BIOTROPICA 37(3): 357–363 2005 10.1111/j.1744-7429.2005.00047.x

Pollination Biology of the Exotic Rattleweed Crotalaria retusa L. (Fabaceae) in NE Brazil1

Claudia M. Jacobi2

Departamento de Biologia Geral/ICB, Universidade Federal de Minas Gerais, Brazil Av. Antonio Carlos 6627,31270-901 Belo Horizonte, MG, Brazil

Mauro Ramalho and Maise Silva

IBIO-Universidade Federal da Bahia, Brazil

ABSTRACT

The rattleweed Crotalaria retusa was introduced in Brazil from Africa, and combines a series of characters that have ensured its establishment in NE Brazil. We focusedon its reproductive biology and pollinator behavior to explain its reproductive success. We performed manual pollination and germination experiments, and monitoredthe behavior of C. retusa’s main pollinators in monospecific plots, and in mixed plots where C. retusa occurred together with two congeners, Crotalaria pallida andCrotalaria lanceolata. Crotalaria retusa is self-compatible and capable of automatic selfing. Inbreeding depression was expressed at the level of percent seed germination,but not seed set. Few insects visited the inflorescences. Legitimate pollinators were two large carpenter bees, Xylocopa frontalis and Xylocopa grisescens which, together,accounted for more than 90 percent of the visits. The former foraged on C. retusa exclusively and has low pollen spread potential. The latter flew longer distancesbetween plants and visited fewer flowers per inflorescence, potentially increasing the extent of pollen carryover, but at the risk of increasing heterospecific pollentransfer, because it visits other Crotalaria species during the same foraging bout. The different foraging strategies, allied to morphological disadvantages representedby pollen overlap on X. grisescens’ body, may partially explain the much lower seed germination observed in C. pallida and C. lanceolata. These results are consistentwith the hypothesis that a reduction in flower constancy may significantly depress viable seed set by increasing the chances of self-pollination.

RESUMO

A leguminosa Crotalaria retusa, vulgarmente conhecida como chocalho ou xique-xique, foi introduzida no Brasil pela Africa, e combina uma serie de caracterısticasque facilitaram seu estabelecimento no nordeste do Brasil. Focalizamos aspectos da biologia reprodutiva e do comportamento dos seu polinizadores para explicar seusucesso reprodutivo. Para isto realizamos tratamentos de polinizacao manual e germinacao de sementes, e monitoramos o comportamento dos principais polinizadoresem areas monoespecıficas e em areas onde C. retusa ocorria junto a duas congeneres, C. pallida e C. lanceolata. C. retusa e autocompatıvel e capaz de autofertilizacaoespontanea. Ocorreu depressao endogamica no nıvel de porcentagem de germinacao de sementes, mas nao no numero. Pouco insetos visitaram as inflorescencias. Ospolinizadores legıtimos foram duas abelhas de grande porte, Xylocopa frontalis e X. grisescens que representaram mais de 90 por cento das visitas. O primeiro forrageiaexclusivamente em C. retusa e tem baixo potencial de dispersao polınica. O segundo voa a distancias mais longas e visita menos flores por inflorescencia, potencialmenteaumentando a extensao de dispersao do polen, mas com o risco de ampliar a transferencia de polen heteroespecıfico, pois visita outras especies de Crotalaria durante ovoo. As diferentes estrategias de forrageio, junto com a sobreposicao de polen no corpo de X. grisescens, explica em parte a baixa taxa de germinacao de C. pallida e C.lanceolata. Os resultados sao consistentes com a hipotese de que uma reducao na constancia floral pode deprimir significativamente a producao de sementes viaveis,ao aumentar as chances de autopolinizacao.

Key words: Crotalaria; exotic weed; floral biology; flower constancy; rattleweed; Xylocopa.

