University Place

cc >> Welcome every one to Wednesday Nite at the Lab. I'm Tom Zinnen. I work here at the UW Madison Biotech Center. I also work for UW Extension Cooperative Extension, and on behalf of those folks and our other co-organizers, Wisconsin Public Television, the Wisconsin Alumni Association, which today is celebrating the 165th Founders Day, and the Wisconsin Science Alliance, thanks again for coming to Wednesday Nite at the Lab. We do this every Wednesday night, 50 times a year. Tonight it's my pleasure to introduce to you Professor Tom Givnish of the botany department here at UW Madison. He was born in Philadelphia, grew up there, and then he traveled the some 45 miles from Philadelphia to Princeton, New Jersey, where he got an undergraduate degree in mathematics. He stayed for his master's and PhD in botany. Then he went to Harvard for eight years and came here in 1985 on the faculty, and he's been here ever since. He has also been, at least 12 times, to South America, about a dozen times to other interesting places like Australia. I help organize a thing on campus called UW Madison Science Expeditions. It's a campus open house. He actually does science expeditions, and sometime I hope he'll take me along. I'm very good at carrying a duffel bag.

LAUGHTER

Tonight, he's going to talk about a hundred million years of Bromeliad evolution in the tropics of America. Bromeliads give us pineapples, and pineapples give us pina coladas, and pina coladas give us hope here in the bowels of winter.

LAUGHTER

Please join me in welcoming Tom Givnish to Wednesday Nite at the Lab.

