2

Evolution of Life History in Three High Elevation (), Leah Veldhuisen and Dr. Rachel Jabaily, Organismal Biology & Ecology, Colorado College

The are known as a hotspot for biodiversity and high species endemism for both and animals. Two important tropical, high-elevation ecosystems in the Andes are the puna in , and Chile between 7° and 27° South, and the páramo in Colombia, Venezuela and Ecuador between 11° North and 8° South; both are found at elevations above 3500 meters. The Puya (Bromeliaceae) is found throughout the puna and the páramo, and is relatively under-studied. Life history of most Puya species is largely unknown, with the notable exception of entirely semelparous , which once right before dying and does not reproduce clonally. Other species in the genus do reproduce clonally to varying degrees; their life history strategies have not been defined. Decreased cloning ability in Puya may be evolving convergently as in other groups endemic to high- elevation, tropical ecosystems. We studied three species of Puya (P. raimondii, P. cryptantha and P. goudotiana) across the two ecosystems in Bolivia and Colombia, and collected data on threshold size at flowering and clonal . Data were also analyzed in conjunction with life history theory to hypothesize each species’ life history strategy. All three species were found to have a consistent and predictable minimum size at flowering, while P. cryptantha was found to also have a minimum size for clonal reproduction. No such evidence was found for P. goudotiana. Our data supported that P. raimondii is fully semelparous, and indicated that P. goudotiana and P. cryptantha may be semi-semelparous.

Keywords: Bromeliaceae, evolution, life history, Andes

3

Acknowledgments: I would like to thank all the generous donors who made this project possible: the Keller Family Venture Grants and matching funds, the OBE Department funds, and the JH

Enderson Research Award from Dr. Sharon Smith. Additionally, Dr. Rachel Jabaily, Dr. Jim

Ebersole, Dr. Julián Aguirre-Santorro, Dr. Mary Carolina Garcia Lino and her family, many

Bolivian students and professors, and everyone on Team Colombia who made the field research possible. Also thank you to Kate McGinn for data analysis and R help.

4

Introduction

Life history theory

Semelparity is a life history strategy that is found in a variety of annual plant species, as well as in the animal kingdom, but is rare in long-lived perennial plants. It occurs when an organism has one, large sexual reproductive effort at the end of its life, which results in death

(Young & Augspurger, 1991). The single reproductive effort of semelparity requires a higher one time energetic input than in each of the repeated reproduction efforts of iteroparity, and can result in one higher reproductive output with larger , more flowers, fruits, and seeds

(Augspurger, 1985; Huxman & Loik, 1997; Young & Augspurger, 1991). Semelparity has been well-studied in some species (Barry, Unwin, Malison, & Quinn, 2001; Lazenby-Cohen &

Cockburn, 1988; Young, 1984, 1985), and its evolution has been studied in some high elevation rosette plants such as the Lobelia genus that evolved convergently with other tropical alpine species (Schaffer & Schaffer, 1977, 1979; Schaffer & Rosenzweig, 1977; Young, 1990; Young &

Augspurger, 1991). These studies resulted in three potential models for the selective pressures that favor semelparity.

One model for semelparity evolution defines a tradeoff between the energy devoted to the one sexual reproductive effort and the elimination of energy for future survival and reproduction; all the plant’s transferable resources are dedicated completely to reproduction to maximize reproductive success (Young, 1990; Young & Augspurger, 1991). In this reproductive effort model, the increase in reproductive success must be an ever-increasing function of reproductive effort (Young, 1990). One example of this model is seen with preference: delayed flowering in semelparous species allows for enough of an increase in rosette and inflorescence size 5 to attract significantly more and expand seed set, thus negating the potential benefits of flowering earlier (Inouye & Taylor, 1980).

There are two other models for the evolution of semelparity: the bet-hedging model and the demographic model. The bet-hedging model predicts that semelparity will be selected for in highly variable and unpredictable environments, although some evidence also suggests the opposite (Schaffer & Rosenzweig, 1977; Young & Augspurger, 1991). Unpredictable environments may favor semelparity over iteroparity, as annual and biennial life history (short- lived semelparous taxa) evolves repeatedly in plant lineages in unpredictable environments

(Young & Augspurger, 1991). Long-lived semelparous rosette plants, such as Lobelia telekii, are common in drought-prone areas that also have highly variable temperatures (Young, 1990; Young

& Augspurger, 1991). The demographic model considers the demographic conditions, such as population growth rate and age-class survivorship, when the loss of all future reproduction is made up for by success of one semelparous reproductive output (Young, 1990; Young & Augspurger,

1991). All three models suggest that semelparity will be selected for as the likelihood of repeated reproduction decreases. Iteroparous species also delay reproduction to minimize reproductive cost, but the reproduction is not lethal (Miller, Williams, Jongejans, Brys, & Jacquemyn, 2012).

Size threshold background

Both iteroparous and semelparous plants proceed through life cycle phases when enough energy has been acquired and stored to move forward, particularly into energetically costly processes like sexual and asexual reproduction (Lacey, 1986). While size and age are often closely correlated, some species reproduce based on a size threshold, while others base their life cycle phases on age (Lacey, 1986). Size in particular is thought to be a good predictor of a wide variety of species, and a consistent minimum size at reproduction has been determined in some species 6

(Lacey, 1986; Miller, Williams, Jongejans, Brys, & Jacquemyn, 2012b; Mora, Chaparro, Bonilla,

& Vargas, 2005; Young, 1984).

In semelparous species in particular, increased vegetative size is an important element of reproduction timing, as it has been shown to correlate with reproductive output and associated pollinator preference for larger (Huxman & Loik, 1997; Rocha, Valera, & Eguiarte,

2005). Fecundity increases with size, often leading to plants reproducing at their largest possible size to maximize fitness. This trend is not true, however, for plants where increased size leads to increased mortality, predation or significantly limited growth, such as Lobelia telekii (Bonser &

Aarssen, 2009; Young, 1985). L. telekii faces more herbivory at larger sizes, despite the fact that it is relatively large compared to other iteroparous species in the genus (Young, 1985). Variability in threshold reproductive size within a species can evolve from inconsistency in the environment or resource availability; plants in resource-limited or disturbed habitats will reproduce at half the size of plants in undisturbed habitats with readily available resources (Bonser & Aarssen, 2009).