THE COLONIZATION ABILITY OF A PLANT SPECIES depends on com-bined aspects of its reproduction and life history attributes. Growthform, longevity, size, mating system, floral allocation strategies, seedset and germination, as well as modes of dispersal, phenotypic plas-ticity, and tolerance to environmental changes are among the mostimportant characteristics correlated with successful establishmentin new environments (Baker 1974). A decisive factor for the successof many of these attributes is the presence of suitable pollinators innew environments, either to effect pollination in the case of self-incompatible plants, or to reduce the levels of inbreeding in self-fertile species. Inbreeding depression is a widespread mechanism toprevent self progeny, reflected in the poor performance frequentlyobserved in the progeny derived from self or close relative crosses(Lande & Schemske 1985, Husband & Schemske 1996, Richards

1 Received 6 February 2003; revision accepted 27 November 2004.2 Corresponding author; e-mail: jacobi@icb.ufmg.br

1997, Husband & Gurney 1998, Culley et al. 1999). In many cir-cumstances, however, self-fertilization is not so strongly prevented,or is even favored by selective forces. The higher the level of uncer-tainty of cross-pollination, the stronger the effect of reproductiveassurance in favor of self-fertilization (Ramsey & Vaughton 1996,Barrett et al. 1997, Petanidou et al. 1998).

The supply of suitable pollination services grants good levelsof both pollen quantity and quality. Pollination services depend onaspects such as the degree of floral constancy of pollinators, a be-havior that is still the focus of ample debate regarding its causes andconsequences, but is generally associated positively with reproduc-tive success (Waser 1986, Kwak & Bergman 1996, McLernon et al.1996, Chittka et al. 1999).

Many legumes (Fabaceae) are among the most widely spreadexotic plants and, besides presenting self-compatibility, they benefitfrom either exotic (Stout et al. 2002) or native pollination services(Etcheverry et al. 2003), although the rates of spread within thefamily vary considerably (Wu et al. 2003). In this study, we will

357

358 Jacobi, Ramalho, and Silva

focus on the causes for the colonization success of a member of thisfamily.

Species of Crotalaria are annual herb native to old-world trop-ics, ranked as pantropical weeds well adapted to colonize openfields, roadsides, cattle pastures, and other human-altered areas(Moore 1978, Polhill 1982, Lewis 1987). They produce hepato-toxic alkaloids (Culvenor & Smith 1957) and are known to be toxicto farm animals (Hooper & Scanlan 1977, Alfonso et al. 1993)and insects (Panizzi & Slanski 1991). Approximately 40 speciesof Crotalaria occur in Brazil, of which ten are exotic, introducedfrom Africa (Oliveira 1992). In the Atlantic coast of NE Brazilthe most abundant exotics are the common rattleweed Crotalariaretusa L., followed by the smooth rattlebox Crotalaria pallida Ait.,and the lanceleaf rattlebox Crotalaria lanceolata E. Mey. The genushas several self-compatible examples (Moore 1978, Almeida 1986,Etcheverry et al. 2003), concurring with the high rates of self-compatibility of Fabaceae (Endress 1996). Seed dispersal is explo-sive and the small reniform seeds are scattered a few meters fromthe mother plant (Almeida 1986), making gene flow highly depen-dent on pollination services. Inflorescences are composed of brightyellow flag-flowers (papilionate flowers), which offer nectar to visi-tors. Pollen is concealed and presented secondarily when legitimatepollinators, typically large bees strong enough to depress the keel,expose the stigma and push out a mass of pollen grains (Endress1996, Westerkamp 1997).

Because of their predominance in disturbed sites and poor per-formance in unaltered environments, exotic Crotalaria may be con-sidered noninvasive colonizers according to Davis and Thompson’s(2000) classification of “invasive” species. However, their successmay vary among populations, and depends on the behavior of na-tive or exotic pollinators they encounter in the new environment.We evaluated the reasons for the colonization success of C. retusawith respect to two less abundant exotic congeners in NE Brazil, C.lanceolata and C. pallida, focusing on reproductive aspects, with em-phasis on the role of visitors. We hypothesized that C. retusa benefitsfrom a better pollen supply, both in quality and quantity, derivedfrom pollinator services, and that this partially explains its higherabundance in the region, given that the taxonomic and ecologicalstatus, and time of introduction in NE Brazil is similar among thethree species.

METHODS

The study took place at the campus of the Federal University ofBahia, NE coast of Brazil (12◦56′ S, 38◦21′ W) from January to May2001. The area is characterized by large patches of open fields andsmall fragments of secondary forest (Queiros 2001). The climate istypically tropical wet and lacks a dry season.

