FIELD STUDIES ON POLLINATION OF PLANTS OF THE GENUS PUYA
by Fernando I. Ortiz-Crespo in J. Brom Soc. 23(1):3-7. 1973 & 23(2): 54-57. 1973
The tubular corollas and copious nectar found in the flowers of many bromeliads point out to birds as possible pollinators, but this cannot be easily confirmed because there are very few recorded observations of bromeliad visitation by animals. To remedy in part this state of affairs, and thanks to encouraging remarks by Amy Jean Gilmartin, I want to summarize here my notes on the subject of Puya pollination; these were made mostly from 1968 to 1970 when I studied hummingbirds in highland Ecuador, but are supplemented here by information gleaned from a number of publications. Support for field work in Ecuador came from the Chapman Fund of the American Museum of Natural History and Dr. O. P. Pearson, Director Emeritus of the Museum of Vertebrate Zoology of the University of California, Berkeley.
The rich diversity of Ecuador's flora and fauna has been known ever since the first European collectors traveled there in the 17th and 18th centuries; it has now been established that the country contains about 12 % of the species in the bromeliad family but amounts to only 3% of the family's area of distribution (Gilmartin, l972). Similarly, over a third of the species of hummingbirds ever described have been collected in Ecuador, which represents less than one fiftieth of the total range of the hummingbird family. These facts should make it clear that the country offers outstanding opportunities for the study of these plants and birds.
The diverse ecology of the northern Andean region results from its mountainous topography and equatorial location. Altitudinal life zones are present that follow climatic strata more or less distinct from one another in temperature and rainfall. Moving up from the warm lowlands to the cooler flanks of the Ecuadorian Andes one notices both a change and a drop in the constituent species of the various communities; luxuriant rain forests give way to simpler montane forest and shrub associations and these in turn are replaced by scrub and bunchgrass prairie above the level of cloud condensation - near the mountaintops and in the rain-shadowed inter-Andean valleys. The bromeliads and hummingbirds of these communities follow this altitudinal trend; in the Quito valley at 2850 m above sea level, my headquarters during the extent of the study, the number of species of these organisms was comparatively small and this facilitated somewhat their identification and observation.
The most striking change in the bromeliads as one leaves the rain and mountain forests below is, of course, the disappearance of epiphytic species. Partly as a corollary terrestrial species increase in relative abundance; the genera Pitcairnia, Tillandsia and Puya are particularly well represented above 3000 m; the latter two prevail on the slopes up to 3000 m, and Puya species are the exclusive bromeliad inhabitants of the Paramo bunchgrass prairie from 3500 to about 4200 m. If one were to plot the size of individual bromeliads against elevation along a transect in Ecuador a positive correlation might be revealed. Clearly some of this result might be attributed to the prevalence of Puya species at high elevations, these being typically large-sized plants, but, because the trend is also seen within the genus and is demonstrated by terrestrial Tillandsia, it is possible that large size might have some adaptive value in the forbidding climate of the Andean heights. The largest bromeliads I found were Puya plants with rosettes about 2 m in diameter growing at 4000 m elev. in the Paramo of Mt. Cayambe and in the Llanganati Mts. ; these plants' size approaches that of the giant of the bromeliad family, Puya raimondii, whose colossal stems, leaves, and inflorescences rising up to 8 m or more can only be seen at very high elevations in the Peruvian and Bolivian Altiplano. This species' exceptional dimensions, as well as those of the largest Ecuadorian Puya, could be related to pollination needs. We shall return to this after we know more about the plants' ecology.
Observations I made at the University of California Botanical Garden first suggested that hummingbirds could be among the natural pollinators of Puya plants. In 1967 and early 1968 I investigated California hummingbirds, mostly the Anna and Allen's Hummingbirds (Caylpte anna and Selasphonus sasin, respectively), at a study area that included the Botanical Garden and adjacent portions of the Berkeley hills. The garden is favored by these birds not only because the plants there are kept lush and green throughout the year, but also because there is an outstanding outdoor collection of exotic plants that produce nectar in abundance. I kept a record of the plants the birds visited and particularly of those where individuals established feeding territories. I was struck by the fact that hummers began exploring new Puya inflorescences days before any flowers had opened, and foraged regularly at an inflorescence as soon and as long as nectar was accessible. Moreover, inflorescences forming a clump were invariably taken over by an individual bird - usually an adult Anna male - that foraged there and excluded other hummers from the site for the one or two months that flower production lasted. The resident bird could often be distinguished by Puya pollen marks left on its front by the stamens of the flowers it had harvested; this and the fact that plants at a territory fruited abundantly led me to believe that successful pollination had been accomplished by the birds (for more details see Ortiz-Crespo, 1969).