Because all environments will be at least somewhat variable, plasticity in threshold reproductive size is quite prevalent in almost all plant populations (Bonser & Aarssen, 2009).

Both semelparous Lobelia telekii and iteroparous Lobelia keniensis are high elevation rosette plants on Mount Kenya, and have a minimum size at sexual reproduction (Young, 1990).

Additionally, other annual, biennial and iteroparous plant species also have consistent minimum reproductive size (Kuss, Rees, Ægisdóttir, Ellner, & Stöcklin, 2008; Werner, 1975; Young, 1984).

Similarly, Puya hamata has a relatively uniform rosette diameter of the vegetative body at the time of flowering (Garcia Meneses & Ramsay, 2014). In resource-limited habitats, the minimum size at reproduction will take longer to reach, and result in a longer life span, like that of long-lived semelparous plants (Young, 1984, 1985). For many plants, both the likelihood of survival and 7 flowering increase as plant size also increases (Bonser & Aarssen, 2009; Metcalf, Rose, & Rees,

2003). This trend holds true specifically in Puya dasylirioides with the number of mature fruits growing exponentially as rosette radii increases (Augspurger, 1985). Additionally, plant size alone is a good predictor of life history evolution since growth rate, chance of survival and reproduction, and reproductive output all depend on size (Augspurger, 1985; Metcalf et al., 2003; Werner, 1975).

There is significantly less evidence for a size threshold in clonal or asexual reproduction, and evidence that does exist varies widely among genera and species (Ashmun & Pitelka, 1985; Mora et al., 2005; Schmid, Bazzaz, & Weiner, 1995; Young, 1984).

In addition to size, plant reproduction timing is influenced by several factors. Age, and especially the interaction between age and size, environmental nitrogen levels, and photoperiod, among other things, play a significant role in when plants reproduce, although size alone is the most influential factor in scaling of reproductive output (Schmid et al., 1995; Song, Smith, To,

Millar, & Imaizumi, 2012). Additionally, vernalization in combination with photoperiods and size are the main triggers for flowering in biennial Oenothera erythrosepala. For this species, there is likely a minimum leaf surface area required to sense photoperiod, indicating that size may interact with a number of other environmental factors to induce reproduction (Kachi & Hirose, 1983).

Puna and páramo

The páramo and the puna are both tropical high elevation ecosystems found above tree line in South America, and are home to plant species with a variety of life history strategies. Puna ecosystems are found South of the equator in Peru, Bolivia and Chile between 7° and 27° South, while páramos exist mainly along and just North of the equator in Colombia, Venezuela and

Ecuador between 11° North and 8° South (Baied & Wheeler, 1993; Balslev & Luteyn, 1992). The puna lies between 3600 and 5000 m above sea level, and experiences more seasonality than the 8 páramo due to its distance from the equator (Brush, 1982). Seasons are just wet and dry, although the puna has a significantly shorter wet season than the páramo, and can go up to six months with no rain (Luteyn, 1999). Despite this, the puna has three sections: dry puna, moist puna, and salt puna. The puna in general has low atmospheric humidity and high solar radiation, and is significantly more xeric than the páramo. Moist puna receives 500-1000 mm of rain annually, although the amount decreases from east to west and north to south (Baied & Wheeler, 1993).

Bunchgrasses, small shrubs, trees and herbs are the most common forms of vegetation, and soil is loose with high quantities of soluble salts (Brush, 1982; Morrone, 2001). The puna experiences extreme diurnal temperature fluctuation; there can be up to a 30°C difference between day and night, and nighttime frost is typical throughout the year, which is also true in the páramo (Luteyn,

1999; Smith & Young, 1987). While productivity can be variable throughout the moist puna, it is generally low (Baied & Wheeler, 1993). Wind is also an important element in both the puna and the páramo, and creates distinct microclimates within each ecosystem (Smith & Young, 1987).

The páramo, while also found between 3000 and 5000 m of elevation, is significantly more humid, with much of the moisture coming from rain, clouds and fog. Humidity is usually between

70 and 85%, and rainfall can be more than 3000 mm annually (Balslev & Luteyn, 1992; Luteyn,

1999). Soils in the páramo are humic, acidic and fertile, and vegetation is dominated by bunchgrasses, dwarfed bamboos, shrubs, sedges, and in the northern páramos, giant rosette plants

(Balslev & Luteyn, 1992). The páramo is also the fastest evolving place on earth, with net diversification rates higher than other quickly-evolving ecosystems. Lineages that are diversifying quickly are more likely to be found in the páramo, and the ecosystem has high species endemism

(Madriñán, Cortés, & Richardson, 2013). Both ecosystems experience unpredictable and extreme 9 temperatures and precipitation, and have significantly shaped the evolution of the species that inhabit them.

Puya background

The Puya genus is comprised of roughly 200 species of rosette-forming terrestrial bromeliads that are endemic to Central and South America from Costa Rica to Chile and Argentina

(Jabaily & Sytsma, 2010; Luther, 2008). Puya species are generally found at high elevations from

1500 to 4000 m throughout the Andes, although there is also a group at sea level in central Chile

(Jabaily & Sytsma, 2010). Puya raimondii, the largest species in the genus and in the Bromeliaceae family, is endemic to the moist and dry puna of Peru and Bolivia, while many other species are endemic to the páramos of Colombia and Ecuador. Still many other species are found in singular, small populations throughout South America. Puya species produce wind-dispersed winged seeds, which is common for many species in tropical high elevation ecosystems (Mora, Chaparro, &

Vargas, 2007; A. Smith & Young, 1987). There is significant variation in vegetative rosette and reproductive display size and life history between species, as well as high levels of endemicity throughout the páramo and puna (Hornung-Leoni & Sosa, 2005; Manzanares, 2005). Growth and reproduction of Puya are controlled by the apical meristem, which builds the vegetative rosette body and then converts to an inflorescence for sexual reproduction. Each rosette can only produce one inflorescence; after the resultant seeds are dispersed the axis does not continue to grow and begins to senesce (Bodine, Bush, Capaldi, & Jabaily, submitted.). Many species of Puya reproduce clonally as well as sexually. Clones are produced by axillary meristems (Benzing et al., 2000), which then have their own shoot apical meristem that is capable of its own sexual reproduction.

