north (Weir Amazonian interface (Fig. 1) emerges as an ef- et al. 2009) diversified first of all in mountain fect of overlap (at the spatial resolution of this ridges of Central America and Colombia (Fjeld- grid) of faunas with different histories. så & Rahbek 2006; Klicka et al. 2007; Sedano To understand the historical separation & Burns 2009; Lovette et al. 2010) and after a of Andean and Amazonian clades we need to rapid initial burst of speciations in this region look into the geological history. The Andean several lineages of finches and tanagars un- habitats are often described as young environ- derwent phylogenetic expansion further south ments, but the orocline as such has a long his- in the Andes and in the tropical and southern tory (Garzione et al. 2008; Mamani et al. 2010). subtropical lowlands (like in amphibians; see As a consequence of the convergence of Santos et al. 2011). Other colonizing groups South America with the oceanic Nazca Plate, arrived through trans-oceanic sweepstakes dis- marine terranes were accreted to the north- persal (vireos, thrushes, mockingbirds, eupho- western margin of the continent (Ecuador to nias), radiated mainly in the lowlands, but over Venezuela) and a proto-Andean chain devel- all with significant flexibility in their life zone oped along the entire western plate boundary. associations. Although oscine and suboscine Here we need to acknowledge that even small groups have markedly different biogeographic mountains near tropical coasts will be distinct histories, the coarse-scale diversity patterns are from the surrounding lowlands in terms of remarkably similar. climate and habitat structure with stunted, microphyllic and epiphyte-laden cloud forest Linking diversity patterns and earth history. The most down below 1000 m elevation (Foster 2001; basal members of the extant suboscine groups Bruijnzeel et al. 2010). This may have allowed inhabit ancient cratonic areas (Guiana and Bra- physiological specialization for these cool/wet zil Shields) and some parts of the southern An- habitats and development of a distinct high- des and there is significant connectance in the land avifauna from already from an early stage. east between the Guianan and Brazilian shield Increasing uplift of mountain chains lead to faunas. Apparently some of the old rainforest formation of several thrusting planes and thick-

20 Diversification of the Neotropical Avifauna ening of the continental crust, and this load Opportunities for diversification in the caused a subsidence of the lowlands to the east, Amazon Basin may have shifted much in first with formation of a narrow axial grove, the course of the conversion from the wide- whereby the water from the Andes could flow northwest-flowing flood basin to the current north to the Caribbean (Maracaibo) or south to eastward-running Amazon basin since the late the Paraná. Thus, the early fauna of the tropical Miocene (Hoorn & Wesselingh 2010) or even Andes (Fig. 2b) may have been more or less iso- as late as the Pliocene (Latrubesse et al. 2010). lated from that of the Brazil-Guiana area. In the early Tertiary, this region was charac- During the mid-Tertiary the Bolivian oro- terized by a sub-Andean grove draining north cline underwent significant crustal thickening to Maracaibo (Venezuela), but gradually shift- and bending (Mamani et al. 2010). The contact ing east to form large tectonically controlled (“Chapare Buttress”) between the thrust front sedimentation basins (Solomões/Pebas) over of the Bolivian Andes and the subsurface edge large parts of western Amazonia. Some dis- of the Brazilian Shield formed a structural di- agreement exists over the nature and timing vide between the Amazonas and Paraná sys- of these wetlands, but apparently there were tems and faunal exchange between the Andes enormous freshwater swamp habitats and ma- region and uplands of southern Brazil, which rine incursions (Lundberg et al. 1998; Hoorn explains why many clades are shared between & Wesselingh 2010) and in the late Miocene the Andes region and the southern Brazilian some tectonic deformation of foreland ridg- uplands (Fig. 2b). As the weight of the Boliv- es (Divisor-Contamana ranges) in northern ian Andes continued to rise there was a large