Crotalaria retusa plants grow in patches that flower year round,as a result of five to six asynchronous annual life cycles within thepopulation. Crotalaria retusa pods are approximately 40-mm long,brown, and dry when ripe. They explode easily when manipulated,propelling most of the seeds within 1–2 m of the mother plant.While most areas of the campus are colonized exclusively by this

species, in some areas it grows mixed with two congeners, C. pallidaand C. lanceolata. In the monospecific patches, C. retusa can reachhigh densities (up to 1.0 individual/m2). In most heterospecificareas, any of the other two Crotalaria species reach only a quarter ofthe density of C. retusa. The three species share most of their floralvisitors and their flowering phenologies overlap. Each C. retusainflorescence has typically few (0–4) flowers open at any giventime, a small number compared to the other two species, which cancarry three to four times that amount. Still, C. retusa’s inflorescencesappear more attractive because of the flower’s larger size, in particularthe flag, which is ca 20-mm long, more than double the size thanthe other two species.

To determine the ability to set fruit and seed in the absence ofpollinators, we performed three pollination treatments, each one ina different inflorescence from 52 plants: (1) agamospermy (N = 28buds), by emasculating buds; (2) spontaneous self-pollination (N =34); and (3) manual cross-pollination (N = 38), by emasculatingbuds and smearing the stigma when the flower opened with pollenfrom at least two different plants distant about 100 m from thefocal plant. Other buds and old flowers were removed from allthe treatment inflorescences, which were immediately bagged withfine mesh cloth to prevent visits, and remained covered until seedswere fully developed or the fruits aborted. The ratio spontaneousselfing/hand outcrossing of mean seed number was used as a self-fertility index, following Lloyd and Schoen (1992). In addition,104 unbagged flowers were used as control (free pollination) tocheck for natural abortions. All the seeds produced by the self andcross-pollinated fruits were sown in a greenhouse, kept moist, andmonitored during 50 d.

Stigma receptivity was determined in the field in ten openand preanthesis flowers by testing the reaction (bubble production)of the stigmatic surface to hydrogen peroxide at 3 percent (Dafni1992) with a hand lens. Anther dehiscence in preanthesis flowerswas checked by removing part of the keel.

Natural seed production and viability in the three genera wasestablished by counting seeds from naturally pollinated fruits takenfrom 31–40 plants of each species, as well as ovules of ten buds fromdifferent plants of each species. All seeds from 40 fruits of C. retusa,9 fruits of C. pallida, and 9 fruits of C. lanceolata were sown andallowed to germinate for 60 d.

We devoted 240 h to observe the foraging behavior of flowervisitors of the three species in different parts of the campus, inparticular their success in exposing the reproductive structures. Ob-servations started before anthesis, around 0700 h, and ended atdusk, around 1900 h.

After establishing the role of each visitor, the floral constancyof the effective pollinators, two large carpenter bees (Xylocopa spp.),was monitored in an area where the three species of Crotalaria grewintermingled. In this plot, C. pallida and C. lanceolata reached theirhighest density, which nevertheless was half or less than that of C.retusa. Twenty-six individuals of Xylocopa frontalis (Olivier) and 32of Xylocopa grisescens Lepeletier were followed from the moment theyentered the study plot (approximately 300 m2) until they flew awayfrom the patch. We recorded the number of flowers available andvisited per plant, as well as the sequence of interplant movements.

Pollination of Exotic Rattleweed 359

The length and thorax width was obtained from ten individuals ofeach species, and the region of pollen deposition in their body wasnoted.

To compare potential pollen spread by the two Xylocopa species,we monitored their foraging flight behavior in a large patch (approx-imately 1000 m2) colonized exclusively by C. retusa. We calculatedthe density of floral resources at the time of the observations fromten 4-m2 quadrats, and registered the number of flowers availableand visited by individual bees for all visited plants. Sixteen individ-uals of X. grisescens and 22 of X. frontalis were followed from plantto plant for up to 17 steps (plant switches). To characterize their for-aging behavior, we measured the beeline between visits to successiveplants and the angle formed by these segments. Significant differ-ences in the frequency distributions of these two flight variablesbetween pollinators were evaluated with a Kolmogorov–Smirnovtwo-sample test (Sokal & Rohlf 1981). To compare the expecteddistance flown by each species, we incorporated these data in anequation that estimates the expected net squared displacement aftern segments E(Rn2), as proposed by Kareiva and Shigesada (1983).In this approach, E(Rn2) depends on the mean segment length andon the mean turning angle between segments. Large segments andsmall turning angles result in large displacements (large inertia). As-suming that the organism exhibits the same probability of turningleft or right (independently of the value of the angle), the expectednet squared displacement is

E(R2

n

) = nE (l2) + E (l )2 c1 − c

(n − 1 − c n

1 − c

), (1)

where l is the mean segment length, c is the mean cosine of theturning angle, and n is the number of steps.

