Pollination of Cacti in the Sonoran Desert

Theodore H Fleming. American Scientist. Volume 88, Issue 5. Sep/Oct 2000.

The Sonoran Desert of northwestern Mexico and southern Arizona is the biologically richest of the world’s deserts. It is filled with many fascinating kinds of plants and animals, but my vote for the creme de la creme among plants has to be the columnar cacti. Anyone thinking about this desert automatically pictures the statuesque saguaro (Carnegiea gigantea) and its relatives. To many of us, these cacti have become the hallmarks of the Sonoran Desert.

For more than a decade, my colleagues, students and I have been studying the pollination biology of four of the seven or eight species of large cacti found in this habitat. In addition to saguaro, these species include cardon (Pachycereus pringlei), organ pipe (Stenocereus thurberi) and senita (Lophocereus schottii). Three of these species (cardon, saguaro and organ pipe) produce large, cream-colored flowers that open at night and contain substantial amounts of nectar and pollen. Their flower morphology and patterns of flower opening and nectar production clearly suggest that they evolved to be pollinated by bats. Senita, in contrast, produces delicate pink flowers that also open at night but often lack nectar. Their flowers are clearly “aimed” at a far different pollinator than a nectar-feeding bat.

Indeed, research conducted in the late 1950s and early 1960s around Tucson, Arizona, indicated that lesser long-nosed bats (Leptonycteris curasoae), white-winged doves (Zenaida asiatica) and honeybees were effective pollinators of saguaro; bats and bees were also effective pollinators of organ pipe. When we began our studies at a site near Bahia Kino in the Mexican state of Sonora, nothing was known about the pollination biology of cardon and senita.

Our initial involvement with this project revolved around a simple observation. Close relatives of these cacti are strictly bat-pollinated in south-central Mexico, whereas bats, insects or birds are effective pollinators of these species in the Sonoran Desert. We might expect cacti to become more generalized in their pollination systems if there were few bats in the Sonoran Desert. In that case, we would also expect competition among columnar cacti for bat pollinators to be high. We decided that the best way to start answering these questions was to determine whether lesser long-nosed bats were less reliable pollinators in the Sonoran Desert, where they are seasonal migrants, than in their southern haunts, where they are year-round residents. We also sought to quantify the relative contribution of nocturnal and diurnal pollinators to fruit set in the three “bat-pollinated” cacti of the Sonoran Desert and to determine yearto-year variation in the abundance of their pollinators.

In the course of answering these questions, we encountered several unexpected twists to the pollination stories, including a major surprise when we began to study the pollination biology of senita. In the end, the pollination biology of these cacti turned out to be much more interesting than we had originally thought.

Flowering Seasons and Competition

The intensity and nature of competition among plants for pollinators are determined by seasonal patterns of flowering. Frequently, plants that rely on the same pollinators avoid interspecific competition and potential hybridization by flowering at different times of the year. To some extent, this is the tactic used by the Sonoran Desert columnar cacti, whose flowering seasons begin in early spring and, in the case of organ pipe and senita, last through midsummer.

Cardon is the first species to begin flowering, often in late March, and is quickly followed by saguaro. Flowering overlap between these two species is extensive, and interspecific competition for bat pollinators is potentially very high. A few individuals of organ pipe begin to flower in April, but the bulk of the population begins flowering in mid-May. Peak flowering in organ pipe occurs about a month and a half after the flower peaks of cardon and saguaro. Senita, which is mothpollinated, begins flowering in midApril and continues through July.

We discovered that interspecific competition for bat pollinators is reduced by another means as well. Although cardon and saguaro have broadly overlapping flowering seasons, these species head off competition by opening and closing their flowers and producing nectar at different times during the day Whereas flowers of cardon and organ pipe (and senita) open right at sunset and usually close well before noon the next day, those of saguaro open one and a half to two hours after sunset and do not close until late afternoon the next day. Peak nectar secretion occurs well before midnight in cardon and organ pipe, whereas nectar production in saguaro has two peaks: one around 2 a.m. and another about two hours after sunrise.

As a result of these differences, saguaro flowers appear to be “aimed” at attracting daytime pollinators, such as birds and bees, rather than at nocturnal pollinators, such as bats. The results of several years of pollinator-exclusion experiments support this hypothesis. Bat visitations account for nearly 90 percent of the fruit set in carrion but only 45 percent or less of fruit set in saguaro. White-winged doves, rather than bats, are the major vertebrate pollinators of saguaro flowers. Bats also account for relatively little fruit set in organ pipe30 percent as opposed to 70 percent from diurnal pollinators. Hummingbirds are the major vertebrate pollinators of organ pipe cactus.