Peru are reflected in Figs 3 b and c. We must geological subsidence, leading to the formation therefore assume that the Amazonian avifauna of the present Beni plains (Hanagarth 1993, persisted by moving around, taking advantage Silva 1994), and today only small fragments of the patch dynamics and high productiv- are left in southeastern Bolivia of the former ity of the floodplains but generally within the high land surface, and the divide between the same region, until the wetlands were filled up Amazon and Paraná basins shifted south (Lun- by megafans of sediment and the drainage dberg et al. 1998: 42). The new, hydrologically towards the Atlantic Ocean was finally es- unstable environments in the Beni-Chaco re- tablished. Thus, the diverse Amazonian and gion became an insurmountable barrier for the Andean faunas could finally merge, leading to birds adapted to rugged and well-drained land- the incredible present species diversity in the scapes, leading to the divergence from the late transition zone (Fig. 2b). The establishment of Miocene to early Pleistocene of Andean and new rivers channels finally lead to distinct vi- South Brazilian sister taxa (see Fig. 2b). cariant patterns and raised species richness in Within the tropical Andes, the most recent grid-cells that include para-species occupying diversification took place in the young eastern opposite river banks (see Figs 3 c and d). Cordillera of Colombia-Venezuela (Fig. 3e) Diversification is tightly linked with geology and Bolivia (Fig. 3d and e). In the southern and landscape evolution in the upper Amazone parts, this corresponds to where jagged con- basin up through the Miocene and Pliocene, figuration of the eastern Andean ridges form and in general there has been a slow-down in climatic pockets that are well protected against diversification after the present drainage pattern the ecological disturbance by south polar was established. In general, the bird populations winds, notably during glacial periods (Fjeldså inhabiting the Brazilian and Guinanan shields et al. 1999). This lead to an extraordinary peak are widespread and have undergone little diver- of specie s diversity 12-18ºS of the equator. sification in recent geological times.

21 Fjeldså

Speciation mechanisms. Low mean root-path dis- Most speciation took place within a narrow tance in the tropical lowland biota has been climate window, with sister species gener- taken as evidence of the tropical niche con- ally replacing each other at similar elevations servation hypothesis, as opposed to assumed on different slopes (Cadena et al. 2011). This innovative adaptations characterizing the more suggests niche conservatism in the therminal terminal radiations in the Andes (Hawkins et niches, while segregation in different eleva- al. 2006; Fjeldså & Irestedt 2009; Löwenberg- tional zones (or in different structural habi- Neto et al. 2011). The high diversity in the tats) on the same slopes were often delayed lowlands is mainly associated with the enor- by several million years (García-Moreno & mous extent and complex geological history Fjeldså 1997). There are few cases of marked of the Amazon basin, but the lack of recent elevational shifts (e.g., Cinclodes, Leptasthenura, speciation in the Amazon lowlands suggests Cranioleuca). Thus, the key to understand inten- that Pleistocene forest refuges (Haffer 1974), sive radiation may be related to conditions that if they existed, were not drivers of specia- allow remnant populations to persist in con- tion. During the upper Pleistocene there was stantly favorable local environments within the intensive speciation (among furnariids) only montane habitat mosaic. This is rarely possible in the western Pampean region and in the An- outside the tropics. Speciation has traditionally des (Fig. 3e). The recent radiations are closely been explained physical barriers between areas, associated with the upper montane forest but many replacements appear to be second- and tree-line zone (Fjeldså & Irestedt 2009), ary contact zones (in places with no obvious mainly in clades that originated south of the physical barriers) and it is important to note Amazon area (a similar situation has now been the importance of factors that promote local described