RESULTS

The percentage of seeded fruits obtained from the different pollina-tion experiments indicate that C. retusa is self-compatible and capa-ble of spontaneous selfing. No fruits developed from agamospermy,while spontaneous self-pollination had 50 percent success in fruitproduction. In comparison, 82.7 percent of the 104 free-pollinatedflowers reached maturity. The rate of aborted cross-pollinated fruitswas high (only 10 fruits lasted, i.e., 26%). This poor maturationrate was probably the consequence of heavy manipulation duringanther removal from inside the keel. This procedure was mandatorysince we confirmed that stigma receptivity and anther dehiscencemay occur before anthesis.

Seeds of C. retusa are brown, reniform (4.2 ± 0.1 mm, N =10), and small (16.8 ± 2.4 mg, N = 10). There was no difference inthe mean number of seeds per pod produced by either spontaneousautogamy or manual cross-pollination, which was 17.1 ± 3.9 (N =17) and 17.4 ± 7.1 (N = 10), respectively, and these did not differfrom the number of seeds in control pods (17.9 ± 3.1, N = 40). Theself-fertility value was 0.98. There was a striking difference in seedgermination, however, according to their paternity (Fig. 1). Thosesired by cross-pollen had a fast germination response and attained

FIGURE 1. Germination of Crotalaria retusa seeds resulting from natural (N =714), spontaneous self-pollination (N = 291), and manual cross-pollination

(N = 174).

97 percent of germination after 50 d, while those sired by self-pollen reached only 44 percent in the same period. Control seedshad the same germination rate of cross-seeds after 20 d of sowing,but decreased their germination rate after that. Indeed, after theinitial 3 weeks, only cross-seeds showed a substantial increase inpercent germination.

Under natural conditions, 80 percent of the ovules of C. retusadeveloped into seeds, a lower rate than the other two Crotalariaspecies, which have approximately twice the number of ovules perpod (Table 1). Seed germination after 2 mo, however, was differentamong the species, and much higher in C. retusa, resulting in thehighest rate (66%) of ovules that developed into viable seeds.

In the study area C. retusa flowers opened by lifting the flagaround 1000 h and wilted by 1800 h. Two native carpenter bees(Xylocopini) accounted for more than 90 percent of all visitsrecorded (N = 214). These bees visited the inflorescences beforeanthesis, forcing the buds in search of nectar, and remained activeall day. Xylocopa frontalis accounted for 49 percent of the visits,followed closely by X. grisescens, with 44 percent. Other than these,few insects visited the inflorescences. Occasional visitors were the

TABLE 1. Ovule and seed number (mean ± SD), and 60-d percent germination

of Crotalaria retusa and two sympatric congeners. Seeds were

obtained from naturally pollinated fruits.

Germination Ovule number Seed number Percent

C. lanceolata 45.02 ± 2.12 42.29 ± 3.00 14.9

(N = 10) (N = 31) (57/383)

C. pallida 51.60 ± 1.67 48.45 ± 4.15 57.0

(N = 10) (N = 31) (245/430)

C. retusa 22.42 ± 1.52 17.85 ± 3.13 83.0

(N = 10) (N = 40) (593/714)

360 Jacobi, Ramalho, and Silva

digger bee Centris leprieuri (Spinola), a legitimate pollinator, and thestingless bee Trigona spinipes (Fabricius), a robber who perforatedthe base of the corolla to obtain nectar. Butterflies of the familiesHesperiidae and Pieridae behaved as thieves, collecting nectar withtheir proboscis without pushing the keel down.

The two carpenter bees were considered the only effectivepollinators because of their high visiting frequency and behavior inthe inflorescence. Their large size allowed them to manipulate theflower mechanism and expose the reproductive structures. Xylocopafrontalis is 29.9 ± 1.1 mm long, with a thorax width of 12.0 ±0.4 mm (N = 10). X. grisescens is slightly smaller, 28.5 ± 1.8 mmlong and a thorax width of 11.2 ± 0.5 mm (N = 10). Both embracedthe flower when they landed on the keel, and pressed it down whilepushing the flag upward to introduce the proboscis between them,forcing the emergence of the stigma and squirting pollen ontothe thorax. Pollen was deposited around the insertion of the thirdpair of legs in both species. Because of their flower’s smaller size,pollen of the other two rattlebox species is deposited closer to theinsertion of the first pair of legs.