The co-occurrence of three species of columnar cacti with flowers corresponding to a bat pollination syndrome in the Sonoran Desert is probably a relatively recent geological event. Evidence from seeds found in fossil packrat middens indicates that saguaro has resided in northern parts of the Sonoran Desert the longest (for at least 8,500 years) and that cardon is the youngest resident. Where and for how long saguaro, cardon, and organ pipe lived together before moving into the Sonoran Desert is currently unknown. Thus, we do not know where these species evolved the phenological differences that reduce their joint reliance for pollination on a single species of nectar-feeding bat.

One hint about this, however, comes from geographic variation in the timing of peak flowering in organ pipe. Where it co-occurs with cardon and saguaro in coastal Sonora, its flowering peak is in mid-to-late June, which reduces its competition with cardon and saguaro for bat and other pollinator visits. But where it co-occurs only with saguaro in southwestern Arizona, its flowering peak is one month earlier. This variation suggests that, compared to organ pipe, cardon is a superior competitor for bat visits and that selection has favored a later flowering schedule in organ pipe when it co-occurs with cardon. Since the ranges of these two species overlap only in the Sonoran Desert, we tentatively conclude that geographic variation in organ pipe’s flowering schedule is a relatively recent evolutionary event.

Cardon’s Breeding System

During the course of our first pollinator-exclusion experiment, we made our initial unexpected discovery. Previous research with saguaro and organ pipe had revealed that both species are hermaphroditic (that is, their flowers have both male and female sex organs) and self-incompatible. This is the most common breeding system in flowering plants. We began our study assuming that cardon, which is the world’s largest cactus, was, like its relatives, a self-incompatible hermaphrodite. By the end of our first field season, however, we knew this was not the case. Our big clue here was that some of our experimental plants set absolutely no fruit, even though they had produced hundreds of flowers. Hermaphrodites do not normally abort all of their flowers.

By the end of the second field season, we knew that cardon had an unusual breeding system termed trioecyliterally “three houses”—in which two kinds of unisexual plants (males and females) co-occur with bisexual plants (hermaphrodites) in the same population. By definition, male plants do not produce fruits, and through sheer bad luck, we had chosen several males as our experimental plants the first season. When we carefully dissected the flowers of many plants, we discovered that our cardon population actually contained four classes of individuals: females, whose flowers have anthers that lack pollen, but whose ovaries contain ovules; males, whose anthers produce pollen but whose ovaries lack ovules; hermaphrodites, with “perfect” flowers containing both pollen and ovules; and, strangest of all, “neuters,” whose flowers lack both pollen and ovules. In the cardon populations we studied, hermaphrodites, which turned out to be self-compatible, and females were equally common and were about 50 percent more common than males. As expected for plants with absolutely no evolutionary fitness, neuters were very uncommon.

The presence of unisexual individuals in the same population with bisexuals raises at least two intriguing evolutionary questions. Is this trioecious breeding system stable (that is, will it persist through time?), or is it in transition from an ancestral condition of hermaphroditism toward dioecy (populations containing only males and females)? If it is stable, how do males and females manage to persist in populations of self-pollinating hermaphrodites?

General theory and basic intuition tell us that unisexual individuals must be twice as good at their single-sex function as bisexuals, which can transmit their genes to the next generation via pollen and seeds, in order to remain in the evolutionary game. We found that indeed, both males and females have developed a competitive strategy to deal with hermaphrodites. We found that males and females produce on average 1.6 times more pollen and seeds, respectively, than do bisexuals. Males produce significantly more flowers per night and per season, and females produce significantly more fruits and seeds per season than do hermaphrodites. Recent work on several additional populations of cardon in Sonora in collaboration with Francisco Molina of the Universidad Nacional Autonoma de Mexico has revealed a similar pattern: males and females nearly always produce up to three times more pollen and seeds annually than do hermaphrodites.

Despite their apparently lower relative fitness, we have several reasons for believing that hermaphrodites are not slowly going extinct in cardon populations. First, fruit production in hermaphrodites is not always lower than that of females, which are mutant hermaphrodites with a male-sterility gene. In some populations in some years, females and hermaphrodites have nearly equal relative fitnesses. Second, the seeds and seedlings of self-fertilized fruits of hermaphrodites do not exhibit lower fitness than the outcrossed seeds and seedlings of females, as sometimes occurs in the evolution of dioecy (a much more common breeding system in plants than trioecy) from self-compatible hermaphrodites. When we measured the growth and survivorship of cardon seedlings for two years in the field, however, we found that hermaphrodite seedlings (both self-fertilized and outcrossed) generally outperformed those of females.