for the upper montane forest zone persistence during periods of environmental in the Sino-Himalayan mountains, where lin- stress, when many species survive only as local eages of northern temperate origin diversified relict populations. The most prominent local intensively during the Pleistocene; see Päckert aggregation of restricted range species in the et al. 2011). This represents a habitat band that southern half of the tropical Andean hotspot may be easily fragmented because of the nar- are in valley systems that are protected by sharp row configuration of this habitat band and bends on the eastern Andean ridges against the steep terrain with many physical barriers the impacts of south polar winds, which rep- (Graves 1988). In addition to this comes the resent a disturbing agent in the Pleistocene complex ways in which the topography creates climate (Fjeldså et al. 1999). The most inten- a mosaic of local climates,with local pockets sive recent radiation took place in resident spe- of very predictable conditions. Areas with cies of dense thickets and elfin forest patches intensive recent speciation in the Andes cor- (Scytalopus, Cranioleuca, Schizoeaca), where the responds closely with areas with many small amount of gene flow may be limited (cf. Kisel distributions (Fig. 1d), in contrast to the lack & Barraclough 2010), but interestingly also the of recent speciation in areas with high mean widespread Cinclodes and Muscisaxicola species range-size on the tropical savanna regions and of the barren alpine habitats maintain rapid much of the Amazon/Guiana region (Fig. 3e). speciation. Apparently, few groups were well adapted Because of the restricted distributions of to cope with highland condition and to radiate many Andean endemics, each species can ac- in this harsh environment, leading to phylo- cumulate local adaptations that allow them genetically clustered communities in the high- to become more abundant, whereas in more lands (Graham et al. 2009; Parra et al. 2010). widespread species the gene flow across sub-

22 Diversification of the Neotropical Avifauna optimal environmental gradients and over be less relevant in large and complex montane long distance may restrict local adaptation and regions, where conditions may vary over very thus the local abundance. Species must either short distances in the complex habitat mosaic be sufficiently generalist to utilize a broad (see Scherrer & Körner 2011). Global climate habitat mosaic, including suboptimal sites, or models based on interpolation between widely they must survive in situ, by tracking specific scattered weather stations cannot capture the habitat zones where they can maintain dense complex local variation within the habitat mo- populations and remain resilient (Williams et saics of large mountain regions, where pockets al. 2009; Reif et al. 2010). The abundance of of extremely stable conditions may exist locally many range-restricted species, which should because of interactions between wind systems, be well known among Andean ornithologists atmospheric stratification and topography. (although not documented quantitatively for these difficult habitats) is contrary to gen- Are communities saturated? The terminal ra- eral theory of range-abundance relationships diations of the super-diverse clades of hum- (Brown 1984; Gaston et al. 1997). Analyzing mingbirds, tyrannids, furnariids and tanagers the mutualistic networks of plants and hum- all took place in areas characterized by low mingbirds at 31 study sites spanning a wide climate change velocity within the tropical range of climate regimes, Dalsgaard et al. Andes region. Some subclades (coquette and (2011) found that the highest degree of mu- brilliant hummingbirds; Cranioleuca, Asthenes/ tualistic specialization in montane forests in Schizoeaca, Scytalopus and mountain tanagers) Costa Rica, Colombia and coastal mountains are really outstanding, considering the short in southern Brazil. Across all 31 study sites, time span and the limited geographical space the degree of mutualistic connectedness was within which they have radiated. With the in- well explained from “climate change velocity” tense speciation in this region we may wonder (Loarie et al. 2009), which, based on global