My work in Ecuador began in late 1968, after the preceding observations had been made; I was thus aware of the possible mutual dependence of Puya and hummers, and when in the field studying the latter I kept an eye open for these plants. In late November, 1968, I located a dense population of Puya aequatorialis growing at Caspigasi on the dry northwestern slopes of Mt. Casitagua at an elevation of some 2900 m. This locality is about 20 Km north of Quito and 4 or 5 Km west of the Equator monument pictured in so many local postcards. When first discovered the plants had nearly full-grown inflorescences but no open flowers; nevertheless a few hummers were found, mostly Sparkling Violetears (Colibri coruscans) and Giant (Patagona gigas). By late December most plants were in full bloom and their flowers were coveted by many violetears and a few giants, the latter having some difficulty in outmaneuvering the smaller violetears around the relatively short -1.5 m - inflorescences. Individual violetears settled at clumps of inflorescences where they fed and from which they excluded other hummers. A clump often included 5-10 stalks rising from one or more plants; each was seen to bear flowers of uniform color, but different plants could have flowers colored differently from dark purple to greenish yellow, presumably representing genetic morphs. The birds were obviously unconcerned about color, since various "morphs" could be included within one territory. I assume this or a similar territorial pattern of Puya use prevailed until about mid-January, but most plants were fruiting and few new flowers were found on January 26, when territorial violetears were no longer in evidence and hummer activity in general had ebbed.
The Caspigasi area was again observed in late 1969-early 1970; a similar pattern of hummer occupancy was then recorded, but probably because I was then better trained than a year earlier I noticed two other species of hummers feeding at the plants, Black-tailed Train-bearers (Lesbia victoriae) and Purple-collared Woodstars (Myrtis fanny). Remarkably enough for plants of a relatively aseasonal region, the flowering of the Puya there was as synchronous in 1969-1970 as it had been in 1968-1969; the peak occurred in early February rather than late December, however. On 3 February 1970 I spent several hours watching a 150 x 40 m hillside stretch covered with inflorescences in full bloom that were defended by about 15 violetears, each of which repelled other violetears and birds of the three other species from its territory. Most intruders were discouraged, but a few individuals of all four species managed to partake of the food there. The Giant Hummingbirds seemed to be the most successful thieves; persistent individuals of this species could easily tackle the smaller violetears (20 vs. 9 g. in body weight). In general, however, the giants seemed to prefer the flowering stalks of neighboring Century Plants, Agave americana, to those of the Puya plants, and there they were capable of keeping off intruders of all other species. Nectar reward might be critical here in addition to aerial mobility; plants producing more nectar than P. aequatorialis, such as Agave, might be easier to defend for giants simply because the profit to be gained by defense is relatively greater.
It is unfortunate that my other Ecuadorian records cannot be as clear as the previous ones on the identity of the plants concerned. However, I might get a measure of sympathy from the reader considering that collection of Puya specimens for identification is an unpleasant task, often made painful by the tough and thorny foliage which characterizes the genus. I, for one, preferred to leave this to the botanists and concentrated on hummingbirds. Thus, on December 28, 1968 I recorded a Great Sapphirewing, Pterophanes cyanopterus, moving about the conspicuous stalks of a green-flowered Puya in Paluguillo Pass, some 35 Km east of Quito as the road Quito-Papallacta winds its way up to an elevation of 3300 m. Two species, Puya sodiroana and P. pichinchae, have been collected there; however, my notes on the plants I saw suggest P. clava-herculis, but this cannot be confirmed in the absence of a specimen. Similarly, on 18 February 1969 I observed a Giant Hummingbird feeding from the stalks of a blue-flowered Puya common in the smaller of the two islands at Lake Cuicocha, a crater lake on. the slopes of Mt. Cotacachi at an elevation of about 3000 m. The plants could be P. glomerifera, but once again no specimens are available to substantiate this possibility. Incidentally, this record constitutes a 40-Km extension of the Giant Hummingbird range northward in South America.
The relation between Puya and hummingbirds implied in the preceding paragraphs can be confirmed by publications dating from as early as over a hundred years ago; however, this information can easily be overlooked because it is found in zoological rather than botanical journals. The earliest reference appears to be by Thomas Bridges, a collector who sent birds from Bolivia, Chile and northwestern Argentina to England around the middle of the Nineteenth Century. Notes accompanying a shipment of his specimens were published by Fraser (1843) : these indicate that "Pourretia coarctata" (= Puya chilensis) is a major food source for the Giant Hummingbird near Valparaiso, Chile, where breeding birds of this species reside in the Austral spring and summer. These observations are independently validated by Landbeck (1876), who states that the Giant Hummingbird not only prefers "Pourretia coarctata" and "Pourretia gigantea" (Puya chilensis and probably Puya berteroniana), but that the bird ranges in central Chile over the same altitudes as these plants. Additionally, Giant Hummingbirds have been recorded visiting 2 m-tall "Pourretia" (Puya) inflorescences in the Peruvian highlands (Dorst, 1956).