The resources in the monospecific C. retusa plot were of 3.5 ±2.4 flowers/m2 during the observations. The number of flowers per

FIGURE 2. Frequency of interplant distances (a) and turning angles (b) of the foraging paths of 16 Xylocopa frontalis and 22 X. grisescens individuals, obtained from

282 and 178 interplant segments, respectively.

plant varied from zero to 35 (mean = 6.6 ± 5.3). A comparisonof the two Xylocopa species foraging behavior in this area indicatedthat X. grisescens is capable of spreading pollen farther from themother plant than the other species. The first important foragingdifference is that X. grisescens visited fewer flowers per inflores-cence (35%) than X. frontalis (47%), potentially increasing pollencarryover.

Other differences were exposed by interplant foraging parame-ters. Although both bees responded to the dense landscape describedabove by flying short distances between plants (more than 67% ofthe interplant flight distances of both species were within 3 m,Fig. 2a), flight segments larger than 10 m represented only 4 per-cent in X. frontalis but reached 9 percent in X. grisescens. About80 percent of the turning angles between plants in both speciesranged between 0◦ and 90◦, either to the left or right, indicatinga strong forward directionality (Fig. 2b). There were no significantdifferences in the flight behavior of the bee species, when comparingthe distribution of either interplant distances (D = 0.073 < D .05 =0.130) or turning angles (D = 0.100 < D .05 = 0.136). This sug-gests a similar range of pollen spread. However, the potential pollenspread evaluated by the expected net squared displacement after

Pollination of Exotic Rattleweed 361

TABLE 2. Mean interplant foraging parameters used to calculate the expected

squared displacement Rn2 of Xylocopa frontalis and Xylocopa

grisescens, obtained from 282 and 178 interplant segments, respec-

tively. See Figure 2 for more details.

X. frontalis X. grisescens t-test

Mean interplant distance (m) 2.80 3.67 P < 0.05

Mean squared interplant distance (m2) 17.82 37.13 P < 0.05

Mean cosine 0.28 0.27 NS

Expected Rn2 after 15 steps (m2) 326.43 651.93

visits to 15 consecutive plants was much higher for X. grisescens(around 650 m2) than for X. frontalis (326 m2). This difference isexplained by the higher mean interplant flight distance of X. gris-escens (t-test; P < 0.05; df = 271), given that the mean cosine ofthe turning angles was not significantly different between species(t-test; P = 0.37; df = 256, Table 2).

The degree of flower constancy, measured in the heterospecificplot, was markedly different between pollinators (Table 3). Thedensity at the time of the observations was of 15 ± 8 flowers/m2.The mean number of flowers/inflorescence was 7.3 ± 4.1, 5.2 ±3.4, and 2.4 ± 2.0, in C. lanceolata, C. pallida, and C. retusa,respectively. However, the proportion of plants favored a higherfloral density of C. retusa, as mentioned before. All 26 X. frontalisbees approached but gave up visits to species of Crotalaria other thanC. retusa. Therefore, all 575 flowers visited by this bee belongedto C. retusa, resulting in a mean of 2.4 flowers visited per plant.On only one occasion, outside of our study, an individual of X.frontalis was seen visiting a plant of C. pallida. In comparison, ofthe 32 X. grisescens individuals studied, only eight visited C. retusaexclusively, and the rest foraged on the three species of Crotalariaindiscriminately. From a total of 1121 visits to flowers, 47 percentwere made to C. retusa, 41 percent to C. pallida, and 12 percent to C.lanceolata. These percentages are in accordance with the proportionof inflorescences available in the study plot, where C. lanceolata wasunder-represented. The mean number of flowers per plant visitedby X. grisescens was 2.3, 4.2, and 5.3 for C. retusa, C. pallida, and

TABLE 3. Floral constancy: interplant foraging flights among three species of Crotalaria performed by their two most frequent pollinators: Xylocopa frontalis (N = 26)

and Xylocopa grisescens (N = 32). Pairs of letters indicate plant switch, either to the same species or a different one. L: C. lanceolata; P: C. pallida; R: C.

retusa. The total number of flowers visited and their relative percentage are also shown.