whether diversification can continue. climate models, describes the rate at which Several studies have demonstrated declin- species must migrate to keep track of climate ing diversification rates over time, suggesting a change. Comparing present-day conditions saturation as the ecological niche space is filled with a global climate model for the last gla- up (e,g,, Phillimore & Price 2008; Rabosky cial maximum, the lowest levels of “climate 2009; Rabosky & Lovette 2008; Sedano & change velocity” were found in tropical mon- Burns 2010). McPeek (2008) considered that tane regions, and explains well the global varia- slow-down may be a general trend, and Rabos- tion in occurrence of local endemic species ki (2010) went on to suggest ecological limits of amphibians, mammals and birds (Sandel et as an alternative to speciation and extinction in al. 2011). Thus, high local endemism can be explaining diversity patterns. Against this view, explained from conditions that allow species Morton et al. 2010 and Wiens (2011) argues to remain in the same geographical area, and that biological diversity continues to grow, and specialize to local biotic and abiotic conditions, indeed the great speciator groups of South in spite of the instability of the global climate. America (core tyrannids, furnariides and tana- It has been predicted that the current global gers; Barker 2011) could be good examples. In change may drive montane species out of the all groups, the most intensive diversification local elevational gradient, leading to large-scale takes place within a rather limited geographical extinction of montane birds (Sekercioglu et al. area in the tropical Andes region. The analysis 2008, La Sorte & Jetz 2010). This might well ap- of furnariid disversification (Derryberry et al. ply to isolated, cone-shaped mountains, but may 2011) suggests a continuously high rate of spe-

23 Fjeldså ciation as the diversification has shifted from Martin Irestedt are thanked for many years of the Amazon lowlands to other ecoregions with phylogenetic work, and Louis Hansen for da- a very high landscape complexity. The most tabase management. The Danish National Re- recent phylogenetic expansion in this group, search Foundation is acknowledged for their as well as of crown group tyrannids, moves support to the Center for Macroecology, Evo- northwards in the tropical Andes region, into lution and Climate. the areas with maximum diversity of nine-pri- maried oscines. Although these groups often References feed together in mixed feeding parties, they differ markedly in feeding methods and micro- Barker F. K. 2011. Phylogeny and diversification of habitats, and may therefore not interfere much modern . Pp 235–256 in G Dyke & G in their use of food resources. Other examples Kaiser (eds) Living Dinosaurs, The Evolutionary of continuously high diversification rates are History of Modern Birds. Wiley-Blackwell, emerging (e.g., Fritz et al. 2011) for groups that Chichester, U.K. have undergone large geographical expansions Barthlott, W., Mutke, J., Rafiqpoor M. D., Kier G., into new geographical areas. Kreft H. 2005. Global centres of vascular plant It is more questionable what happens in diversity. Nova Acta Leopoldina 92: 61–83. ecoregions that are structurally less complex Batalha-Filho H., Fjeldså J., Fabre P.-H. & Miyaki and where the species are widespread, move in C. Y. (2013) Connections between the Atlantic response to climatic harshness and instability and the Amazonian forests avifaunas represent and thus are under shifting selective pressures. distinct historical events. J.Orn.154: 41-50. Here, the general range-abundance relationships Bates J. M., Cracraft J. & Hackett S. J. 1998. Area- (Brown 1984) could well apply, and the generalist relationships in the Neotropical lowlands: species may squeeze out the local specialists. An hypothesis based on raw distributions of Further studies of diversification rates need birds. J. Biogeogr. 