The ability of different species to clonally reproduce differs, and is often unknown or poorly characterized. 10

Life history strategies for Puya species, as well as other plant groups, are generally characterized as semelparous (reproducing once sexually) or iteroparous (reproducing sexually repeatedly from each interconnected clonal rosette, also known as pups), though no standardization of terminology in the bromeliad scientific community exists (Bodine et al., submitted.). Life history, often portrayed as a binary state (Stearns, 1992), may be more accurately described as a continuum between semelparous and iteroparous (Hughes, 2017). Puya species span this continuum, with some like P. raimondii that are semelparous, some with reduced pupping ability, and many with prolific pupping ability. Jabaily and Sytsma (2013) use the term “semi- semelparous” to describe Puya species that pup only once they hit reproductive age and not all individuals do pup; they place Puya goudotiana in this category based on limited field observations. Species also vary in scaling of their vegetative bodies and sexual displays. The genus is thus a great model for studying adaptations of life history characteristics in the high Andean hotspots of biodiversity.

Study species

Puya raimondii is the largest bromeliad in the world, and relatively well-studied compared to other species in the genus. It is endemic to the puna of southern Peru and Bolivia at high altitudes between 3000 and 4800 m (Sgorbati et al., 2003). Fifty-two additional species of Puya grow in the

Bolivian puna, although most at slightly lower altitudes; very few other species of the genus can be found at the highest range of this distribution (Garcia Lino, 2005). Puya raimondii inhabits dry and moist puna. A P. raimondii inflorescence can be 4-6 m tall (Sgorbati et al., 2003), while many of the lower leaves often senesce or are burned. The inflorescence has potentially hundreds of thousands of white flowers, and the leaves are long and lined with thorns (Hornung-Leoni & Sosa,

2004; Sgorbati et al., 2003). Additionally, populations of the species often live separately from 11 each other in “rodales” that are found across Peru and Bolivia on rocky hillsides in nutrient-poor soils (Augspurger, 1985; Castillo, Baldarrago, Poma, & Raimondo, 2010; Garcia Lino, 2005;

Hornung-Leoni & Sosa, 2005). Like many species, P. raimondii is threatened by human impacts, including fire, livestock, and climate change, and is classified as “endangered” on the IUCN Red

List of Threatened Species (Castillo, Baldarrago, Poma, & Raimondo, 2010; IUCN, 2008). The isolation of Puya raimondii populations has led to low genetic diversity, making it especially vulnerable to human and environmental threats (Sgorbati et al., 2003). As a long-lived, semelparous species, P. raimondii only flowers once in its lifetime, which is estimated to be 80-

150 years in the wild (“San Francisco Botanical Garden - News - Rare in Bloom at SF

Botanical Garden,” 2006). Of all bromeliads, P. raimondii produces the most seeds per inflorescence, and its flowers are likely pollinated by hummingbirds and passerine due to the large quantities of glucose and fructose in the nectar (Hornung-Leoni, González-Gómez, &

Troncoso, 2013; Hornung-Leoni & Sosa, 2004; Hornung-Leoni, Sosa, & López, 2007; Jabaily &

Sytsma, 2013).

Puya raimondii is the only Puya species that is well known enough to be defined as fully semelparous (Jabaily & Sytsma, 2013). While the evolution of semelparity has been relatively thoroughly studied in similar long-lived plants in the Yucca, Agave and Lobelia genera (Schaffer

& Schaffer, 1977, 1979; Smith & Young, 1987; Young, 1990), little is known about the evolution of semelparity in Puya raimondii. Jabaily and Sytsma (2013) hypothesize that semelparity is evolving repeatedly within the Puya genus and only P. raimondii has currently fully evolved this life history strategy. None of the semelparity evolution models have been investigated in the case of Puya raimondii, despite being the only known entirely semelparous Puya species (Garcia

Meneses & Ramsay, 2014). Any model or combination of these models could potentially explain 12 the evolution of P. raimondii, as the puna is quite variable because of its tendency toward drought, and there is no data on the survival or likelihood of reproduction for different age classes. In Yucca and Agave, semelparity been shown to evolve based on the reproductive effort model, which suggests that continuously-increasing reproductive effort is favored when there is an ever- increasing relationship between effort and reproductive success per unit of reproductive effort, which can also be seen in the concave upward curve of the relationship between effort and success

(Schaffer & Schaffer, 1977, 1979; Young, 1990; Young & Augspurger, 1991). In Agave specifically, low pollinator density has been found to select for semelparity, as pollinators tend to prefer larger inflorescences (Schaffer & Schaffer, 1977, 1979). Again, this could be the case for

P. raimondii, since it is only pollinated by hummingbirds and passerine birds, but it has not been tested (Hornung-Leoni et al., 2013). Similarly, increased leaf surface area leads to more viable seeds in semelparous Yucca whipplei whipplei through increased photosynthate production and more available energy for reproduction. This trend was not true for iteroparous Yucca whipplei caespitosa (Huxman & Loik, 1997), and again has not been investigated in P. raimondii.

In addition to P. raimondii, this research focuses on P. cryptantha Cuatrec. and P. goudotiana Mez., species endemic to the páramos of Colombia. Both páramo Puya species in this study are smaller than P. raimondii. Puya goudotiana is one of the largest Puyas, reaching 5 m tall when in flower with leaf lengths of over one meter (Smith & Downs, 1974). To our knowledge, no ecological studies have focused on Puya goudotiana.

Unlike P. raimondii, P. goudotiana and P. cryptantha’s life history strategies have not been clearly defined; both reproduce clonally as well as sexually to some degree but there is no definitive published research. A study by Mora et al. (2005) in the Piedras Gordas section of

Chingaza National Park found that P. cryptantha generally produces one to three pups at a leaf 13 length between 8.8 and 20.8 cm, and usually flowers at a leaf length between 12.0 and 35.7 cm.

Flowering occurs in July and August, the rainiest months in the páramo (Mora et al., 2005).