APPLAUSE

>> Thanks, Tom, for those very kind words. It's a pleasure to be here, and I'm glad that we have such a fine turnout, especially considering how cold it is outside. We'll be talking mainly about tropical plants tonight, so perhaps things will warm up. I guess we can just dive right in here because we have a hundred million years to cover in 50 minutes, and that's two million years a minute so I better get cracking. So, our first glimpse of Bromeliads did not begin auspiciously. On Columbus' second voyage to the New World in 1493, he and his crew waded ashore on the island of Guadalupe to scout a Carib Indian village. It had been deserted in great haste with only a few infants crying in the dust as their terrified mothers fled. The log homes were thatched with huge palm fronds and ornamented with serpents carved on their pillars. Everywhere was the squawking of parrots, the quacking of domesticated ducks, and the bubbling of pots and pots of human flesh on the simmer, the slaughter from the fierce Carib's most recent raid. There were gnawed thighbones and skulls used as goblets scattered all around. Uninvited guests to this ghoulish banquet, Columbus and his crew were obviously horrified by the scenes and the stench of cannibalism all around them. But in a nearby garden, they found a curious fruit. Golden brown in color, beautifully patterned in spiral scales, delightfully fragrant and intensely sweet and piercingly acid when eaten. The Spaniards dubbed them pinas, or pine cones, after their visual appearance. These were, after all, the first pineapples ever seen by Europeans. The Carib and Tupi Indians named them Ananas, meaning excellent fruit. Explorers from England later called them pineapples after their resemblance to pine cones. Indeed, what we now call pine cones were called pineapples in English until 1694 when pine cone took over. Pineapples do not grow naturally in the Caribbean and were brought there by centuries of long distance trade among the Indians from dry forests a continent away on the frontiers of what are now Brazil and Paraguay in southern South America near famous Iguazu Falls, shown here. Columbus brought back several pineapples to the Spanish court were Ferdinand and Isabelle deemed them the best fruit they had ever eaten, fruit that was fit truly for a king and queen. Indeed for a long time, royalty were the only westerners with access to the few pineapples brought back by explorers. Pineapples remained so rare and coveted a commodity that in 1675 King Charles II of England posed for an official portrait receiving the first pineapple produced via hothouse culture by the royal gardener. At that time, the value of a single pineapple was estimated to have been the present day equivalent of $10,000. In time, pineapples were spread round the world in the tropics, first by the Spaniards, then the Portuguese, and then the Americans. In colonial America, pineapples were a symbol of hospitality, a mark of the wealth and sophistication of the hostess. Less well off families in an effort to make an ostentatious display would often rent but not eat pineapples, passing the fruit from family to family until it was consumed at an affair of surpassing elegance. In the meantime, several species related to pineapples made their way back to Europe so that in 1754 Carl Linnaeus described 14 such species, most of which he placed in the genus Bromelia, named after Olaf Bromelius, the city doctor of -- and foremost Swedish botanist in the generation before Linnaeus. Soon after, the plant family Bromeliaceae was created by a French botanist to house Bromelia and its relatives. Members of this family, the Bromeliads have tongue-like leaves arranged in a distinctive circle or rosette. In many Bromeliads, particularly those that grow as epiphytes as you see here on the branches of trees, the bases of those leaves fit together so tightly that they form a tank that can impound rainwater from which water and mineral nutrients can be absorbed. Like most monocots, Bromeliads have leaves with parallel veins and flowers with three petals and sepals. Uniquely among monocots, Bromeliads bear tiny umbrella-like hairs or trichomes on their leaves. In many Bromeliads, especially epiphytes, such hairs absorb water and nutrients, which is a useful capability for plants that have no contact with the soil on the ground. Today, we recognize the Bromeliaceae as having some 58 genera and roughly 3200 species. It's the largest of the 37 families of flowering plants found mostly or exclusively in the New World tropics of South America, Central America, and the Caribbean. Bromeliads include more epiphytic species than any family worldwide except the orchids, which is the largest family of flowering plants. Bromeliads range from the ancient -- or sandstone table mountains of southern Venezuela to the sun-baked granitic domes of the Brazilian shield, from the cloud forests of the Andes and the mountains of Central America to the Cypress swamps of the southeastern United States festooned with Spanish moss, another Bromeliad, and from the frigid Andean puna to the arid Atacama Desert. Many species have tanks, bear obstructive trichomes, and possess what's called chem photosynthesis and so take up CO2 at night to minimize water losses like many cacti do. Tank Bromeliads also often house a great diversity of insects including some that have adverse impact on human health and a variety of other arthropods, spiders, crabs, frogs, salamanders, and snakes. In a single acre of cloud forest, the tanks of epiphytic Bromeliads can impound thousands of gallons of rainwater and hundreds of pounds of humus