To these more or less casual records we must add the observations of Federico Johow (1910), a botanist who taught at the Universidad de Chile in Santiago around the turn of the century and who wrote an excellent flora of the Juan Fernandez Islands (1896). Although his writings are a monument to Chilean science they have remained relatively obscure because they appeared in Chilean and German journals. One of the few English references to his work is by Robinson (1950), but this author cites only a portion of Johow's finding on Puya. It seems worthwhile therefore to review the highlights of Johow's work.
He had noticed that Chilean Puya present a peculiar pollination syndrome because of their showy tubular corolla, lack of floral scent, abundant nectar, sticky pollen and simple stigma. Taken together these features seem to point to birds and rule out insects or wind as pollinators. In fact, by denying birds access to Puya flowers one could easily prove that seed production is upset; furthermore Johow noted that except for the introduced Honeybees no other insects were particularly fond of Puya flowers at least as pollinators. He kept a list of the visitors he saw in the plants and noticed that the most consistent one is the Giant Hummingbird, recorded at P. chilensis, P. "caerulea" (P. berteroniana) and P. venusta. Another common visitor is the Austral Blackbird, Curaeus curaeus,, which perches on the projecting spikelets of the inflorescences of P. chilensis and P. "caeurulea" and has no trouble introducing its face and relatively short bill into the shallow corolla tubes of these plants to drink nectar. After this happens a bird might show a bright smear of Puya pollen on its face; these blackbirds are omnivorous and regularly feed on Puya nectar when the plants bloom in the Austral spring, so that whole flocks seem to consist of pollen-dusted individuals at places where the plants bloom in concentrations.
Interestingly enough, spikelets are missing and flowers are comparatively narrow and deep in P. venusta, which appears to be visited by giant hummers but no other pollination agents; the accessible cuplike flowers of the other two species provide nectar for hummingbirds, blackbirds and even for some finches and flycatchers which, like the blackbirds, gain access to the nectar from a perching position. Johow concluded from this that adaptation to particular pollinators is shown not only by floral characters but also by the shape of the inflorescence in these plants, the broadly ornithophilous forms providing shallow blossoms and perches for short-billed birds unable to hover.
Besides birds an insect may pollinate Chilean Puya, namely, Castnia eudesmia, a large and colorful diurnal moth which is known to pass its larval and pupal stages on P. chilensis and P. alpestris; the adult form feeds on the plants' nectar much like a hummingbird, and like one gets dusted with pollen and thus it is presumed to effect pollination. Futhermore, individuals occasionally drive other insects and birds from their food plants. This information was reported by Gourlay (1950), who based it on an account in Spanish by Reed (1934), and who affirms that the association between Puya and this remarkable insect was known as early as in 1884. In that year M. North, on assignment for the Royal Botanical Garden at Kew, completed a painting entitled "Blue Puya with Moths" in the vicinity of Santiago. This account and an attempt to establish the proper names for Chilean species are in a second paper by Gourlay (195?). His nomenclature has been used in the foregoing as no other taxonomic treatment seems satisfactory.
It is unfortunate that experiments on whether seeds are set by self-pollination have not been widely performed in Puya species; however, according to Johow (1910) and Gourlay (1950) at least two Chilean species are known to require outcrossing for production of viable seeds. Self-sterility is likely to be more widespread in the genus, and thus it is somewhat surprising to find many Puya species living vegetatively for several years before flowering. It would seem that these species, like the "Century Plant," save their reproductive effort for one big occasion; this might involve some risk because when that occasion comes the pollinator might be unavailable, with the result that many years of preparation and much energy would be wasted from the plant's point of view. Perhaps because of this most Puya have adapted to rely on bird rather than insect pollinators, the latter being relatively less mobile and abundant than birds at high elevations.
Dependence on birds could even be of selective value on the trend for increased body size with elevation mentioned previously; the higher a Puya lives the more isolated it becomes relative to alternate nectar sources a bird might use, and thus the plant must cater to a relatively hungrier pollinator by providing greater amounts of nectar; these can presumably be supplied by a large plant better than by a small one.
This explanation is probably a gross oversimplification; attraction of scarce pollinators is just one of the ecological problems that must be faced and solved by Puya plants. However, one can hardly escape thinking along these lines when one considers that these organisms are among the selected few able to survive in high mountains, and, furthermore, among the handful of plants able to reproduce there through the agency of animal pollinators. If animals are required for successful seed set in the lofty giant, Puya raimondii, then the pollinator-attracting mechanism the plant must use has to be very precise. This species is known to include long-lived individuals that go for scores of years without giving rise to either flowers or vegetative shoots; the reproductive effort of these plants comes just once at the end of their lives. Then a massive inflorescence is produced by each including enough flowers to potentially bear as many as 6.5 million seeds; if production of the latter depends on an animal's visiting the inflorescence and bringing pollen from another plant, then survival of Puya raimondii as a species rests on a delicate basis, having so far been achieved by virtue of refined adaptations.