It is possible that P. goudotiana and P. cryptantha are semi-semelparous, with some individuals flowering without pupping and others pupping before flowering, but baseline data is necessary for confirmation. Also, physiological cues from body scaling for clonal and sexual reproduction in these specific species are unknown. For species that can pup at some level like P. cryptantha, threshold size at reproduction is determined by a tradeoff between the benefits of beginning to reproduce and the cost for future reproduction and survival. Semelparous species such as P. raimondii do not face this same tradeoff (Wesselingh, Klinkhamer, de Jong, & Boorman,

1997). For clonal reproduction, there is only evidence of a minimum size in some species (Ashmun

& Pitelka, 1985; Mora et al., 2005; Schmid et al., 1995; Young, 1984).

Questions and hypotheses

Three different Puya species, P. raimondii, P. goudotiana and P. cryptantha, were studied in two different ecosystems (puna and páramo) for this investigation. The goals of this study were to determine threshold sizes for clonal and sexual reproduction, and to test for evidence of tradeoff in reproductive effort, pup number, and vegetative body size. We also sought to determine if clear life history categories could be determined for all three species, and made applicable to other species, as well. 14

Figure 1: Photos A-C show P. raimondii in the puna of Bolivia. D shows prolific pupping in P. goudotiana in the páramo, E shows post-flowering P. goudotiana, and F shows post-flowering P. cryptantha. A-C, E by Leah Veldhuisen, D, F by Rachel Jabaily.

15

Methods

Bolivia – Puna

Study sites

To collect size and reproductive data on Puya raimondii, we visited four different populations near La Paz and Cochabamba, Bolivia in February 2018. These locations were the only known populations of P. raimondii easily accessible from the cities of La Paz and

Cochabamba. Location One is outside of La Paz, and is adjacent to the tiny town of Comanche at

16.57462°S, 68.25185°W. The population of P. raimondii covers a rocky hill that functions as a quarry and livestock grazing area for local Bolivians, and many of the plants were growing on the sides of nearly-vertical cliffs. Various species of ferns, cacti and grasses were also growing on the hill. Plots ranged from 4069 to 4143 m of elevation. Plot Three at this location was located directly beneath a cliff, and had higher numbers of small individuals than any other plot in the study.

Locations Two and Three, Totora Kasa and Rodeo, are outside of the city of Cochabamba and in the Vacas Municipality. Location Two is at 17.3956°S, 65.3509°W, and Three is nearby at

17.2344°S, 65.2115°W. Both populations cover large rolling and rocky hills. The two locations are on opposite sides of a valley that is divided by a paved highway, and both sides are scattered with small farms. Plot Five at Location Two was partially in a livestock grazing area, and plants seemed to have more fire damage than the other plots. Elevation ranged from 3721 to 3901 m.

Location Four, called Toro Huarko, consisted of only nine individuals and one plot because those nine were the only P. raimondii in the area. The plot was also in the Vacas Municipality, at

17.3643°S, 65.3218°W and 3292 m above sea level. The nine plants were immediately next to a gravel road and across the road was a creek and some small farms. Location Four is significantly 16 lower than the other three, was much wetter and had a wider variety of vegetation than the other three populations.

Data collection

At each of the four populations, we recorded general information including GPS location coordinates, weather, elevation, angle and aspect of slope(s), additional vegetation, and human impacts. Each location was also thoroughly photographed.

We constructed four plots of ten plants in each population using a randomly placed plot center. General plot locations were selected to cover the most space possible across the P. raimondii-inhabited area. The ten plants closest to each plot center were selected to collect data, and the distance from the plot center was recorded. We selected plants based only on distance from the plot center, and not based on size, and each plot was thoroughly searched for small plants. For each plant, we measured six parameters: longest leaf length, longest leaf width, rosette height, rosette width, reproductive category, and distance from plot center. For each plot within a population, we also recorded human impact, elevation, and slope angle and direction. Most of the size data were not collected for dead plants, although dead individuals were included in the plots.

The only information collected for dead plants was rosette height and reproductive category.

All measurements specific to one leaf were taken from the longest leaf, which we selected visually, and we took all size measurements using a 100-m transect tape. To measure the longest leaf length, we placed the tape on the leaf and fed into the center of the rosette until it hit the center.

We measured the width one third of the way in on the same leaf. Rosette width was measured with two people holding the transect tape alongside the plant. We measured rosette height with a clinometer for the plants above 150 cm, and by holding the transect tape next to the plant for individuals below 150 cm. The transect tape was used to measure the distance of each plant from 17 the plot center. We determined reproductive category based on the presence of an inflorescence: pre- and post-flowering.

Colombia – Páramo

Study sites

All study sites were in Chingaza National Park, east of Bogotá, Colombia. Data were collected on the 17th and 18th of October 2018. The first study site was below the La Piedra

Research Station, on a southwest facing slope at 4.3132°N, 73.4616°W and 3234 m of elevation.

Site One had dense shrubs low to the ground, as well as some taller bushes. Puya cryptantha was living in sympatry with P. goudotiana, and P. nitida was on the steeper slopes just above the data collection site. Some areas of Site One were quite wet with Sphagnum throughout, indicating a more acidic site.

Site Two, at an altitude of 3353 m, was located at 4.311°N, 73.4224°W, and was wetter than Site One, with significant standing water. Site Two had P. cryptantha and P. trianae (not studied) living in sympatry, as well as Espeletia.

The third data collection site was adjacent to the Chingaza Visitor’s Center at 4.74365°N,

73.8493722°W, and was also quite wet with standing water.

Data collection

We collected data on rosette size, reproductive category, size and details of inflorescence for Puya goudotiana and P. cryptantha. At Site One, we constructed two 5 by 5 m plots. We selected a random cardinal direction and a random number of steps to walk to determine the location of each plot. All individuals of P. goudotiana were measured within these plots; longest leaf length and width, rosette height and width, reproductive status and number of pups were recorded for all individuals. Pups were determined by proximity to larger rosettes. High numbers 18 of pups were recorded for these individuals, although none were dug up, so we don’t know if these individuals were actually pups or just seedlings in close proximity to each other.

Additionally, at Site One we collected data on post-flowering P. goudotiana individuals.

Individuals were selected based on the presence of an inflorescence to ensure collection of data for individuals that had reproduced. For these plants, we collected data on longest leaf length and width, rosette height and width, inflorescence height, number of fruits and number of pups. At the second site, we collected data on P. cryptantha, but only for living individuals with inflorescences.