Interplant switches Flowers visited

RR RL RP PP PL PR LL LP LR R P L

X. frontalis 240 0 0 0 0 0 0 0 0 575 0 0

(%) 100 0 0 0 0 0 0 0 0 100 0 0

X. grisescens 194 7 26 56 1 34 23 6 6 528 462 131

(%) 55.0 2.0 7.4 15.9 0.3 9.6 6.5 1.7 1.7 47.1 41.2 11.7

C. lanceolata, respectively, reflecting the higher number of availableflowers per inflorescence in the last two species.

DISCUSSION

The high frequency of self-compatibility in annual herbs has beeninterpreted as a response to the degree of uncertainty of pollina-tion during flowering (Endress 1996, Ramsey & Vaughton 1996,Barrett et al. 1997). In this scenario, reproductive assurance in C.retusa is reflected in its ability to spontaneously self-pollinate, con-sidering that its annual life cycle and ample geographical distributionmight expose the species to the risk of temporary absence of pollina-tors. The high rate of spontaneous self-pollination in C. retusa andother congeners (Etcheverry et al. 2003) is probably associated withthe absence of a membrane in the stigma, characteristic of tribeCrotalarieae (Shivanna & Owens 1989). In most self-compatibleFabaceae, however, there is usually no spontaneous germination ofa self-pollen tube, because the stigma is covered by a membranethat needs to be abraded first. This is achieved during the process ofexposing the reproductive organs by a legitimate pollinator (Endress1996).

Although seed set in Crotalaria is granted even in the case ofabsence of visits, and involves delayed selfing, as pointed out for C.micans (Etcheverry et al. 2003), the low germination rate of self-seeds in C. retusa is a clear indication that appropriate pollinatorsare needed to prevent inbreeding depression. Papilionate flowersdemand specific pollinators that combine strength and behavior toexpose the reproductive structures. In our study, only two carpenterbees conformed to these specifications: X. frontalis and X. grisescens.Although these native bees were frequent visitors of C. retusa andits two congeners, these last two showed considerably lower germi-nation rates. A partial explanation for the low seed viability of C.pallida and C. lanceolata is the difference in the floral constancy ofthe bees.

Stigma obstruction by heterospecific pollen may occur in thethree Crotalaria species since X. grisescens is promiscuous and vis-ited the three species during the same feeding bout. This wouldreduce the percentage of available cross-pollen, intensifying the ger-mination of self-pollen tubes and leading to inbreeding depression.

362 Jacobi, Ramalho, and Silva

However, the chances of C. retusa receiving cross-pollen are highercompared to the other two exotics, because the percentage of vis-its from the promiscuous pollinator is diluted with the visits of X.frontalis, which foraged on this plant exclusively. In addition, sinceits flower is larger, C. retusa pollen is spatially segregated, reducingthe chance of stigma contamination.

The advantage of a constant pollinator, however, may not becontinuous over time, since it has been shown that pollinators mayreduce their specificity according to overall resource densities (Kunin1993, Kunin & Iwasa 1996). In the study site, the presence oftwo other Crotalaria species may have enhanced attractiveness andbenefited all species by pollination facilitation, even at the risk ofreduced seed set due to heterospecific pollen transfer, as suggested bytheoretical models (Feldman et al. 2004). Whereas their combinedfloral display may have enhanced attraction of visitors to the threespecies, the behavior of the pollinators suggests it produced a biasthat has favored C. retusa’s chances of outcross progeny.

Another difference in pollinator behavior was related to pollencarryover. Xylocopa grisescens is potentially capable of transportingpollen from a focal plant farther away than X. frontalis, becauseof its larger mean interplant flight distances and lower percentageof flowers visited per plant. Considering the short-distance seeddispersal, a larger area of pollen spread is beneficial because it reducesthe chances of inbreeding as a consequence of mating with siblings,as already pointed out.