Finally, a series of hand-pollination experiments revealed that fruit production is higher when flowers of hermaphrodites receive pollen from other hermaphrodites rather than pollen from males. In contrast, pollen source does not appear to affect fruit production in females. Hermaphrodites, therefore, appear to be more selective than females about their mates. This selectivity favors the perpetuation of hermaphrodite genotypes in cardon populations. Our overall conclusion from these studies is that cardon’s trioecious breeding system appears to be evolutionarily stable. All three sex classes will continue to coexist through time.

But the cardon breeding-system story really does not end here. We surveyed sex ratios in cardon populations throughout its geographical range in coastal Sonora and most of Baja California and found the breeding system to vary The system is trioecious in the northern half of its range in Sonora and in the southern two-thirds of its range in Baja California. Elsewhere, it is gynodioecious-females coexist with hermaphrodites. Male plants were missing in the southern half of the range in Sonora and in the northern one-third of the range in Baja California.

What could account for the “missing” males? Theory tells us that mutant hermaphrodites containing a femalesterility gene-in other words, male plants-should have a much harder time becoming established in populations of self-compatible hermaphrodites than do female plants. This is because the initially rare males are in very strong competition with the much more abundant hermaphrodites for access to unfertilized female and hermaphrodite ovules. This competition is even greater when hermaphrodites can self-fertilize, which automatically reduces the number of hermaphrodite ovules available for fertilization by other individuals. Thus, opportunities for the spread of female-sterility genes in hermaphroditic or gynodioecious populations are very limited.

However, it is possible to favor the spread of female-sterility genes in areas where pollinators are abundant. Many pollinators can visit many flowers, and this, coupled with higher levels of pollen production in males, can potentially reduce competition for ovules.

If pollinator abundance has played an important role in the evolution of cardon’s trioecious breeding systems, we might expect trioecious populations to be clustered around major roosts of lesser long-nosed bats, which form maternity colonies containing tens of thousands of individuals at widely scattered locations in the Sonoran Desert in the spring. When we compared the locations of known Leptonycteris roosts with those of trioecious populations of cardon, we found a rather good fit. Most trioecious populations were within 50 kilometers of a known nectar-bat roost, whereas most gynodioecious populations were much farther away from a nectar-bat roost. This geographic fit is not perfect, however, and it is likely that something else (for example, geographic variation in the cytoplasmic factors involved in sex determination) besides the abundance of nectar-feeding bats may be responsible for the striking geographic pattern that we have uncovered.

Early Flowering in Organ Pipe

Unlike cardon and saguaro, whose blooming seasons last for about 10 weeks, organ pipe has a prolonged flowering season lasting at least 15 weeks. As mentioned previously, most individuals of organ pipe begin to flower in mid-May, but some individuals (the same ones each year in our marked population) begin to flower in April during the flowering peaks of cardon and saguaro. Since organ pipe individuals tend to produce many fewer flowers per night than those of cardon and saguaro, it would seem that early-flowering organ pipes are in an extremely precarious situation in terms of competition for bats and other pollinators. In mid-April, for example, cardon flowers outnumber organ pipe flowers per hectare by a factor of 50 to 100. If lesser long-nosed bats are not selective about which cactus flowers they visit (and our flower-choice experiments indicate that they are not), then, more often than not, organ pipe flowers are likely to receive the wrong pollen in April and early May.

What are the consequences for fruit set when organ pipe flowers receive the wrong pollen? Normally, when flowers receive foreign (heterospecific) pollen, they abort and do not set fruit. Natural selection usually strongly penalizes heterospecific matings. This certainly is the case in cardon and saguaro, whose flowers abort when they receive foreign pollen. But this is not the case in organ pipe, as we discovered when we conducted a series of hand-pollination experiments in 1996.

Instead of aborting, a substantial fraction of organ pipe flowers that received only cardon pollen set fruit. We repeated and extended these experiments in 1999 and obtained essentially the same results. Fruit set in flowers that received either cardon or organ pipe pollen was about twice as high as fruit set in open-pollinated (control) flowers. Organ pipe thus appears to be capable of “making babies” with its competitor’s pollen. When we measured the growth rates of fruits that received different kinds of pollen, we found that both open-pollinated fruits and those produced with cardon pollen grew at significantly lower rates than those produced with organ pipe pollen. This result indicates that most organ pipe fruits produced during April and early May are the products of heterospecific pollination.