For this species, we measured longest leaf length and width, total plant height, inflorescence circumference, number of pups, reproductive state, and number of fruits. We collected data on all plants in the area without constructing plots. We determined reproductive category based on groups, but again we simplified these data to pre- and post-flowering for comparison purposes.

Finally, we collected data on both species at the third site next to the Chingaza Visitor

Center. For P. cryptantha, we took the same measurements as Site Two, and for P. goudotiana, metrics were the same as those for post-flowering individuals at Site One.

Results

Best measures of size

Longest leaf length seems to be the best available predictor of rosette size for P. raimondii, as it is more closely correlated with rosette width than the other size metrics considered, such as longest leaf width or rosette height (Figure 1: R2=0.7135). In other semelparous species, longest leaf length has also been found to be a good indicator of rosette size (Augspurger, 1985; Young,

1984). Additionally, no other variables are as closely correlated as longest leaf length and rosette 19 width, indicating that leaves stop lengthening at a certain point while rosette height continues to increase.

The maximum leaf length for P. raimondii was 177 cm, although there were only three individuals above 150 cm and many between 120 and 150 cm. Only one plant flowered with a leaf length below 120 cm.

Figure 2: Longest leaf lengths of P. raimondii plotted against rosette width. Data indicate a strong correlation between leaf length and overall plant size, especially in smaller plants.

Threshold size at flowering

For P. goudotiana, there is no overlap between sizes of pre- and post-flowering individuals

(Fig. 2). Only plants with leaves longer than 77 cm had flowered, and pre-flowering individuals did not have leaves longer than 53 cm. Pupping had a much lower size threshold; individuals with pups have leaves starting at 17 cm, although it is nearly impossible to tell consistently which individuals are pups without genetic information or digging up the plants. Puya cryptantha also shows a threshold size for sexual reproduction, as none of the flowering individuals had leaves less than 18 cm long, although all measured individuals had flowered. A threshold size for flowering is also found in P. raimondii, where all but one of the post-flowering individuals had 20 leaves longer than 120 cm; there are also many individuals that are above this size that are not yet flowering.

Figure 3: Distribution of leaf length for individuals without inflorescences (pre-flowering) and with inflorescence (post-flowering) for Puya cryptantha, P. goudotiana and P. raimondii.

Threshold for pupping

In P. goudotiana, there seems to be a slight tradeoff between leaf length and number of pups for pre-reproductive individuals, as plants with four or more pups have shorter leaves than those with zero to two pups (Fig. 5). Plants with one or two pups seem to have longer leaves than those with zero.

Figure 4. Longest leaf length plotted with number of pups for P. cryptantha, showing the relationship between number of pups and size of the plant 21

Figure 5. Longest leaf length plotted with number of pups for P. goudotiana, showing the relationship between number of pups and size of the plant Tradeoff between inflorescence height & fecund portion and pups

In P. goudotiana, there seems to be no relationship between the portion of the inflorescence that is fecund, leaf length, number of fruits and number of pups. Longer leaves, or a bigger plant, does not necessarily lead to more pups or more fruits, and neither does a larger fecund portion of the inflorescence.

Tradeoff between pups and fruit number

In P. cryptantha, there is a trend towards longer leaf lengths, more fruits and no pups and, alternatively, shorter leaves, fewer fruits and more pups (Fig. 4). This trend is less clear from

Figure 3, as individuals with one or two pups seem to have longer leaf lengths than individuals with no pups. Despite this, the individuals with four or more pups have shorter leaves than the individuals with fewer pups.

For P. goudotiana, there seems to be no relationship between number of pups, number of fruits and leaf length (Figs. 4, 6). 22

Figure 6. Leaf length of P. cryptantha is plotted against number of fruits and categorized by pup presence. There is a pattern between shorter leaves, fewer fruits and pups, and, alternatively, more fruits, longer leaves and no pups.

Figure 7. Leaf length of P. goudotiana is plotted against number of fruits and categorized by pup presence.

Discussion

Overall, these data indicate a consistent narrow threshold vegetative size for sexual reproduction within all three species of Puya, as well as a threshold size for clonal reproduction in

P. cryptantha. Additionally, they suggest a tradeoff in P. cryptantha between number of pups, leaf length and fruit number; this pattern is not visible in P. goudotiana. The tradeoff for P. goudotiana may also be harder to see since there are fewer measurements of post-reproductive P. goudotiana individuals, and there was no certainty about which plants were indeed pups. 23

Threshold for sexual reproduction for all species

The evidence for a threshold sexual reproductive size in the three studied Puya species differing in overall size, habitat, and life history, fits with a larger pattern of threshold size that has been described previously in other species (Augspurger, 1985; Garcia Meneses & Ramsay, 2014;

Kuss et al., 2008; Mora et al., 2005; Young, 1984). While a minimum size for flowering is an established phenomenon, reasoning for this threshold is more variable across different species and more unknown in general. This threshold may be due to meeting the minimum of stored fixed carbon, or the point when production of resources requires more energy than maintenance by the amount necessary to flower (Young, 1984). Additionally, semelparous species’ inflorescence size is more dependent on rosette size at reproduction than iteroparous species, which would make minimum rosette size particularly important for P. raimondii (Young, 1984). There is no evidence in our data that P. raimondii shows less variation in minimum flowering size than P. cryptantha or P. goudotiana, but multiple studies have found semelparous species’ reproductive output to be more sensitive to size at reproduction than that of iteroparous species (Schaffer & Schaffer, 1977,

1979; Young, 1990).

The lack of variation between the species in this study may also be due to the fact that P. raimondii data were collected across a wider geographic range, or that P. cryptantha and P. goudotiana are currently evolving towards semelparity. Plants with larger flowers and less variable inflorescence size also tend to flower at larger sizes (Schmid et al., 1995), which may be the case for P. raimondii which has large, open flowers like other members of Puya subgenus Puya, in contrast to the typical narrow flowers of most other species of the genus. Semelparous Yucca whipplei whipplei’s reproductive output has been found to be highly responsive to increased photosynthate production from increased leaf surface area, while increased leaf surface area and 24 photosynthate production had no impact on iteroparous Yucca whipplei caespitosa’s reproductive output (Huxman & Loik, 1997). Although P. cryptantha and P. goudotiana are not entirely semelparous, minimum size at flowering is still significant and makes sense with prior studies in other genera (Mora et al., 2005; Young, 1984).