The association between large carpenter bees and Crotalaria iscommon. Xylocopa bees constitute a guild of pollinators active allyear round, with ample geographical distribution in the wet trop-ics (Hurd & Moure 1963, Gerling et al. 1989, Minckley 1998).This makes them relatively predictable in space and time. Even so,their visitation rates are probably lower when the density of floralrewards is low. Also, the ample geographical distribution of Crota-laria (Polhill 1982) expose individuals to the risk of a temporaryor local lack of pollinators, in the boundaries of its geographicalor population expansion and, evidently, in regions where they areexotic. In both cases, or during unpredictable drop in pollinationservices, automatic self-pollination increases the chances of seedproduction. The subsequent establishment of C. retusa in the studyarea and, likewise, in other portions of the coastal landscape seemsto depend, fundamentally, on the successful attraction of its mainpollen vector, Xylocopa species. Potentially different roles in the re-productive success of C. retusa were recognized for each pollinator;X. frontalis shows flower constancy favoring outcrossing, but wouldspread pollen over a smaller area, while X. grisescens is capable oflarger pollen carryover favoring genetic variability, but at the risk ofreaching the wrong Crotalaria species.

ACKNOWLEDGMENTS

The authors wish to thank Marcos M.A. Monsao and Camila M.Pigozzo for their help during fieldwork. CMJ thanks CAPES for thefellowship that made possible a 3-mo stay at the Federal University ofBahia. The comments of two anonymous reviewers greatly improvedan early version of the manuscript.

LITERATURE CITED

ALFONSO, H. A., L. M. SANCHEZ, M. D. FIGUEREDO, AND B. C. GOMEZ. 1993.

Intoxication due to Crotalaria retusa and C. spectabilis in chickens and

geese. Vet. Human Toxicol. 35: 539.

ALMEIDA, E. C. 1986. Biologia floral e mecanismo de reproducao em Crotalaria

mucronata Desv. Revta. Ceres 33: 529–540.

BAKER, H. G. 1974. The evolution of weeds. Annu. Rev. Ecol. Syst. 5: 1–24.

BARRETT, S. C. H., L. D. HARDER, AND A. C. WORLEY. 1997. The com-

parative biology of pollination and mating in flowering plants. In J.

Silvertown, M. Franco, and J. L. Harper (Eds.). Plant life histories. Ecol-

ogy, phylogeny and evolution, pp. 57–76. Cambridge University Press,

Cambridge.

CHITTKA, L., J. D. THOMSON, AND N. M. WASER. 1999. Flower constancy, insect

psychology, and plant evolution. Naturwissenschaften 86: 361–377.

CULLEY, T. M., S. G. WELLER, A. K. SAKAI, AND A. E. RANKIN. 1999. Inbreeding

depression and selfing rates in a self-compatible, hermaphroditic species,

Schiedea membranacea (Caryophyllaceae). Am. J. Bot. 86: 980–987.

CULVENOR, C. C. J., AND L. W. SMITH. 1957. The alkaloids of Crotalaria retusa

L. Aust. J. Chem. 10: 464–473.

DAFNI, A. 1992. Pollination ecology: A practical approach. Oxford University

Press, New York.

DAVIS, M. A., AND K. THOMPSON. 2000. Eight ways to be a colonizer; two ways

to be an invader: A proposed nomenclature scheme for invasion ecology.

Bull. Ecol. Soc. Am. 81: 226–230.

ENDRESS, P. K. 1996. Diversity and evolutionary biology of tropical flowers.

Cambridge University Press, Cambridge.

ETCHEVERRY, A. V., J. J. PROTOMASTRO, AND C. WESTERKAMPP. 2003. De-

layed autonomous self-pollination in the colonizer Crotalaria micans

(Fabaceae: Papilionoideae): Structural and functional aspects. Plant Syst.

Evol. 239: 15–28.

FELDMAN, T. S., W. F. MORRIS, AND W. G. WILSON. 2004. When can two plant

species facilitate each other’s pollination? Oikos 105: 197–207.

GERLING, D., H. H. W. VELTHUIS, AND A. HEFETZ. 1989. Bionomics of the

large carpenter bees of the genus Xylocopa. Annu. Rev. Entomol. 34:

163–190.

HOOPER, P. T., AND W. A. SCANLAN. 1977. Crotalaria retusa poisoning of pigs

and poultry. Aust. Vet. J. 53: 109–114.

HURD, P. D. JR, AND J. S. MOURE. 1963. A classification of the large carpenter

bees (Xylocopini) (Hymenoptera, Apoidea). Univ. Calif. Publ. Entomol.

29: 1–365.

HUSBAND, B. C., AND J. E. GURNEY. 1998. Offspring fitness and parental effects

as a function of inbreeding in Epilobium angustifolium (Onagraceae).

Heredity 80: 173–179.

——, AND D. W. SCHEMSKE. 1996. Evolution of the magnitude and timing of

inbreeding depression in plants. Evolution 50: 54–70.