high in the canopy while providing key sources of food for many primates and birds. A few tank Bromeliads are carnivorous plants. At least one is known to benefit for the prey captured by spiders living in the Bromeliad. Many Bromeliads are protected or fed by ants. Pollinators include a wide range of insects as well as hummingbirds, bats, and a very few perching birds. In many cases, the whole plant becomes a showy display to attract pollinators. The inflorescences, the flower spikes of -- from the Andean puna are the most massive of any flowering plant, while those of some dwarf Brocchinia and Tillandsia can be only a few inches tall. Some species have fleshy fruits dispersed by birds, or occasionally mammals, while others are wind dispersed. Finally, Bromeliads contribute a large share of the total number of species of epiphytes in the tropical forests and are particularly diverse at mid elevations around 1500 to 2500 meters. Previous attempts to understand the evolution of the Bromeliads, that is, where and when did they arise, how are they related to each other, what was their pattern of geographic spread, what key traits were fundamental to their spread and ecological success, and why they're so diverse, how foundered on three fundamental problems. First, Bromeliads are remarkably isolated in form from other flowering plants, making it hard to infer from morphology which groups would be their closest relatives or ancestors. Second, within the family there appears to have been extensive convergence in many traits which confounds efforts to infer relationships within the family based on form. And finally, within the family rates of DNA evolution are exceptionally slow, providing little information with which to infer relationships based on the sharing of mutations in their DNA. To overcome these problems and reach some decisive conclusions regarding Bromeliad evolution, I put together an international consortium of labs from Austria, Germany, Australia, England, Panama, and the US, especially including my long-term colleague Kent, who's here this evening, in order to sequence eight different loci, or regions, of chloroplast DNA in 90 species of Bromeliads. Within each leaf cell there are dozens to hundreds of chloroplasts, and within each chloroplast there are dozens to hundreds of loops of chloroplast DNA, which then replicate as individual chloroplasts and leaves do and are passed clonally, with the occasionally mutation, to succeeding generations, first within species and then ultimately to species derived from ancestral forms. Sequencing large numbers of chloroplast genes are better of the more rapidly evolving spacers within and between genes can provide a powerful tool for inferring relationships among plants based on literally hundreds of characters. Because early studies showed such a slow rate of chloroplast evolution of Bromeliads, lower than in almost any monocot group except for the palms, we decided to sequence eight fast evolving genes, introns, and spacers to infer relationships within the Bromeliads. For more than a century, based on morphology, Bromeliads have been classified into three subfamilies, illustrated here from this ritual being conducted by Indians in the Atacama Desert. The Pitcairnioideae, I know it's a mouthful but let's bear with it, have winged seeds. Tillandsiod or Tillandsiods have -- seeds, and Bromeliodeae has fleshy fruits. Similarly, based on morphology, Bromeliads have long been thought to be related to a wide range of monocots including gingers or sedges or agaves, the families Rapateaceae or Velloziaceae. And this rag-tag collection of groups spans four of the 12 orders of monocots. Research over the past decade culminating in a recent analysis based on the sequences of 81 genes from the chloroplast genome obtained by another research group I lead places the Bromeliads in the order Poales. And I'll just highlight here the Bromeliads. There they are. So they're in the order Poales, the order of the grasses, sedges, and their relatives. The Poales in turn are a sister to two orders composed of dayflowers and their relatives and another order composed of gingers, bananas, and their relatives. And those groups, in turn, are related to the palms and the odd woody family -- from Australia. Now, in this phylogeny, or family tree, the inferred ancestor of any two species of genera or families is represented as a branch that splits into two. Such ancestors are, in almost every case, today extinct. Branches deeper in the tree represent ancestors of ancestors, so the overall branching topology, or pattern, represents the inferred pattern of evolutionary descent. Our data from the International Consortium places Bromeliads sister to all other families of Poales, with the cattails and their relatives being sister to the next subset of families, then Rapateaceae to the next nested subset, then sedges and rushes, then grasses and -- and ultimately the grasses. Now, to resolve relationships within and among the Bromeliads are consortium sequence nearly 10,000 bases of chloroplast DNA from eight rapidly evolving genes, introns, and spacers for representatives of 46 of 58 genera. The genera we didn't include add up to less than 2% of all Bromeliads, and their relationships to the other genera are pretty clear based on form. When we analyze these data, it produced a well resolved, well supported family tree for the Bromeliads with eight major branches which I've color coded for us here. And that led us to recognize eight subfamilies rather than the traditional three. Our molecular data did support two of the traditional subfamilies, the -- seeded