25: 783–793. to be comprehensive, covering large taxonomic Brown J. H. 1984. On the relationship between assemblages at continent-wide or global scale, abundance and distribution of species. Am Nat and should include data on community struc- 124: 255–279. ture, phylogenetic clustering as well as niche Bruijnzeel L. A., Scatena F. N. & Hamilton L. S. 2010. segregation. This next generation of studies Tropical Montane Cloud Forests. Science for can start now, as large distributional databases Conservation and Management. Cambridge U.P., and large phylogenies become available, and as Cambridge. we obtain more ecological data: we now need Burns K. J. & Racicot R. A. 2009. Molecular to look at networks and community structure, phylogenetics of a clade of lowland tanagers. analyzing packing of niches according to mor- Implications for avian participation in the Great phological segregation (as done already for fur- American Interchange. Auk 126: 635–648. nariids; Derryberry et al. 2011) and resource Cadena C. D., Kozak K. H., Gómez J. P., Parra J. L., use, and relatedness within local communities, McCain C. M., Bowie R. C. K., Carnaval A. C., and also on large geographical scales. Moritz C., Rahbek C., Roberts T. E., Sanders N. J., Schneider C. J., VanDerWal J., Zamudio K. Acknowledgements R. & Graham C. H. 2011. Latitude, elevational climatic zonation and speciation in New World First of all, I will thank the organizing com- vertebrates. Proc. R. Soc. B. 279: 194-201. mittee of the IX Neotropical Ornithological Dalsgaard B., Magård E., Fjeldså J., Gonzáles A. Congress for inviting me. Jan Ohlson and M. M., Rahbek C., Olesen J. M., Ollerton J.,

24 Diversification of the Neotropical Avifauna

Alarcón R., Araujo A. C., Cotton P. A., Lara 2011. Diversification and biogeographic patterns C., Machado C. G., Sazima I., Sazima M., Tim- in four island radiations of passerine birds. mermann A., Watts S., Sandel B., Sutherland Evolution 66: 179-190 W. J. & Svenning J.-C. 2011. Specialization in Garzione C. N., Hoke G. D., Libarkin J. C., Withers plant-hummingbird networks associated with S., MacFadden B., Eiler J., Ghosh P. & Mulch A. Quaternary climate-change velocity. PloS One 2008. Rise of the Andes. Science 320: 1304–1307. 6, e25891. Garcia-Moreno J., Arctander P. & Fjeldså J. 1998. Pre- Derryberry E. P., Claramunt S., Derryberry G., Pliostecene differentiation among chat-tyrants. Chesser R. T., Cracraft J., Aleixo A., Pérez-Emán Condor 100: 629–640. J., Remsen J. V. & Brumfield R. T. 2011. Lineage Gaston K J., Blackburn T. M. & Lawton J. H. 1997. diversification and morphological evolution in a Interspecific abundance-range size relationships: large-scale continental radiation: the Neotropical An appraisal of mechanisms. J. Anim. Ecol. 66: ovenbirds and woodcreepers (Aves: Furnariidae). 579–601. Evolution, 65: 2973-2985 Graham C. H., Parra J. L., Rahbek C. & McGuire Fjeldså J., Lambin E. & Mertens B. 1999. Correla- J. A. 2009. Phylogenetic structure in tropical tion between endemism and local ecoclimatic hummingbird communities. Proc. Natl. Acad. stability documented by comparing Andean Sci. USA 106: 19673–19678. bird distributions and remotely sensed land sur- Haffer J. 1974. Avian speciation in tropical South face data. Ecography 22: 63–78. America. Publ. Nuttall Orn. Club 14: 1–390. Fjeldså, J., Irestedt, M. & Ericson, P. 2005. Molecular Hanagarth W. 1993. Acerca de la geoecologia de data reveal some major adaptational shifts in the las sabanas del Beni en el noreste de Bolivia. early evolution of the most diverse avian family, Instituto de Ecología, La Paz. the Furnariidae. J. Orn. 146: 1–13. Hawkins B. A., Diniz-Filho J. A. F. & Soeller Fjeldså J. & Irestedt M. 2009. Diversification of the S. A. 2005. Water links the historical and South American avifauna: patterns and implica- contemporary components of the Australian tions for conservation in the Andes. Ann. Miss. bird diversity gradient. J. Biogeogr. 32: 1035– Bot. Garden 96: 398–409. 1042. Fjeldså J., Irestedt M., Jønsson K A., Ohlson J. I. Hawkins B. A., Diniz-Filho J. A. F., Jaramillo C. A. & & P. G. P. Ericson 2007. Phylogeny of the ov- Soeller S. A. 2006. Post-Eocene climate change, enbird genus : A case of independent niche conservatism, and the latitudinal diversity adaptations for terrestrial life. Zool. Scripta 36: gradient of New World birds. J. Biogeogr. 33: 133–141. 770–780. Fjeldså J., Johansson U., Lokugalappatti L. G. S. & Hoorn C. & Wesselingh, F. P. eds; 2010. Amazonia: Bowie RCK 2007. Diversification of African Landscape and Species Evolution. A Look into greenbuls in space and time: linking ecological the Past. Wiley Blackwell. 