Organ pipe is not particularly closely related to cardon and saguaro (it belongs in a different subtribe) and does not hybridize with either of these species. Thus, it is highly unlikely that the wrong pollen is actually fertilizing organ pipe ovules. Instead, it appears that the presence of foreign pollen on their stigmas stimulates organ pipe flowers to develop into a fruit with mature seeds via some form of asexual reproduction. This phenomenon is not uncommon in plants, including many of our most important crop plants. Further work is needed to fill in the details of this process in organ pipe, but one thing seems certain. Instead of being disadvantageous, early flowering in this cactus is potentially advantageous because plants can use foreign pollen to increase their annual fruit production.

Early-flowering plants would thus appear to have the best of both worlds. They produce fruit asexually early in the season when conspecific pollen is scarce and then undergo conventional sexual reproduction when such pollen becomes more common than that of other cacti. Whether or not early-flowering plants have higher fitness than later-flowering plants, however, is currently unknown. To answer this question, we will need to determine whether the fitness of asexually produced seeds is the same as that of sexually produced seeds.

Senita’s Pollination System

In the early years of our study, attention was focused on the pollination biology of the three “bat-pollinated” cacti. It was not until 1995 that we began to study the pollination biology of senita. But it only took one night of flower-watching to discover that something special was also going on in this species. Based on its nocturnal flower opening and small flower size, previous cactus workers had assumed that senita was pollinated by hawkmoths. Rather than hawkmoths, however, senita flowers turn out to be pollinated by a small pyralid moth (Upiga virescens), which we have dubbed the “senita moth.” By day these longitudinally striped moths rest among the long spines at the tops of senita branches. Females begin to arrive at flowers as they open at sunset. Once they land on a flower, they rub their abdomens on the pollen-laden anthers. In short order, their long, feather-like abdominal scales become covered with masses of pollen. Then they leave the flower, presumably to fly to another plant in this self-incompatible, hermaphroditic species. Upon arriving at a different flower, females immediately walk to the upright stigma, where they assume a head-down position and carefully rub their pollen-covered abdomen over the stigma.

This brief act of very deliberate “active pollination” is nearly unique in the animal kingdom. Prior to 1995, only two other kinds of pollinators-yucca moths and fig wasps-were known to deposit pollen on flower stigmas actively The vast majority of flower visitors, including lesser long-nosed bats, white-winged doves, honeybees and hawkmoths, are “passive pollinators,” which usually effect pollination as an accidental byproduct of their feeding activities. Not so with active pollinators, which have specialized structures-mouthparts in yucca moths and abdominal scales in senita moths-for collecting and depositing pollen on flower stigmas. So the interaction between the senita cactus and the senita moth has entered the pantheon of ecology and natural history as the third independently evolved case of very specialized pollination.

Anyone familiar with the natural history of the interactions between the yucca plant and the yucca moth or between the fig and the fig wasp knows there is a dark side to these stories. In both cases, females deposit one or more of their eggs into the ovaries of the flowers they pollinate. Their larvae then destroy part of the seed crop their mothers helped produce. In some cases, seed loss to larval predation can be as high as 50 percent, although it is often much lower than this.

Does this same dark side exist in the interaction between senita and senita moth? Of course it does. After a female pollinates a flower, she returns to its petals, where she lays one egg before flying off. The egg hatches after the flower closes the next morning, and in the next few days, the tiny larva crawls to the base of the corolla, where it chews into the swelling ovary and begins to eat ovules and other plant material. It then chews out of the fruit, which is doomed to abort, and tunnels into the cactus branch to pupate. One hundred percent of the fruits attacked by senita larvae die without producing seeds. But, as in the case of the yucca and fig interactions, only a fraction, usually about 30 percent, of all mothpollinated flowers end up aborting because of larval attacks. High mortality rates of eggs and tiny larvae tip the scales of this interaction firmly in favor of the plant.

Our pollinator-exclusion experiments and other observations indicate that female moths produce about three times more fruits and seeds than are destroyed by their larvae. Like the interaction between the yucca and the yucca moth or that of the fig and the fig wasp, the highly specialized interaction between senita and senita moth can confidently be classified as a mutualism rather than as larval predation. Populations (and individuals) of both plants and pollinators clearly benefit from this interaction.