Tillandsiodeae, shown here in gold, and the fleshy-fruited Bromeliodeae, shown here in red, as being monophyletic. That is to say, as comprising the set of all descendants from a single ancestor. No more, no less. By contrast, the traditional subfamily Pitcairnioideae, marked by winged seeds and highlighted now by the green bars I just put on the slide, is paraphyletic. That is to say, some of the ancestors that lie outside the traditional subfamily in, for example, Tillandsiodeae and Bromeliodeae. So the old style Pitcairnioideae is not a natural group. And apparently, winged seeds that characterize the old Pitcairnioideae were ancestral to the Bromeliads as a whole, so the possession of winged seeds is not a reliable guide to clades, what are called clades or derived lineages, monophyletic groups within the Bromeliads. The pattern of branching that we see here forces us to recognize at least eight subfamilies if each of these families is to be natural, is to be descended from a single ancestor in each case. As you see, as you look as this tree, you can see that the branching pattern is essentially ladder-like, with Brochinioideae, this subfamily of Brochinioideae, sister to all other subfamilies. Then Lindmanioideae is sister to all the rest. Then the Tillandsioides, then the Hechtioides, then the Navioides, then the Pitcairnioides in a new narrow sense, and then finally at the top, the Puyoides, sister to the Bromelioides. Within the Bromeliaceae, the remarkable genus Brocchinia forms the subfamily Brocchinioideae, a small clade of just 20 species, sister to all the other Bromeliads. Brocchinia is endemic, restricted to the Guiana Shield in northern South America, a large area of ancient rock crowned by sandstone table mountains. And Brocchinia is nearly restricted to the extremely nutrient poor sands and sandstones atop the Tepuis, the flattop sandstone mesas, and adjacent sand plains. In response to the extreme nutrient poverty in these areas, Brocchinia has undergone an adaptive radiation in mechanisms of nutrient capture unparalleled at the generic level in angiosperms, including carnivorous species, ant-fed species, tank epiphytes, tree forms, and nitrogen fixers, in addition to a number of more typical forms without tanks that depend on roots for nutrients. My colleagues and I documented the ecology, adaptations, and phylogeny of Brocchinia based on several expeditions that we've made to the Tepuis and surrounding areas. Tepuis cut off from the surrounding tropical lowland forests and savannas by soaring cliffs of the ancient sandstone, uplifted beds of the sea were long a mystery to biologists, leading Sir Arthur Conan Doyle to pen his famous novel The Lost World in 1912, envisioning a plateau isolated from time and inhabited by dinosaurs. Alas, there are no T Rex atop the Tepuis, but there are many plants and animals found no where else on Earth. These include many genera of such families as the Iridaceae, the Rapateaceae, the pipehorse of --, all of them closely related to Bromeliads, as well as the -- related to tea and similar shrubs and woody members of the daisy family that are pollinated by hummingbirds. The flora also includes a remarkable number of terrestrial orchids and recognized carnivorous plants, including, as you see here, the sun pitcher Heliamphora and also large numbers of sun dews, the reddish plants here highlighted against the sand. On one of our expeditions to the tallest and wettest Tepuis, to the tallest and wettest Tepuis, my colleagues and I demonstrated that --, this bizarre but quite beautiful shrub in the tea family resembling an artichoke atop a pole was fire adapted. The stem was three-quarters bark, and the massive leaf rosettes protected the single terminal bud from the brief fires that run across the rocky substrates. Now, the soils atop the Tepuis are extremely infertile, derived from the ancient sandstone and heavily leeched by the extremely rainy, humid conditions atop the Tepuis. Today, there are over 50 such sandstone table mountains, usually over 2,000 meters of elevation in southern Venezuela and adjacent countries. Some are quite small, like Cerro Autana in Amazonas, while others are gigantic, like Cerro Duida along the upper reaches of the Rio Orinoco. The ecological isolation provided by the steepness of the cliffs and their sheer elevation is not to be underestimated. Early explorers had to access the summits via long canoe trips followed by arduous hikes up perilous steep trails. Latter day explorers, like my colleagues and I, had it much easier using motorboats and helicopters to access many peaks. But helicopters are not without their own peril. Once while we were working in Amazonas in Venezuela, a helicopter coming to pick Dan and I up crashed and burned. On another day I was sitting inside a copter with a pilot on a hand-cut helipad, 2,000 meters elevation on remote Cerro de la Neblina, the foot of a 300-meter cliff with a steep slope to our rear. The pilot, in cowboy fashion, decided to lift off, pivot 180 degrees, and dive down to gather speed rapidly all in one motion rather than doing it one step at a time, which was great except at just that moment a downblast came down the cliff, and we started to sink like a stone faster than the ground was falling way below us. I watched the air speed dial as the forest canopy came up closer and closer. Forty miles per hour, then 50, and I was thinking we're not going to really walk away from this. There were alarms going off in the cabin as the engine was overtorquing, and fortunately we pulled out at just the last minute as you might have guessed because I'm here tonight.