482 pp. and historical processes. J. Orn. 148 (S2): 359- Irestedt M., Fjeldså J., Dalén L. & Ericson P. G. P. 367. 2009. Convergent evolution, habitat shifts and Fjeldså J. & Rahbek C. 2006. Diversification of tan- variable diversification rates in the - agers, a species rich bird group, from lowlands woodcreeper family (Furnariidae). BMC to montane regions of South America. Integr. Evolutionary Biology 9: 268. Comp. Biol. 46: 72–81. Jetz W., Rahbek C. & Colwell R. K. 2004. The Foster P. 2001. The potential negative impacts of coincidence of rarity and richness and the global climate change on tropical montane cloud potential signature of history in centres of forest. Earth-Science Reviews 55: 579–582. endemism. Ecol. Letters 7: 1180–1191. Fritz S. A., Jønsson K. A., Fjeldså J. & Rahbek C. Kisel Y. & Barraclough T. G. 2010. Speciation has a

25 Fjeldså

spatial scale that depends on levels of gene flow. and phylogenetic classification of the antbirds, Amer. Nat. 175: 316–334. ovenbirds, woodcreepers and allies (Aves: Pas- Klicka J., Burns K. & Spellman G. M. 2007. Defining seriformes: infraorder Furnariides). Cladistics a monophyletic Cardinalini: a molecular 25: 386–405. perspective. Mole. Phyl. Evol. 45: 1014-1032. Müller P. 1976. The Dispersal Centres of Terrestrial La Sorte F. A. & Jetz W. 2010. Projected range con- Vertebrates in the Neotropical Realm. Dr. W. tractions of montane biodiversity under global Junk, The Hague. warming. Proc. Royal Soc. B doi:10.1098/ Ohlson J. I., Fjeldså J. & Ericson P. G. P. 2008. Tyrant rspb.2010.0612. flycatchers coming out in the open: phylogeny Latrubesse E., Cozzual M., Rigsby C., Silva S,. Absy and ecological radiation of Tyrannidae (Aves: M. L. & Jaramillo C. 2010. The Late Miocene Passeriformes). Zool. Scripta 37: 315–335. paleogeography of the Amazon basin and the Ohlson J. I., Irestedt M., Ericson P. G. P. & Fjeldså evolution of the Amazon River. Earth Science J. 2013. Phylogeny and classification of the Reviews 99: 99–124. New World Suboscines (Aves, Passeriformes). Lovette I. J., Pérez-Emán J. L., Sullivan J. P., Banks Zootaxa. R. C., Fiorentino I et al. 2010. A comprehensive Päckert M., Martens J., Sun Y.-H., Severinghaus multilocus phylogeny for the wood-warblers L. L., Nazarenko A. A., Ting J., Töpfer T. & and a revised classification of the Parulidae Tietze D. T. 2011. Horizontal and elevational (Aves). Mol. Phyl. Evol. 57: 753-770. phylogeographic patterns of Himalayan Löwenberg-Neto P., de Carvalho C. J. B. & Hawk- and Southeast Asian forest patterines (Aves: ins B. A. 2011. Tropical niche conservatism as a Passeriformes). J. Biogeogr. 39: 556-573. historical narrative hypothesis for the Neotrop- Parra J. L., McGuire J. A. & Graham C. H. 2010. ics: a case study using the fly family Muscidae. J. Incorporating clade identity in analyses of Biogeogr. 38: 1936–1947. phylogenetic community structure: an example Lundberg J. G., Marshall L. G., Guerrero J., Horton with hummingbirds. Amer. Natur. 176: 573–587. B., Malabarba M. C. S. L. & Wesselingh 1998. Phillimore A. B. & Price T. D. 2008. Density- The stage for Neotropical fish diversification: dependent cladogenesis in birds. PLoS Biol. 6: a history of tropical South American rivers. In e71. Reis RE, Vari RP, Lucena ZM & Lucena CAS Rabosky D. L., Donnellan S. C., Talaba A. L. & (eds) Phylogeny and Classification of Neotropi- Lovette I. J. 2007. Exceptional among-lineage cal Fishes. Porto Alegre, Edipucrs: 13-48. variation in diversification rates during the ra- Mamani M., Wörner G. & Sempere T. 2010. Geo- diation of Australia’s most diverse vertebrate chemical variations in igneous rocks of the clade. Proc. Roy. Soc. B: Biol. 274: 2915–2923. Central Andean Orocline (13º to 18ºS): Trac- Rabosky D. L. 2009. Ecological limits and diversifi- ing crustal thickening and magma generation cation rate: alternative paradigms to explain the through time and space. Geol. Soc. Am. Bull. variation in species richness among clades and 122: 162–182. regions. Ecology Letters 12: 735–743. McPeek M. A. 2008. The ecological dynamics of Rabosky D. L. 2009. Ecological limits and diversifi- clade diversification and community assembly. cation rate: alternative paradigms to explain the Am. Nat. 172: E270–E284. variation in species richness among clades and Morton H., Potts M. D. & Plotkin J. B. 2010. Infer- regions. Ecology Letters 12: 735–743. ring the dynamics of diversification: a coales- Rabosky D. L. 2010. Primary controls on species cent approach. PLoS Biology 8(9) e1000493. richness in higher taxa. Systematic Biology 59: Moyle R. G., Chesser R. T., Brumfield R. T., Tello J. 634–645. G., Marchese D. J. & Cracraft J. 2009. Phylogeny Rahbek C. & Graves G. R. 2001. Multiscale assess-