Just what pollinators and which adaptations are used by other Puya has been our subject in this discussion, which should emphasize the need for field research in the Andes to understand better the fascinating ecology of these bromeliads. Even as a zoologist I recognize that they offer exceptional material to explore questions of plant evolution and community structure, and I hope that some answers will be forthcoming through support and interest by people who enjoy bromeliads.
Genus Puya Molina (Pitcairnioideae): Relocation of Several Rare Species and Some Preliminary Remarks on Geographic Distributions and Species Divergence by G.S. Varadarajan in J. Brom. Soc. 38(6): 243 etc 1988 & 39(1):3-7. 1989
INTRODUCTION
Puya, with about 185 species, is the second largest member of the Bromeliaceae subfamily Pitcairnioideae. Its geographic range is widespread throughout the Andes, from Costa Rica to Chile. Puya is perhaps one of the few bromeliads that occur over extensive altitudes, from sea level to nearly 5,000 meters. Despite these outstanding distributional characteristics, only about 40% of the species may be biologically well known judging primarily from the literature, representation in herbaria, and cultivation in botanical gardens. Our knowledge concerning the remaining 60% is often limited to the type collection.
Massive growth habit is an unusual trait in the Bromeliaceae. In Puya, it is very striking and distinctive in a number of species such as P. raimondii Harms, P. chilensis Molina, and P. gouditiana Mez. The gigantic plant form is almost always associated with stiff, spiny foliage, massive, strobilate, cylindrical inflorescences with highly compacted flowers. These attributes are mostly responsible for the difficulties that one encounters while collecting in the field and processing the plants for herbarium. In a number of instances, the material available for a taxon is limited to a single leaf (often without the sheath) and a few flowers. For this reason, the descriptions of the habit, scape, inflorescence type and the like are often far from being complete. These problems are compounded by the narrow distributional ranges of species in high altitude mountains, relatively small population sizes, and highly variable flowering and/or fruiting periods. In my opinion, most, if not all of these factors may have caused the paucity of plant material as well as the under-representation of many puyas in herbaria.
Although "rare species" or "rare taxa" may often be ambiguous phrases, I have narrowed their use to two specific situations: taxa that were poorly collected previously and, therefore, not well represented in herbaria; taxa that are represented by very few individual plants in the natural habitats (as opposed to being abundant). As far as possible, I have attempted to differentiate these conditions through my field explorations. Field studies also aided my assessment of population size, phenological changes, as well as comparisons of specimens represented in herbaria and in natural habitats.
My first exploration in 1983-1984 covered Argentina, Bolivia, and Venezuela and some of the outcomes of these studies appeared in the Journal last year (Varadarajan 1987a, b). My later investigations continued in Ecuador and Chile as well as in several new field areas in Argentina and Bolivia.
OBJECTIVES
I propose to summarize in this article some of my field observations on several rare species of Puya and to present a few preliminary remarks on the geographic distributions from the perspective of phylogeny. Also, I would like to discuss new field localities that are extensions of the previously known geographic range of a few taxa. Some of the inferences presented are tentative and require corroboration.
PUYAS OF ECUADOR
The geography. Two mountain chains, the Cordillera Oriental and Cordillera Occidental, constitute the principal Andean structural units in Ecuador that include Puya. These cordilleras, although near the Peruvian ranges in the south and the Colombian in the north, have their own geologic history, floristic and biogeographic peculiarities. Further, the two cordilleras have some striking differences in their ecosystems, climate, and topography (Gilmartin 1973). With some exceptions, most species of Puya in the Ecuadorian cordilleras are apparently endemic to narrow ranges and valleys in the north-central and south-central regions.
North-central Ecuador. Judging from the specimens in the herbaria, one may find Puya retrosa Gilmartin (formerly known as a part of P. fastuosa Mez) to qualify for the rare species category. In my expeditions, I have encountered P. retrosa in three localities besides the type locality on the ChimborazoThungurahua frontier in central Ecuador: two sites in Pichincha province in Paramo de Huamani at about 3,100 and 3,300 m elevations respectively, and one in Chimborazo province in a subparamo about 3,200 m above Juan de Velasco. Despite these additional collections, the present range of P. retrosa may not be considered extensive by any standard. Populations, with the exception of those in the type locality, are limited to less than fifteen individuals. The plants are 2-3 m tall and grow in rosettes. Massive inflorescences (ca. 75 cm long and 10-15 cm wide) are conspicuously placed well above the level of formidable, spiny foliage and act as firm platforms for the visiting black hummingbirds. The rust-colored scales of the inflorescence and green petals provide a remarkable visual contrast.