Unreliable Pollinators in the Desert Our research indicates that the pollination systems of cardon, saguaro and organ pipe are more generalized than those of their bat-pollinated relatives. Several species of birds and bees, in addition to lesser long-nosed bats, are effective pollinators of their flowers. Daytime visitors to flowers of tropical columnar cacti are not effective pollinators, either because the stigmas of these flowers lose their receptivity or the flowers actually close before sunrise. Why have the Sonoran Desert cacti expanded their range of pollinators by remaining open and receptive well after sunrise? Alfonso Valiente-Banuet and his associates at the Universidad Nacional Autonoma de Mexico have suggested that generalized pollination is favored at the northern distributional limits of columnar cacti because of year-to-year variation in the abundance (and hence reliability) of migratory nectar-feeding bats. Whenever a plant’s effective pollinators are unreliable in space and time, natural selection should favor traits that increase its range of pollinators or that favor a switch from an unreliable to a reliable pollinator. If daytime flower visitors are effective and more reliable pollinators than nocturnal ones, then selection should favor traits that expose flowers to them.

So the question we must now address in order to understand our observations regarding pollination of the columnar cacti is whether lesser longnosed bats are less reliable pollinators in the Sonoran Desert than in arid tropical habitats. The answer to this question depends on how one chooses to define reliability.

Pollinators can be unreliable for at least three reasons. First, some are dietary generalists and do not restrict their foraging to one or a few flower species. Their fidelity to a particular species depends an the availability of alternative food sources. Or they can be dietary specialists, but their abundance varies widely from year to year, which results in variable fruit set in their preferred food species. Finally, they may be dietary specialists, but their abundance may be chronically low relative to the availability of flowers and/or other potential pollinators.

Based on the results of our studies, the third scenario appears to explain best the Sonoran Desert situation. During spring in the Sonoran Desert, lesser long-nosed bats are feeding specialists. At our study site, they feed at the flowers of only four species of plants: cardon, saguaro, organ pipe and a paniculate agave (Agave subsimplex). Since the three cacti are much more abundant than is A. subsimplex, this bat is effectively a specialist on columnar cactus flowers.

Whereas cactus densities-and, most important, cactus-flower densities-tend to be high during spring, nectar-feeding bat densities tend to be low (or zero) except near maternity roosts. For example, cactus-flower densities at our study site in Mexico and at Organ Pipe Cactus National Monument in southwestern Arizona, where we worked in 1997, were one to two orders of magnitude higher than the densities of lesser long-nosed bats. The energy supply contained in those flowers was three to four times the energy demand of pregnant and lactating bats. Densities of cactus-pollinating birds were an order of magnitude higher than those of Leptonycteris bats. Based on these density differences, birds are likely to be more frequent (and hence more reliable) visitors to cactus flowers than are bats in many parts of the Sonoran Desert. This is especially true for saguaro, whose northern range limit extends at least 100 kilometers north of the range limits of the lesser longnosed bat. In contrast, the range limits of the white-winged dove, a migratory species that is the most effective diurnal pollinator of saguaro during its breeding season, coincide very closely with those of the saguaro cactus. Blair Wolf and Carlos Martinez del Rio of the University of Arizona, among others, have observed that this dove relies heavily on saguaro flowers and fruit for most of its energy and water. It is a common and highly reliable visitor to saguaro cactus flowers.

Our flower and pollinator-density data suggest that the lesser long-nosed bat is an unreliable cactus-flower visitor only because of its low densities relative to those of cactus flowers and other pollinators in much of the Sonoran Desert. According to data published by Alfonso Valiente-Banuet and his coworkers, nectar-feeding bat densities are also low in tropical arid lands such as the Tehuacan Valley of southeastern Mexico, where bats are the exclusive pollinators of many species of columnar cacti. The main difference between this site and the Sonoran Desert is that the match between cactus flower density and bat density appears to be much closer. Compared with cardon and saguaro, columnar cacti in the Tehuacan Valley produce far fewer flowers per night. As a result, despite their low density, bats are able to pollinate nearly 100 percent of the available flowers in Tehuacan, whereas they pollinate only a fraction of the available flowers in the Sonoran Desert.

Given the chronically low density of nectar-feeding bats in the Sonoran Desert, natural selection has favored the evolution of more generalized pollination systems in what were formerly exclusively bat-pollinated cacti. But this situation raises a further question: Why is flower production in cardon and saguaro so much higher than in their Tehuacan relatives? What selective factors have favored the evolution of larger flower crop sizes in certain northern columnar cacti in the face of chronically low densities of nectarfeeding bats? These and other unanswered questions will motivate future studies of these interesting cacti.