26 Diversification of the Neotropical Avifauna

ment of patterns of avian species richness. shifts and bird extinctions. Conservation Biol- Proc. Natl. Acad. Sci. U.S.A. 98: 4534–4539. ogy 22: 140–150. Rahbek C., Gotelli N. J., Colwell R. K., Entsminger Silva J. M. C. 1994. Can avian distribution patterns G. L., Rangel T. F. L. B. B. & Graves G. R. 2006. in north Argentina be related to gallery-forest Predicting continental-scale patterns of bird expansion-retraction caused by quaternary cli- species richness with spatially explicit models. matic changes? Auk 111: 495-499. Proc. Roy. Soc. London, Ser. B, Biol. Sci. 274: Tello J. G., Moyle R G., Marchese D. J. & Cracraft J. 165–174. 2009. Phylogeny and phylogenetic classification Reif J., Hořák D., Sedláček O., Riegert J., Pešata L., of the tyrant flycatchers, cotingas, manakins Jrázslý Z., Janeček Š. & Storch D. 2006. Unu- and their allies (Aves: Tyrannides). Cladistics 25: sual abundance-range size relationship in an 429–467. Afromontane bird community: the effect of Tognelli M. F. & Kelt D. A. 2004. Analysis of de- geographical isolation. J. Biogeography 33: terminants of mammalian species richness in 1959–1968. South America using spatial autoregressive Ribas C. C., Aleixo A., Nogueira A. C. R., Miyaki models. Ecography 27: 427–436. C Y. & Cracraft J. 2011. A palaeobiogeographic Weir J. T. 2006. Divergent timing and patterns of model for biotic diversification within Amazo- species accumulation in lowland and highland nia over the past three million years. Proc. Royal Neotropical birds. Evolution 60: 842–855. Soc. B 274: 2399-2408. Weir J. T., Bermingham E. & Schluter D. 2009. The Sandel B., Arge L., Dalsgaard B., davies R. G., Gas- Great American Biotic Interchange in birds. ton K. J., Sutherland W. J. & Svenning J.-C. PNAS 106: 21737–21742. 2011. The influence of late quaternary climate- Wiens J. J. 2004. Speciation and ecology revisited: change velocity on species endemism. Science phylogenetic niche conservatism and the origin 334: 660-664. of species. Evolution 53: 193–197. Sanín C., Jiménez J., Fjeldså J., Rahbek C. & Cadena Wiens J. J. 2011. The causes of species richness pat- C. D. 2012. How much spatial variance in spe- terns across space, time, and clades and the role cies richness of suboscines in South America of “ecological limits”. The Quaterly Review of can be accounted for by diversification rates. Biology 86: 75–86. North American Ornithological Conference Wilf P., Cúnero N. R., Johnson K. R., Hicks J. F., 2012: PS1.28. Wing S. L. & Obradovich J. D. 2003. High plant Santos J. C., Coloma L. A., Summers K., Caldwell J. diversity in Eocene South America: Evidence P., Ree R. & Cannatella D C. 2011. Amazonian from Patagonia. Science 300: 122–125. amphibian diversity is primarily derived from Williams S. E., Williams Y. M., VanDerWal J., late Miocene Andean lineages. PLoS Biol. 7:3. Isaac J L., Shoo L. P. & Johnson C. N. 2009. Scherrer D. & Körner C. 2011. Topographically Ecological specialization and population size in controlled thermal-habitat differentiation a biodiversity hotspot: How rare species avoid buffers alpine plant diversity against climate extinction. Proc. Natl. Acad. Sci. USA. 106: warming. Journal of Biogeography 38: 406–416. 19737–19741. Sedano R. E., Burns K. J. 2008. Molecular Zachos J., Pagani M., Sloan L., Thomas E. & Billups phylogenetics and biogeography of a clade K. 2001. Trends, rhythms, and aberrations in of Neotropical tanagers (Aves: Thraupini). J. global climate 65 Ma to present. Science 292: Biogeogr. 37: 325–343. 686.693. Sekercioglu C. H., Schneider S. H., Fay J. P. & Loarie S. R. 2007. Climate change, elevational range

27