KAREIVA, P. M., AND N. SHIGESADA. 1983. Analyzing insect movement as a

correlated random walk. Oecologia 56: 234–238.

KUNIN, W. 1993. Sex and the single mustard—Population density and pollinator

behavior effects on seed-set. Ecology 74: 2145–2160.

——, AND Y. IWASA. 1996. Pollinator foraging strategies in mixed floral arrays:

Density effects and floral constancy. Theor. Pop. Biol. 49: 232–263.

KWAK, M. M., AND P. BERGMAN. 1996. Early flowers of Bartsia alpina (Scro-

phulariaceae) and the visitation by bumblebees. Acta Bot. Neerl. 45:

355–366.

Pollination of Exotic Rattleweed 363

LANDE, R., AND D. W. SCHEMSKE. 1985. The evolution of self-fertilization and

inbreeding depression in plants. I. Genetic models. Evolution 39: 24–40.

LEWIS, G. P. 1987. Legumes of Bahia. Royal Botanical Gardens, Kew, 369 p.

LLOYD, D. G., AND D. J. SCHOEN. 1992. Self- and cross-fertilization in plants,

I. Functional dimensions. Int. J. Plant. Sci. 153: 358–369.

MCLERNON, S. M., S. D. MURPHY, AND L. W. AARSEN. 1996. Heterospecific

pollen transfer between sympatric species in a midsuccessional old-field

community. Am. J. Bot. 83: 1168–1174.

MINCKLEY, R. L. 1998. A cladistic analysis and classification of the subgenera

and genera of the large carpenter bees, tribe Xylocopini (Hymenoptera:

Apidae). Nat. Hist. Museum 9: 1–47.

MOORE, L. R. 1978. Seed predation in the legume Crotalaria. Oecologia 34:

185–202.

OLIVEIRA, A. L. P. C. 1992. Evolucao cariotıpica no genero Crotalaria L. (Legu-

minosae). Ph.D. Dissertation. Escola Superior de Agricultura Luis de

Queiros, Sao Paulo, Brazil.

PANIZZI, A. R., AND F. SLANSKI. 1991. Suitability of selected legumes and the

effect of nymphal and adult nutrition in the Southern green stink bug

(Hemiptera, Heteroptera, Pentatomidae). J. Econ. Entomol. 84: 103–

113.

PETANIDOU, T., A. C. ELLIS-ADAMS, J. C. M. DEN NIJS, AND J. G. B. OOSTER-

MEIJER. 1998. Pollination ecology of Gentianella uliginosa, a rare annual

of the Dutch coastal dunes. Nord. J. Bot. 18: 537–548.

POLHILL, R. M. 1982. Crotalaria in Africa and Madagascar. A. A. Balkema,

Rotterdam.

QUEIROS, E. P. 2001. A subfamılia Faboideae (Leguminosae) nas restingas da

costa Norte do Estado da Bahia. M.Sc. Thesis. Universidade Federal da

Bahia, Salvador, Brazil.

RICHARDS, A. J. 1997. Plant breeding systems. Chapman & Hall, London.

RAMSEY, M., AND G. VAUGHTON. 1996. Inbreeding depression and pollinator

availability in a partially self-fertile perennial herb Bladfordia grandiflora

(Liliaceae). Oikos 76: 465–474.

SHIVANNA, K. R., AND S. J. OWENS. 1989. Pollen-pistil interactions (Papil-

ionoideae). In C. H. Stirton and J. L. Zaruchi (Eds.). Advances in

legume biology. Monogr. Syst. Bot. Gard. 29, pp. 157–182.

SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry. W. H. Freeman, San Francisco.

STOUT, J. C., A. R. KELLS, AND D. GOULSON. 2002. Pollination of the invasive

exotic shrub Lupinus arboreus (Fabaceae) by introduced bees in Tasmania.

Biol. Conserv. 106: 425–434.

WASER, N. M. 1986. Flower constancy: Definition, cause, and measurement.

Am. Nat. 127: 593–603.

WESTERKAMP, C. 1997. Keel blossoms: Bee flowers with adaptations against

bees. Flora 192: 125–132.

WU, S. H., S. M. CHAW, AND M. REJMANEK. 2003. Naturalized Fabaceae

(Leguminosae) species in Taiwan: The first approximation. Bot. Bull.

Acad. Sin. 44: 59–66.