Threshold for clonal reproduction

Our data indicate a size threshold for clonal reproduction in P. cryptantha, similar to Mora et al. (2005), who more extensively studied the species. A minimum size for branching was determined in iteroparous Lobelia keniensis (Young, 1984). Because Lobelia and Puya are comparable in their growth forms and tropical high elevation habitats, a pupping size threshold in

Lobelia could suggest a similar pattern in iteroparous Puya species.

Tradeoffs to maximize reproductive output

Regardless of life history, all plants have evolved and are evolving to maximize reproductive output, partitioning their finite amount of energy towards competing goals of reproduction and individual growth, defense and maintenance. Our data show evidence of this tradeoff in P. cryptantha, as individuals with pups tend to be smaller and produce fewer fruits, while larger individuals have no pups but more fruits. For this tradeoff to benefit the plant, the pups must compensate with sufficient reproductive output to offset the loss from the shorter inflorescence of the mother rosette; larger inflorescences are strongly correlated with higher reproductive success, partially due to increased pollinators and a higher portion of set seeds

(Huxman & Loik, 1997; Inouye & Taylor, 1980; Schaffer & Schaffer, 1977, 1979; Young, 1990).

It is unknown if the pups of P. goudotiana or P. cryptantha flower on their own, which is important when analyzing this potential tradeoff. Iteroparous Lobelia keniensis inflorescence size does not increase with higher soil moisture, but the number of rosettes per individual does, indicating that 25 the clonal pup rosettes are of significant importance to the plant, and are the preferable investment

(Young, 1990).

Categorize life history

Puya raimondii is a well-known semelparous species, as it never pups and dies after flowering (Smith & Downs, 1974). This categorization fits with our data, as P. raimondii individuals that we visited had one terminal inflorescence and did not have pups, and all with inflorescences appeared dead or dying. Puya cryptantha and P. goudotiana both pup as well as flower, but their life history strategies are harder to categorize into just semelparous or iteroparous.

The term “semi-semelparity,” coined by Jabaily and Sytsma (2013), may be the correct categorization for both P. cryptantha and P. goudotiana, as each mother rosette produces one inflorescence and both species have significantly reduced pupping ability compared to low- elevation Puya species and sympatric P. nitida. Additionally, we never saw either species’ pups flower on their own, indicating that pups may have lost the ability to flower while evolving towards true semelparity. It is hard to tell if a flowering rosette is the initial seed-grown mother, or if the flowering rosette was the product of clonal reproduction from a long-decayed mother. This idea contradicts Jabaily and Systma’s (2013) prior conclusion that phylogenetic evidence indicates semi-semelparity is not a transitional stage between iteroparity and full semelparity, and merits further study.

Environment as a factor in life history evolution

While it has been established that P. raimondii is a semelparous species, there is not yet evidence for why this life history evolved in the species (Padilla, 1973; Smith & Downs, 1974).

Neither P. cryptantha nor P. goudotiana’s life history strategies have been officially categorized or evaluated in terms of evolution. From other comparative studies, there is evidence of 26 semelparity existing and evolving in a drier site of one ecosystem, or the drier of two similar ecosystems (Young, 1984). This mirrors the trend found in the three Puya species studied here, as the puna is significantly drier than the páramo and is the home of the only fully semelparous Puya species. Similarly, Young and Ausgsperger’s (1991) bet-hedging model for semelparity evolution suggests that highly variable and unpredictable environments may favor semelparity, especially in habitats prone to unpredictable drought like the puna. This trend offers a plausible explanation for why so many iteroparous and non-fully semelparous species inhabit the páramo, as its rainfall is consistent throughout the year and experiences significantly less seasonality than the puna. Plants that have high survivorship, reproduce frequently and live in wet sites require 13 times more reproductive output than iteroparous species to develop semelparity, while plants in dry sites with low survivorship and less frequent reproduction only require five times the reproductive output to evolve semelparity (Young, 1990). This finding is supported by the fact that semelparous plants are more resource-dependent than their iteroparous counterparts (Huxman & Loik, 1997; Young,

1984, 1990). For semelparous plants, resources are essential to building an inflorescence, which takes longer in the resource-limited environments where large semelparous plants are common

(Smith & Young, 1987; Young, 1990). Giant inflorescences are predicted by Young and

Augsperger’s (1991) reproductive effort model of semelparity, which states that ever-increasing reproductive effort may favor semelparity, as larger inflorescences can be favored by pollinators and thus result in increased reproductive output (Rocha et al., 2005; Schaffer & Schaffer, 1977,

1979). While pollinators have not been found to be major factors in the reproductive effort model

(Young & Augspurger, 1991), their presence and activity may be important in further determining how semelparity evolved in P. raimondii.

27

Future study

While these data address important questions about the evolution and life history of P. raimondii, P. cryptantha and P. goudotiana, there are still many areas for continued investigation.

Specific age-class survivorship data for P. raimondii would allow for evaluation of the demographic, bet-hedging and reproductive efforts models of semelparity evolution. Data on pollinator preference and reproductive effort, output and success would also help to answer this question, and clarify habitat’s role in life history evolution for the Puya genus. Comprehensive data on the environmental conditions of the puna and páramo would also be useful. Finally, reproductive output and success for semelparous and iteroparous species has been quantified in other genera, and would be interesting to analyze for Puya as evidence of the tradeoff between semelparity and iteroparity (Young, 1990).

28

References

Ashmun, J. W., & Pitelka, L. F. (1985). Population Biology of Clintonia borealis: II. Survival and

Growth of Transplanted Ramets in Different Environments. Journal of Ecology, 73(1), 185–198.

https://doi.org/10.2307/2259777

Augspurger, C. K. (1985). Demography and Life History Variation of Puya dasylirioides, a Long-Lived

Rosette in Tropical Subalpine Bogs. Oikos, 45(3), 341–352. https://doi.org/10.2307/3565569

Baied, C. A., & Wheeler, J. C. (1993). Evolution of High Andean Puna Ecosystems: Environment,

Climate, and Culture Change over the Last 12,000 Years in the Central Andes. Mountain

Research and Development, 13(2), 145–156. https://doi.org/10.2307/3673632

Balslev, H., & Luteyn, J. L. (1992). Paramo: An Andean Ecosystem under Human Influence. San Diego,

CA: Academic Press Limited.