South-central Ecuador. Puya sodiroana Mez, previously known from less than ten specimens, readily fits into the rare group. This species is confined to steep slopes of the subparamo localities in Azuay province. Populations of P. sodiroana are discrete, with about fifteen to thirty individuals, and range from 2,500 to 2,800 m elevation. The plants display a number of striking morphological features, e.g., thick stems, 2-3-meter long, compound inflorescences with a series of conical lateral branches, and densely packed flowers. P. sodiroana is also remarkable in that the petals are bright green at anthesis and pale yellow when postfloral.
Azuay province may represent a significant center of species diversity for Puya in Ecuador. Several grass-dominated (Calamogrostis, Stipa) paramos in the northern sectors of Azuay adjoining the provinces of Canar on the west and Morano-Santiago on the east include type localities of some other interesting species. At about 3,000 m elevation, I encountered two rather closely related species: P. compacta L.B. Smith and P. maculata L.B. Smith in northern Azuay. Populations of both species are relatively small and never exceed fifteen individuals. The plants are medium sized (70-80 cm tall), characterized by dense, cylindrical, compound inflorescences with highly reduced lateral branches. Enclosed by a single foliaceous bract and are two to three fertile flowers and one notable pulvinus (a swollen, tumorlike structure). The latter probably represents a reduced flower-producing structure. Woolly hair, frequently noted in great density in the inflorescence of several high altitude puyas such as P. clava-herculis Mez & Sodiro and P. glomerifera Mez & Sodiro, occurs only in a moderate level in P. compacta and P. maculata.
Expeditions to a few shrub-dominated (Yaccinium, Hypericum) paramos, namely, Paramo de Tinajillas and Paramo de Castillo (Azuay province) yielded two more remarkable species at about 3,300 m elevation. These locations may represent a few instances where Puya species are sympatric. P. nutans L.B. Smith and P. pygmaea L.B. Smith both dwarfish, occasional herbs, occur in small, restricted patches along the swampy meadows. In some areas, these species share their habitats with P. glomerifera or with P. clava-herculis. P. nutans is a beautiful species characterized by its decurved, cylindrical inflorescence with highly compacted flowers. Its dark brown papery bracts provide a clear contrast with the dark green flowers that they subtend. Black hummingbirds spotted in these areas visit these colorful inflorescences in particular.
Although neither the flowers nor the entire inflorescence of P. pygmaea is attractive, some of its characteristics are notably distinct. The flowers, tubular at anthesis, are bluish green. They appear to be somewhat buried within the much condensed inflorescence axis in which the dark, stiff and foliaceous, subtending bracts provide adequate mechanical support. Scarcely more than a third of the scape is raised above the level of the foliage. These attributes suggest that pollination is done by an agent different from that of P. nutans.
Range extension of a Colombian puya. Puya trianae Baker was known originally from central Colombia (Cundinamarca dept.) and several collections were made there. I collected this species in the state of Tachira in the Merida Andes of Venezuela in 1987 (Varadarajan 1987a). During a recent exploration I found P. trianae in a grass paramo in the Azuay province with P. compacta and P. clava-herculis at least 2,300-2,500 kilometers south of the Colombian sites. At this time, long distance dispersal is a good explanation of the interrupted distribution of P. trianae in Venezuela, Colombia, and Ecuador.
Summary. The Andes of Ecuador, despite their contiguity with the Peruvian and Colombian ranges, include their own assembly of Puya species that may have radiated locally. Most of the rare species I collected in Ecuador exhibit a relatively small population size and are apparently endemic to single mountain ranges.
PUYAS OF BOLIVIA
Geography. Puya species are widely represented in two regions of the Andes in Bolivia, namely, the Altiplano and the Cordillera Oriental (Varadarajan 1986; 1987b). The Altiplano is a vast expanse of high Andean plateau extending from central Peru to northwestern Argentina. It contains, in large part, a treeless, steppe-like arid vegetation known as puna. The Cordillera Oriental is an extension of the Peruvian structural unit of the Andes. A few transitional regions between the Altiplano and the Cordillera Oriental are also important from the point of view of distributions of Puya. In Bolivia, with the possible exception of P. ferruginea (Ruiz & Pavon) L.B. Smith, the nearly 44 species may readily fit into the rare category. Although I encountered a remarkable number of these puyas in my field trips in Bolivia, the scope of this article limits my reference to a few only.