LAUGHTER

The pilot left the very next morning. His nerve was gone. I stayed and flew many other days to collect data and observations. Now, carnivorous species of Brocchinia have steep yellow leaves that form a vertical cylinder coated on the inside with a fine, waxy dust that readily exfoliates and tumbles unwary ants or bees into the tank, which is highly acid and full of their remains, as you can see. The tank fluid emits a sweet, nectar-like odor that serves with the bright yellow leaf color to attract prey. We demonstrated through field experiments with radioisotopes that the peltate leaf trichomes can absorb mineral nutrients and amino acids, and hence nitrogen, at higher rates. The two carnivorous species of Brocchinia are specialized on ants versus bees and wasps, and they often are able to colonize bare rock and presumably are aided in this by eating ants that are scrambling around on the rocks. They also, in large numbers, can colonize a recently burned sites. I hope you appreciate the scale here. We're looking at acres upon acres of these carnivorous plants. Another species, Brocchinia acuminata, is ant-fed. The ants build their nest among the inflated, incredibly tough, non-photosynthetic leaf bases. Notice that those leaf bases come together toward the tip. That prevents large amounts of rainwater from entering and flooding the ant nest but allows some to get in. Then you see that the long leaf tips are green and photosynthetic. This species has unusually large trichomes that can absorb nutrients and water at high rates. Large air channels connect the non-green leaf bases to the green leaf tips. I've not yet had the opportunity to test the idea that these channels allow the plants to essentially turbocharge photosynthesis with ant breath, CO2 rich ant breath. And on some of the slopes of the Tepuis, we have a gigantic tree-like Brocchinia, Brocchinia micrantha. Here it's growing in large numbers by Kaieteur Falls in Guyana. This is four times taller than Niagara Falls. The amount of mist is so great that no woody plants really can grow in the mist zone, and that zone is dominated by plants that are up to 10 feet to 20 feet tall of Brocchinia micrantha. Inside many of these plants are golden arrow poison frogs. It's not clear what their function is. Clearly the frogs benefit from the enormous amounts of rainwater and mist water they collect in their leaf axils. But it's possible that the frogs assist the plant by pooping in the tank and adding nutrients to the tank. And I should say that just that has been demonstrated very recently in one species of Tillandsioideae Bromeliads where a tree frog is involved with pooping and aiding the plant. A quarter of the nitrogen in the plant comes from tree frog poop. Such poop assistance occurs in other plants that impound rainwater in pool-like areas. So in Borneo we have a large number of tropical pitcher plants in the genus Nepenthes, most of these eat insects. They're simple carnivorous plants. But one, quite recently, has been shown to serve literally as a toilet for tree shrews that come to the pitcher, they perch on the pitcher, and they lick away all the nectar on the lid, and while they're doing so, they defecate and urinate and the plant's better off. And another remarkable story has also developed for another species of Nepenthes where the pitcher really isn't catching insects so much as it is serving as a bivouac, as a tent for a bat, small, wooly bat, and it releases nutrients, phosphorus and nitrogen, in its excrement, and as a consequence helps the plant along. We don't know whether systems like that exist in Bromeliads. They could quite possibly. We do know that in at least one species, spiders feeding on prey and then crapping into the tank add to the growth rate of the Bromeliads. In another case we know that the nymphs of damselflies that live inside of the tanks speed up the decomposition of prey or of animals or plant material that fall into the tank so much that it benefits the plant. All right, let's return to our family-wide phylogeny. And note that above Brocchinioideae is the subfamily Lindmanioideae, the next rung on the ladder, sister to all the Bromeliads except for Brocchinia. Lindmania and Connellia within this genus are terrestrials. They don't occur up in the trees. They have non-impounding leaves, and they are restricted to the Guiana Shield, like Brocchinia, suggesting that the family may have indeed arisen there. Tillandsioideae, the subfamily most specialized for epiphytism and including one-third of all Bromeliads, diverge next. Most taxa occur in cloud forests in the Andes, Central America, and the Caribbean. Next to diverge were the drought-adapted species of Hechtia. These have shiny and spiny leaves. They're found in the semi deserts of Central America, Mexico, and southern Texas. Hechtia's position here low on the tree is somewhat surprising because it shares a number of leaf traits with other genera that are up close to the top of the tree, like Puya, Dyckia, and Encholirium. But these are, while they occur in dry areas, are not at all, as you can see, closely related to Hechtia. The Navioides, the next rung on this ladder, shown here in purple, are sister to the remaining Bromeliads, and include Cottendorfia from the Brazilian Shield and Brewcaria, Navia, and Sequencia from the Guiana Shield. All of them are terrestrial plants not found up in the trees with tough and non-impounding leaves. Navia is the most species rich Bromeliad genus in the Guiana Shield, but it's ecological very stereotyped. Its diversity appears to be driven not by adaptive radiation, but by poor dispersal. It's the only genus that lacks any specialization for seed dispersal. So the plants don't get around very much and they speciate locally. The single species of Sequencia was originally placed in Brocchinia, but our DNA sequencing showed that it belonged in Navia instead, and hence we applied this name. The redescribed, redrawn Pitcairnioides include Pitcairnia, which is sister to the remaining genera, Fosterella at mid elevations, and then Dyckia, found in dry habitats throughout much of Brazil, and then Encholirium up in the most arid part of northeastern Brazil, and Deuterocohnia, a cushion plant found high in the Andes, above tree line, and again under very dry conditions. Pitcairnia grows from sea level to the high Andes, mainly in the understories of rainforest and cloud forest, as marked by broad leaves and showy, reddish flowers, pollinated primarily by hummingbirds. As I mentioned, Fosterella occurs at mid elevation. Deuterocohnia grows as a cushion plant up at high elevations in the Andes. Puya is centered in the Andes, and several taxa bear spectacular iridescent flowers and immense terminal inflorescences. Puya consist of two natural lineages, as Rachel -- and Kent -- have shown. One of these is composed primarily of Andean species from high elevations, and a second consists of a small number of species from coastal Chile. It was once argued that the gigantic Puya raimondii, which you see here, was carnivorous, a carnivorous plant, sort of frightening to think about, a plant taller than you being carnivorous, because small birds often nest among its spiny leaves were occasionally found impaled on the spines which would perhaps result in nutrients dripping from their corpses onto the ground and being taken up by the plant. The Brits, who made this argument, also alluded to the fact that the leaf spines are recurved, they bend back toward the center of the plant rather than projecting outwards, which to them suggested that the spines could not be defensive in nature. They couldn't defend against llamas or other herbivores running around the puna. However, I pointed out that the Andean spectacled bears often eat the inflorescences of up to 90% of the individuals in some Puya populations, which effectively kills them because the Puyas only flower once and then die. And you can imagine, as I did, that as an ursine shimmies his way up a Puya, the leaves will bend under its weight, reorienting the spines, and ultimately bringing them, in fact, to bear.