Barry, T. P., Unwin, M. J., Malison, J. A., & Quinn, T. P. (2001). Free and total cortisol levels in

semelparous and iteroparous chinook salmon. Journal of Fish Biology, 59(6), 1673–1676.

https://doi.org/10.1111/j.1095-8649.2001.tb00230.x

Benzing, D. H., Bennett, B., Brown, G., Dimmitt, M., Luther, H., Ramirez, I., … Till, W. (2000).

Bromeliaceae: Profile of an Adaptive Radiation (1 edition). Cambridge, UK ; New York, NY,

USA: Cambridge University Press.

Bodine, E.N., C. Bush, A. Capaldi, & R.S. Jabaily. Quantifying Differences in Reproductive Effort

Between Iteroparous and Semelparous Reproductive Strategies in Bromeliaceae (Submitted:

Journal of Theoretical Biology April 2019).

Bonser, S. P., & Aarssen, L. W. (2009). Interpreting reproductive allometry: Individual strategies of

allocation explain size-dependent reproduction in plant populations. Perspectives in Plant

Ecology, Evolution and Systematics, 11(1), 31–40. https://doi.org/10.1016/j.ppees.2008.10.003 29

Brush, S. B. (1982). The Natural and Human Environment of the Central Andes. Mountain Research

and Development, 2(1), 19–38. https://doi.org/10.2307/3672931

Castillo, J. S., Baldarrago, F. C. D., Poma, I., & Raimondo, F. M. (2010). Diagnostico del estado actual

de conservación de Puya raimondii en Arequipa-Perú, 10.

Garcia Lino, M. C. (2005). Estado de conservacion de Puya raimondii Harms en el valle de Araca.

Municipio Cairoma, La Paz - Bolivia. Universidad Mayor de San Andres, La Paz, Bolivia.

Garcia-Meneses, P. M., & Ramsay, P. M. (2014). Puya Hamata Demography as an Indicator of Recent

Fire History in the Paramo of El Angel and Volcan Chiles, Ecuador-Colombia. Caldasia, 36(1),

53–69. https://doi.org/10.15446/caldasia.v36n1.43891

Hornung-Leoni, C., & Sosa, V. (2005). Morphological variation in Puya (Bromeliaceae): an allometric

study. Plant Systematics and Evolution, 256(1–4), 35–53. https://doi.org/10.1007/s00606-005-

0302-z

Hornung-Leoni, C. T., González-Gómez, P. L., & Troncoso, A. J. (2013). Morphology, nectar

characteristics and avian pollinators in five Andean Puya species (Bromeliaceae). Acta

Oecologica, 51, 54–61. https://doi.org/10.1016/j.actao.2013.05.010

Hornung-Leoni, Claudia, & Sosa, V. (2004). Uses in a giant Bromeliad: Puya raimondii. Journal of the

Bromeliad Society, 54, 3–8.

Hornung-Leoni, Claudia, Sosa, V., & López, M. G. (2007). Xylose in the nectar of Puya raimondii

(Bromeliaceae), the Queen of the Puna. Biochemical Systematics and Ecology, 35(8), 554–556.

https://doi.org/10.1016/j.bse.2007.02.001

Hughes, P. W. (2017). Between semelparity and iteroparity: Empirical evidence for a continuum of

modes of parity. Ecology and Evolution, 7(20), 8232–8261. https://doi.org/10.1002/ece3.3341 30

Huxman, T. E., & Loik, M. E. (1997). Reproductive patterns of two varities of Yucca whipplei

(Liliaceae) with different life histories. International Journal of Plant Sciences, 158(6), 778.

Inouye, D. W., & Taylor, O. R. (1980). Variation in generation time in Frasera speciosa

(Gentianaceae), a long-lived perennial monocarp. Oecologia, 47(2), 171–174.

https://doi.org/10.1007/BF00346816

IUCN. (2008). Puya raimondii: Lambe, A.: The IUCN Red List of Threatened Species 2009:

e.T168358A6482345 [Data set]. International Union for Conservation of Nature.

https://doi.org/10.2305/IUCN.UK.2009-2.RLTS.T168358A6482345.en

Jabaily, Rachel S., & Sytsma, K. J. (2013). Historical biogeography and life-history evolution of

Andean Puya (Bromeliaceae). Botanical Journal of the Linnean Society, 171(1), 201–224.

https://doi.org/10.1111/j.1095-8339.2012.01307.x

Jabaily, Rachel Schmidt, & Sytsma, K. J. (2010). Phylogenetics of Puya (Bromeliaceae): Placement,

Major Lineages, and Evolution of Chilean Species. American Journal of , 97(2), 337–356.

Kachi, N., & Hirose, T. (1983). Bolting induction in Oenothera erythrosepala Borbás in relation to

rosette size, vernalization, and photoperiod. Oecologia, 60(1), 6–9.

https://doi.org/10.1007/BF00379312

Kuss, P., Rees, M., Ægisdóttir, H. H., Ellner, S. P., & Stöcklin, J. (2008). Evolutionary demography of

long-lived monocarpic perennials: a time-lagged integral projection model. Journal of Ecology,

96(4), 821–832. https://doi.org/10.1111/j.1365-2745.2008.01374.x

Lacey, E. P. (1986). Onset of reproduction in plants: Size-versus age-dependency. Trends in Ecology &

Evolution, 1(3), 72–75. https://doi.org/10.1016/0169-5347(86)90021-2 31

Lazenby-Cohen, K. A., & Cockburn, A. (1988). Lek promiscuity in a semelparous mammal, Antechinus

stuartii (Marsupialia: Dasyuridae)? Behavioral Ecology and Sociobiology, 22(3), 195–202.

https://doi.org/10.1007/BF00300569

Luteyn, J. (1999). Páramo Ecosystem. Retrieved September 25, 2018, from

http://www.mobot.org/MOBOT/research/paramo_ecosystem/introduction.shtml

Luther, H. (2008). An alphabetical list of bromeliad binomials (Eleventh). Sarasota, Florida: Bromeliad

Society International.

Madriñán, S., Cortés, A. J., & Richardson, J. E. (2013). Páramo is the world’s fastest evolving and

coolest biodiversity hotspot. Frontiers in Genetics, 4. https://doi.org/10.3389/fgene.2013.00192

Manzanares, J. (2005). Jewels of the Jungle, Bromeliaceae of Ecuador, Part II (Vol. 2).