Species in the Altiplano. Puya raimondii is perhaps the most spectacular species of the Altiplano. Its habit, habitats, and phytosociological features have been discussed elaborately in the past (Foster 1950, Rees and Roe 1980). In the main context of this paper, I would like to stress just one character that may qualify this giant Puya for the rare group although neither the herbarium specimens nor the number of individuals in a stand may justify this idea. Of the two sites in Bolivia from which P. raimondii is presently known, no more than a single individual has been seen in flower or even known to be in any reproductive phase in a given time. This finding is corroborated by several plant explorers, ecologists, and ornithologists who have visited the sites from time to time. For this reason I regard P. raimondii a rare species.
Aside from the giant Puya, the Altiplano in central Bolivia includes sites of several dwarf members including P. humilis Mez, P. leptostachya L.B. Smith, P. tristis, L.B. Smith, P. tunarensis Mez. Populations of these species are more or less evenly distributed in a given area. Their slender inflorescences are barely visible above a rough, rocky, and stony ground layer, while the remainder of the plant body is completely hidden.
Species in the cordillera and transitional regions. Puya brittoniana Mez, P. fosteriana L.B. Smith, and P. mollis Baker ex Mez are a few interesting but poorly known species. These puyas occur in a few restricted sites of the cordillera and the transitional regions between the Altiplano and the cordillera. For the first two taxa, my collections are next only to the type and for P. mollis they may be the sixth or seventh that exist in the herbaria. The natural habitats of these species are found along open, relatively moist, rocky areas within an approximate 75-km radius of La Paz. A large, cylindrical, strobilate inflorescence with dense woolly indument characterizes P. brittoniana, P. fosteriana, and P. mollis. While a simple raceme is evident in P. brittoniana, a compound, bipinnate inflorescence reminiscent of some Ecuadorian species (P. glomerifera, P. retrosa) is found in P. fosteriana and P. mollis.
The type specimen of P. brittoniana was described from fragments only. The plants are nearly 60 cm tall and caulescent. The inflorescence is about 30 cm long, cylindrical, and is borne on a 20-cm long, curved scape. The native habitats of P. brittoniana are found in a semi-dry puna, characterized by various grasses, at about 3,800 m. I collected P. brittoniana from Sorata area in Larecaja province which is less than 15 km from the type locality.
A spectacular photograph of Puya fosteriana appeared recently in this Journal in the reprinted article by M.B. Foster (1950). The specimen I collected was very similar to what one finds in the Foster photograph. Here, I would like to add a couple of points. The petals of P. fosteriana display gradual color changes from blue (very early preanthesis), bluish green (preanthesis), grass green (anthesis), to pinkish purple (postanthesis). These changes are evident even in a single inflorescence. P. fosteriana populations consist of nearly 35 individuals and are sparsely distributed over a moist puna characterized by Stipa and Baccaris community. My collection of P. fosteriana is from the cordillera Tres Cruces near Quime, at about 4,000 m elevation. This collection site is outside the type locality.
Puya mollis was also described only from fragments. The plants are nearly 1.5 m tall and acaulescent. At 75 mm long, 5-8 cm thick, the stiff, erect scape bears a 30-35-cm long cylindrical inflorescence. The bipinnate inflorescence displays dense lateral branches which are clothed with creamy yellow wool. I collected P. mollis from Zongo valley, La Paz, at about 3,300 m elevation. The surrounding vegetation is a moist shrubland characterized by Buddleja and Oreopanax that merges with patchy grass-dominated puna. Unlike P. brittoniana and P. fosteriana, populations of P. mollis occur in small discrete groups over an altitudinal range of 3,000-3,500 m.
In cloud forests of the Cordillera Oriental, some moist rock outcrops often provide suitable habitats for a few species of Puya. In several middle elevation sites (1,500-2,500 m) in La Paz and Cochabamba departments, I encountered very large populations of P. pearcei (Baker) Mez . The plants are 1.5-2 m tall with rosettes of leaves, and a spectacularly reddish scape. A profusely branched particulate inflorescence with at least 30-cm tall, lateral branches is similar to that of P. floccosa (Linden) E. Morren ex Mez. The dark blue preanthesis petals clearly contrast with their dark green color at anthesis and after.
Expeditions to a few, semidry, eastern slopes of the Cordillera Oriental in Santa Cruz department yielded two other dwarf species, Puya nana L.B. Smith and P. tuberosa Mez. They grow on exposed, rocky substrates with shrub-dominated secondary vegetation. Puya nana displays an odd kind of inflorescence that is most common in the pitcairnioid genus Navia and in several members of the Bromelioideae. The scape is very short or may even be completely absent. The sessile inflorescence is basically compound, umbelliform, and consists of numerous highly condensed axes of the lateral branches and their sub-bending involucral bracts. The dark reddish color of the bracts and the lateral branches contrasts remarkably with the dark green foliage on the outside and the bluish green petals within. P. tuberosa has a slender, paniculate inflorescence that is barely visible above the foliage. The primary bracts are serrate and cover nearly half of- the lateral branches. Populations of P. tuberosa are sparsely distributed over a 3-km range, between 1,000-2,200 m elevation, whereas those of P. nana are locally common between 1,700-1,800 m and are notably absent elsewhere.