LAUGHTER

I know, it's a long build up. Finally, the flesh-fruited Bromeliads are sister to Puya. The earliest divergent Bromeliads are all small terrestrial genera, including Fascicularia, Ochagavia from low elevations in southern Chile, Bromelia from throughout South America, and then followed by Ananas, the pineapple genus, from the Brazilian Shield. All Bromeliads except pineapple have separate berry-like fruits. In pineapple these have become fused to produce its distinctive compound fruit. Some of the Bromelioides are consumed locally for food. Wild pineapples, by the way, do set seed which decreases the value of the fruit, and it is expressly for that reason that Hawaii has prohibited the import of hummingbirds because they might fertilize all of the pineapples that are grown there. Now, we used our molecular phylogeny for the Bromeliads, calibrated against time using monocot fossils, and tied to the distributions of present day species to infer the tempo and the geographic pattern of spread of the family. Here are the historical biogeography of present day Bromeliads and ancestral taxa as shown by branch color for inferences based on one kind of logic, namely parsimony, by the colors of the pie diagrams in each note of Bayesian logic, and by the larger pie diagrams using yet a more convoluted form of logic. As you can see, however, the conclusions reached using these three different analytical approaches are quite similar. Our phylogeny implies, based on the color of the branches that we've color coded the ancestral forms, that the Bromeliads arose in the Guiana Shield a hundred million years ago. Now, the origin of the family is well off the screen. It's about here. So a hundred million years ago. And it also implies that the genera now in existence began to diverge from each other only about 20 million years ago. So there's an 80 million year gap between when Bromeliads ancestors diverged from other monocots and when the living Bromeliad genera started diverging from each other. That 80 million year gap points to a lot of extinction and helps account for the difficulty that botanists have had using either morphology or DNA to infer the closest relatives of the Bromeliads. Again, based on this reconstruction, Bromeliads began spreading outside the Guiana Shield about 15 million years ago. Initially into the Andes, Central America, the northern South American coast, and the Caribbean. There were two major invasions into the Andes around 15 million years ago by ancestors of the Tillandsioides and of the Pitcairnioideae, Puyoideae, Bromeliodeae lineage. Hechtia also invaded Central America at about the same time, with several Tillandsioides later spreading into that area. There were two major invasions of the Brazilian Shield, that's the light colors here, by Dyckia and Encholirium about 11 million years ago, and then by the core Bromelioides about nine million years ago. The latter invasion was actually confined mostly to the very humid Atlantic forest region along the raised coastal edge of the Brazilian Shield. Pineapples arose about six million years ago in southern Brazil, again near Iguazu Falls. Tragically, however, it wasn't until 1952, only 61 years ago, that they were used to invent pina coladas.