Quito, Ecuador: Imprenta Mariscal.

Metcalf, J. C., Rose, K. E., & Rees, M. (2003). Evolutionary demography of monocarpic perennials.

Trends in Ecology & Evolution, 18(9), 471–480. https://doi.org/10.1016/S0169-5347(03)00162-9

Miller, T. E. X., Williams, J. L., Jongejans, E., Brys, R., & Jacquemyn, H. (2012a). Evolutionary

demography of iteroparous plants: incorporating non-lethal costs of reproduction into integral

projection models. Proceedings of the Royal Society B: Biological Sciences, 279(1739), 2831–

2840. https://doi.org/10.1098/rspb.2012.0326

Miller, T. E. X., Williams, J. L., Jongejans, E., Brys, R., & Jacquemyn, H. (2012b). Evolutionary

demography of iteroparous plants: incorporating non-lethal costs of reproduction into integral

projection models. Proceedings. Biological Sciences, 279(1739), 2831–2840.

https://doi.org/10.1098/rspb.2012.0326

Mora, F., Ángela Chaparro, H., Bonilla, M. A., & Vargas, O. (2005). Rasgos de historia de vida de Puya

cryptantha una bromelia monocárpica perenne. In Estrategias adaptativas de plantas del páramo 32

y del bosque altoandino en la cordillera Oriental de Colombia (pp. 289–306). Universidad

Nacional de Colombia, Facultad de Ciencias Departmento de Biologia.

Mora, Francisco & Angela Chaparro, Hooz & Vargas, Orlando & Bonilla, Maria Argenis. (2007).

Dinámica de la germinación, latencia de semillas y reclutamiento de plántulas en Puya

cryptantha y P. trianae, dos rosetas gigantes de los páramos colombianos. Ecotropicos, 20, 31-

40.

Morrone, Juan. (2001). A formal definition of the biogeographic Paramo-Punan subregion and its

provinces, based mainly on animal taxa. Revista del Museo Argentino de Ciencias Naturales, 3,

1-12. https://doi.org/10.22179/REVMACN.3.105.

Padilla, V. (1973). Bromeliads. Crown Publishers, New York.

Rocha, M., Valera, A., & Eguiarte, L. E. (2005). Reproductive ecology of five sympatric Agave littaea

(Agavaceae) species in central Mexico. American Journal of Botany, 92(8), 1330–1341.

San Francisco Botanical Garden - News - Rare Flower in Bloom at SF Botanical Garden. (2006, July

19). Retrieved August 2, 2018, from https://www.sfbotanicalgarden.org/news/pr/puya-

raimondii.htm

Schaffer, W. M., & Schaffer, M. V. (1977). The adaptive significance of variations in reproductive habit

in the Agavaceae. In Evolutionary Ecology (pp. 261–276). Palgrave, London.

https://doi.org/10.1007/978-1-349-05226-4_22

Schaffer, William M., & Rosenzweig, M. L. (1977). Selection for Optimal Life Histories. II: Multiple

Equilibria and the Evolution of Alternative Reproductive Strategies. Ecology, 58(1), 60–72.

https://doi.org/10.2307/1935108

Schaffer, William M., & Schaffer, M. V. (1979). The Adaptive Significance of Variations in

Reproductive Habit in the Agavaceae II: Pollinator Foraging Behavior and Selection for 33

Increased Reproductive Expenditure. Ecology, 60(5), 1051–1069.

https://doi.org/10.2307/1936872

Schmid, B., Bazzaz, F. A., & Weiner, J. (1995). Size dependency of sexual reproduction and of clonal

growth in two perennial plants. Canadian Journal of Botany, 73(11), 1831–1837.

https://doi.org/10.1139/b95-194

Sgorbati, S., Labra, M., Grugni, E., Barcaccia, G., Galasso, G., Boni, U., … Scannerini, S. (2003). A

Survey of Genetic Diversity and Reproductive Biology of Puya raimondii (Bromeliaceae), the

Endangered Queen of the Andes. Plant Biology, 6(2), 222–230. https://doi.org/10.1055/s-2004-

817802

Smith, A., & Young, T. (1987). Tropical Alpine Plant Ecology. Annual Review of Ecology and

Systematics, 18, 137–158.

Smith, L. B., & Downs, R. J. (1974). Flora Neotropica: Pitcairnioideae (Bromeliaceae). Flora

Neotropica, 14(1), 1–658.

Song, Y. H., Smith, R. W., To, B. J., Millar, A. J., & Imaizumi, T. (2012). FKF1 Conveys Timing

Information for CONSTANS Stabilization in Photoperiodic Flowering. Science, 336(6084),

1045–1049. https://doi.org/10.1126/science.1219644

Stearns, S. C. (1992). The Evolution of Life Histories (1 edition). Oxford ; New York: Oxford University

Press.

Werner, P. A. (1975). Predictions of fate from rosette size in teasel (Dipsacus fullonum L.). Oecologia,

20(3), 197–201. https://doi.org/10.1007/BF00347472

Wesselingh, R. A., Klinkhamer, P. G. L., de Jong, T. J., & Boorman, L. A. (1997). Threshold Size for

Flowering in Different Habitats: Effects of Size- Dependent Growth and Survival. Ecology,

78(7), 2118–2132. https://doi.org/10.2307/2265949 34

Young, T. P. (1984). The Comparative Demography of Semelparous Lobelia telekii and Iteroparous

Lobelia keniensis on Mount Kenya. Journal of Ecology, 72(2), 637–650.

https://doi.org/10.2307/2260073

Young, T. P. (1985). Lobelia telekii Herbivory, Mortality, and Size at Reproduction: Variation with

Growth Rate. Ecology, 66(6), 1879–1883. https://doi.org/10.2307/2937383

Young, T. P. (1990). Evolution of semelparity in Mount Kenya lobelias. Evolutionary Ecology, 4(2),

157–171. https://doi.org/10.1007/BF02270913

Young, T. P., & Augspurger, C. K. (1991). Ecology and evolution of long-lived semelparous plants.

Trends in Ecology & Evolution, 6(9), 285–289. https://doi.org/10.1016/0169-5347(91)90006-J