Some deep canyons of the Cordillera Oriental include "prepunas" (Cabrera 1957, 1971; Sarmiento, 1975) which are semi-arid woodlands with thorny, dicotyledonous species such as Acacia, Capparis, and Prosopis. Open, rocky substrata of the prepunas provide suitable habitats for Puya sanctae-crucis (Baker) L.B. Smith. Plants are nearly 1.5 m tall with laxly particulate inflorescence and the populations are limited to about fifteen individuals in a given site.
Bolivia may be yet another region where the species of Puya have proliferated to a remarkable degree. A number of species here display compound, large, cylindrical inflorescences, Puya hamata-type seeds and a few other traits in common (Varadarajan 1986; Varadarajan & Gilmartin 1988). These taxa, probably differentiated from the same parental stock, are particularly concentrated in the high punas. On the other hand, a species group partly represented in Bolivia share some common features, e.g. a laxly branched paniculate inflorescence, Puya ferruginea-type seed. This lineage has apparently colonized deep canyons and valleys as well as relatively moist, low areas of the Bolivian cordillera, while the putative counterparts occur in Argentina (see below). In attaining the present diversity, a group of dwarf, but phyletically somewhat unrelated species, also seem to have played a key role. These taxa seem to have radiated in arid, semiarid, and mesic places in Bolivia.
Summary. As in other geographic areas, Bolivia includes a number of species that are apparently confined to single mountain ranges or valleys. Yet, some taxa perhaps expanded their range over fairly extensive regions and, occasionally, also became sympatric. A wide array of species assembly, distribution patterns, and morphological traits indicate that puyas of Bolivia have diversified to a degree comparable to various other regions such as Colombia and Peru. The speciation processes, however, seem to have occurred in a regime of intrinsic and extrinsic factors that appear to be unique to Bolivia.
PUYAS OF ARGENTINA
Argentina, particularly its northwestern sector, includes a wide variety of vegetational types and habitats of puya (Varadarajan 1986, 1987b). Despite the practical difficulties involved in visiting the field areas, herbarium collections are moderately available for a number of species as compared with those of Peru and Colombia. Plant explorers, especially Castellanos, Hieronymus, Hunziker, Schreiter, Venturi, have been responsible, in particular, for the extensive field collections.
From the list of rare species I collected, I wish to make a few remarks on P. assurgens L.B. Smith, P. castellanosii L.B. Smith, P. harmsii (Castellanos) Castellanos, P. yakespala Castellanos. These are probably good examples of local endemics, which are also widely separated from their related species.
An apically simple (undivided) and basally branched (pinnate) inflorescence type is seen in P. assurgens and is perhaps unusual, although there are species with predominantly simple inflorescences that produce occasional lateral branches or vice versa (P. meriana Wittmack, P. mirabilis [Mez] L.B. Smith). My collection of P. assurgens is from the Yala River area in Jujuy province, which is its type locality. Smith and Downs (1974) separate in their artificial key P. assurgens from P. dyckioides (Baker) Mez. Closer comparisons reveal that the related species of' P. assurgens are probably in the more general group of dwarf species, some of which are widespread in Bolivia.
Puya castellanosii is the only representative of Puya subgenus Puya in Argentina. There are extensive populations of this species in the Brealito area, near Cachi, Salta province. Its peculiar, compound inflorescence characterized by sterile, terminal portions of the branches identifies it with Chilean species P. berteroniana Mez and P. chilensis. P. castellanosii occurs in a remote and somewhat isolated range separated from its supposed allies by at least a few thousand kilometers. This distribution pattern may suggest that the latter species originated by biogeographic splitting of habitats.
Puya harmsii is a strikingly robust species that grows in dense aggregations along the arid, high punas of Tucuman and Catamarca provinces. At least a few hundreds of individuals constitute the population in a given locality, which is rather unusual for a majority of species. The plants are nearly 2 m tall, with a relatively thick scape and conspicuous primary bracts that cover nearly a third of lateral branches. It resembles several members of a complex of species in Argentina: P. lilloi Castellanos, P. spathacea (Grisebach) Mez. This observation is supported by the gross inflorescence type and foliar anatomy (Varadarajan 1986). However, P. harmsii exhibits a much narrower and somewhat isolated geographic range compared with the other members of the putative complex. The magnitude of distribution of P. harmsii is especially limited to the highly arid punas while its related taxa seem to exhibit a much wider range of tolerances.