LAUGHTER

Now, the ancestor of the single species of Bromeliad to occur outside the New World in West Africa, we infer that that occurred via long distance dispersal no more than nine million years ago, which rules out continental drift. The southern Atlantic opened up about 90 million years ago. Now, we, by we I mean Ken, myself, and our colleagues, have made innovative use of this Bromeliad phylogeny to test a series of hypotheses about how environmental factors and plant traits could have shaped Bromeliad distribution and diversification. So this slide, which looks a little complicated but just follow through, it's actually not that complicated, summarizes our hypotheses. First, fertile montane habitats favor the evolution of epiphytes by creating high humidity and rainfall, low evaporation, and a rich rain of nutrients. Second, epiphytism should favor the tank habit, absorptive trichomes, and CAM photosynthesis in order to provide and conserve water and nutrients absent any contact with the soil. Third, pardon me, before we get to third, the tanks should be most effective at capturing moisture and nutrients in fertile, rainy, montane habitats. And tanks, absorptive trichomes, and CAM should favor epiphytism in return. We also know that arid sites on the ground should favor CAM photosynthesis. Third, epiphytism should select for entangling seeds. That is to say, seeds in fleshy fruits or with fine plumes facilitate attachment to twigs in the branches of epiphytic hosts. Such seeds would also favor epiphytism and result in strong dispersal ability. Fourth, cool, wet, montane conditions should select for thermal regulating pollinators, notably hummingbirds, in the New World. Insects are not going to be very active under such damp, cold conditions. Fifth, epiphytism should favor high rates of net species diversification by providing access to an extensive, a dynamic, an ecologically diverse substrate, namely the forest canopies, that are inhabited by few other plants. Sixth, floral coevolution with rapidly radiating hummingbirds with different bill sizes and shapes should also favor a lot of speciation. And finally, moderately high dispersal associated with entangling seeds combined with life in extensive topographically complex mountain chains like the Andes or the -- in coastal Brazil should promote high rates of diversification, allowing lineages to disperse down such chains occasionally and then diverge and speciate in parallel at several points simultaneously. Now, our group tested these ideas by mapping the evolution of various traits and ecological conditions on our family tree and then conducting formal tests of correlated evolution. And we conducted formal tests to see whether particular traits were associated with higher rates of species diversification. And finally, we quantified the extent to which several different radiation, several different kinds of divergence ecologically contributed cumulatively to Bromeliad diversity. Here, and in the next slides, we are plotting the rise of several ecologically important traits including by branch color as a function of time, which is horizontal position, and of space, indicated by the background shading. I don't expect that any of you can read the names. It's not important that you read the names. Look at the colors of the branches. And, as shown on the left, there were two major origins of epiphytism in the Tillandsioides in the Andes 15 million years ago just as the northern Andes were beginning their major uplift, and in the Bromelioides in the Brazilian Shield. That's the second reddish set of branches up near the top, 5.7 million years. Again, as Cerro de Mar and other coastal ranges began their uplift. So the evolution of the tank habit, shown on the right, largely parallels that of epiphytism. The two traits are very closely correlated. Entangling seeds arose at the same time as epiphytism, and the tank habit in Tillandsioides but before either evolved in the Bromelioides at the top of the tree. CAM photosynthesis, shown in red on the right, evolved at least five times. Mostly in association with arid terrestrial sites starting at 15 million years ago at a time when atmospheric CO2 was going down, where CAM photosynthesis would have been favored. Also at a time when there was a lot of mountain building and rain shadows being created, leading to drying. CAM photosynthesis has persisted in a large group of tank epiphytic Bromelioides in Brazil and arose sporadically in a number of Tillandsioides and Puya. Avian pollination, which is mainly hummingbird pollination, arose at least twice in Tillandsioides and the Pitcairnioides, Puyoides, Bromelioides clade in the Andes about the time the northern Andes uplifted and when the two major clades of Andean hummingbirds started to diversify. Interestingly, all origins of bat pollination evolved from bird pollination, mostly at low elevations. High elevations work against bat pollination given the high evaporative potential of naked wing membranes at lower temperatures at high elevations. And also bat pollinated plants need to produce more photosynthesis because they have to produce much more sugar to attract bats versus hummingbirds. And the plants at higher elevations, cooler