Puya yakespala is presently known from the Yakespala puna slopes near Santa Victoria in northwestern Argentina. Some characters of this species are interesting: nearly 3-m tall caulescent plants, a dry scape, more or less fibrous, papery floral bracts, inflorescence decurved and with creamy yellow, woolly indument. Only a few populations were noted over a nearly 45-km range and each had no more than 5 individual plants. The habitats of P. yakespala are rock outcrops in open, moist punas at 4,000 m, characterized by a wide variety of grass species (Calamagrostis, Festuca, Stipa). This simple-inflorescence puya appears to have no closely related member in the nearby region, and P. nivalis Baker, a Colombian species, is probably its only allied taxon.
Most of the species of Puya in Argentina are confined to relatively small, isolated regions, single mountain ranges and valleys. Field and herbarium studies suggest that sympatric species association is rare. Compared with other regions, the species lineages are much less diverse and are geographically rather widely separated from their closely related members in other parts of South America. A species subset including P. lilloi Castellanos, P. micrantha Mez, P. mirabilis, P. pearcei, P. sanctae-crucis, P. smithii Castellanos, P. spathacea (Grisebach) Mez is represented in Argentina and to a limited degree in Bolivia. It is characterized by a laxly paniculate inflorescence, Puya ferruginea-type seed, and a few other common traits (Varadarajan 1986; Varadarajan and Gilmartin 1987, 1988). The diverse species in the lineage appear to have evolved from the same parental stock before their proliferation into the presently known regions. Further, for another group that has an apparent center of primary expansion in Chile (Puya subgen. Puya), the species somewhat remotely distributed in Argentina (P. castellanosii), Peru, and Bolivia (P. raimondii, P. weddelliana (Baker) Mez may represent instances of peripheral radiation and divergence.
PUYAS OF CHILE
Compared with other geographic regions, the proportion of puya in Chile is significantly low. Six species are presently known to be native and a new species, Puya gilmartinii Varadarajan and Flores is being described. All are limited to the hills, sand dunes, and scrubs along the coast, except that P. berteroniana ranges from sea level to the Andean cordilleras (ca. 2,000 m).
Puya gilmartinii appears to be closely related to P. boliviensis Baker and P. chilensis. Several morphological characteristics of P. gilmartinii include a combination of different traits from the allied members, suggesting that it may be a hybrid. In the absence of ploidy level chromosomal data, which often is used to infer the hybrid status of a species, the distributional aspects seem to provide some clues.
A relatively small population of about 15 individuals of P. gilmartinii was found in the type locality, La Serena region, situated between the limits of the ranges of P. boliviensis and P. chilensis. This locality is approximately 200-300 km south of the southern limits of P. boliviensis (Anthofagasta-Atacama region) and at least 300 cm north of the northern limits of P. chilensis (Coquimbo region). In hypothesizing hybridization, it must be emphasized that both P. boliviensis and P. chilensis had much wider distributions in the past and their overlapping sites approximated the La Serena region. Notable absence of the parental populations from that region is probably an instance of local extinction that may have been caused by some minor environmental changes.
Superficial morphological comparisons may suggest an instance of supposed hybrid origin also for P. alpestris (Poeppig) Gay, probably involving parents similar to P. berteroniana and P. chilensis. It is interesting, however, that the present range of distribution of P. alpestris (south-central Chile) is overlapped only by P. chilensis but not by P. berteroniana. Further, in about twenty localities in north-central and central Chile where I found P. berteroniana and P. chilensis co-occuring, I did not encounter populations or individual plants that would conform to any intergrading category.
Summary. The species of Puya in Chile present perhaps the most interesting, well-defined examples suited for detailed investigations of the patterns and processes of speciation. While it would be premature even to speculate on the probably migratory directions, a few general statements may be important to remember. Species native to Chile are unknown from other regions, and no member from any other region seems to exhibit even some peripheral extension into Chile. Pending further explorations of P. alpestris and P. gilmartinii, the population size of every other species in Chile is strikingly large. There are several hundreds of individuals in more than a few localities, unlike the situation in a majority of taxa occurring elsewhere.
ERR.COLL.>100: 185 species, is the second largest member of the Bromeliaceae subfamily Pitcairnioideae. Its geographic range is widespread throughout the Andes, from Costa Rica to Chile. Puya is perhaps one of the few bromeliads that occur over extensive altitudes, from sea level to nearly 5,000 meters. Despite these outstanding distributional characteristics, only about 40% of the species may be biologically well known judging primarily from the literature, representation in herbaria, and cultivation in botanical gardens. Our knowledge concerning the remaining 60% is often limited to the type collection
-Distribution: About 219 species distributed from Costa Rica to Chile and Argentina, principally Andean; seven species in Chile. —SeeZizka et al. 2013