temperatures, cloudier conditions aren't as productive. Now, as highlighted here in red, and again let's not worry about reading all of this, just look at how much red there is, all associations between pairs of plant traits and ecological conditions the we predicted by our model in fact do show significant patterns of correlated evolution after we take relationships into account. The association between epiphytism and the tank habit is especially strong. Seventeen of the 25 pair-wise comparisons in which we did not predict correlated evolution, in fact fail to show any significant correlation. However, as highlighted in blue, there are a few cases where there are unexpected patterns of correlated evolution. One unexpected correlation makes sense after the fact. It turns out that tanks are frequently associated with avian pollination, and that makes sense because the tanks would attract small insects that the hummingbirds feed on, and they also provide a lot of fluid required for the copious production of nectar in hummingbird flowers. Now, when we use the tree to identify accelerations of net speciation, we found that there were three points in our simplified family tree where there were accelerations. And these included the Tillandsioides, shown in gold. The Navioides, pardon me, the ancestor of the Navioides, Pitcairnioides, Puyoides, and Bromelioides, that's the point highlighted with the green arrow. And then a further acceleration within that lineage up at the top, that red arrow pointing to the blue triangle, in the crown tank epiphytes of the Bromelioides. Here, horizontal distance is proportional to time, and the height of the triangles is proportional to the number of species in each lineage. When we further analyzed the pattern of diversification, it showed that tank formation and epiphytism and hummingbird pollination were strongly correlated with the speciation rate, the net speciation rate, but, actually, life in extensive, fertile, montane regions like the Andes and the Cerro de Mar was the strongest effect. The net rates of species diversification are five times higher in the Andes and the Cerro de Mar than elsewhere. Now, the acceleration of net species diversification by particular traits or environments is an important feature that underlies Bromeliad diversity. But we also have to consider the additive effect of several broad scale radiations on the full range of conditions invaded by the family as summarized by this slide. Based on the association of traits with the invasion of particular regions by particular lineages, seven major radiations appear to have generated more than 85% of all Bromeliad species. First, the rise of primarily C3 epiphytes in the Tillandsioides, shown here in dark green, accounts for about 1200 species in the Andes, Central America, and the Caribbean. Second, the rise of CAM epiphytes in Bromelioides, shown in bright green, accounts for about 600 species in the mountains of coastal Brazil. Third, fourth, fifth, and sixth, the origins of CAM photosynthesis in terrestrial plants accounts for about 50 species of Hechtia in Central America, 170 species of Deuterocohnia, Dyckia, and Encholirium in arid portions of the Andes and Brazil, 100 species of the early divergent Bromelioides in semiarid regions of Chile, and over 200 species of Puya, mostly in the arid to semiarid elevations of the Andes. Finally, the diversification of the broad-leaved Pitcairnia in tropical forest understory, shown in blue here, especially in and near the Andes, generated nearly 400 species. The three major clades that did not expand outside the ancestral Guiana Shield, outside those poor mountains, those nutrient poor mountains, those very wet mountains account for only 200 species in present day diversity. In previous analyses, it's been impossible to tell which came

first

adaptive radiation or species diversification. Our approach here shows it incorporating the impact of trait evolution on the physiological capacity to spread into new areas. Why CAM plants invade dry areas, why tank epiphytes proliferate in montane areas with rich bedrock can suggest that the arrow of causality flies from adaptive divergence to ecological innovation and the invasion of new areas, and then finally to species richness, to diversification. Rather than species richness somehow being correlated with adaptive radiation via some sampling artifact. We believe that this kind of approach where we're testing a priority hypothesis can throw additional light on the relationships among morphological evolution, geographical spread, and species diversity in many other groups and make historical biogeography a hypothesis driven enterprise. Clearly the approach has yielded many insights into the evolution of Bromeliads, and these insights are, to our knowledge, unmatched in studies of any other major plant lineage today. In conclusion, my colleagues and I would like to thank the National Science Foundation, the Austrian Academy of Sciences, the -- and the -- Initiative for financial support, as well as the Selby Botanical Garden, the Palmengarten Frankfurt, and the Heidelberg Botanical Gardens for access to critical plant material. Thank you all very much, and I'd be happy to take questions.

APPLAUSE