xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xlink http:www.w3.org1999xlink xmlns:xsi http:www.w3.org2001XMLSchema-instance
leader 00000nas 2200000Ka 4500
controlfield tag 008 000000c19749999pautr p s 0 0eng d
datafield ind1 8 ind2 024
subfield code a M39-00237
Quinn, Andrew M.
Cambio de color floral y mantenimiento de las flores Viejas en Lantana camara (Verbenaceae)
Floral color change and maintenance of old flowers in Lantana camara (Verbenaceae)
Young flowers of Lantana camara are yellow and positioned in the center of the inflorescence with a ring of older, red flowers around them. This study investigates how changes in flower color and presence/absence of nectar influence robbers and pollinators of L. camara. It was found that a significant proportion of older, red flowers (mean = 20.3 percent) contain nectar. In addition, the majority of yellow and red flowers (estimated 90 percent)
appear to have pollen and are receptive to pollination. Both pollinators and robbers preferred yellow flowers. Pollinators took nectar from a greater proportion of yellow flowers (e.g. 57 percent) versus red flowers (15 percent). In addition, pollinators showed no preference for inflorescences with the characteristic bulls-eye pattern: all yellow inflorescences (AY) that were the same size as normal (N), bulls-eye inflorescences, received more visits from
butterflies (22 versus 10 per hour), and had more nectar removed (only 12.6 percent of AY flowers had nectar versus 37.5 percent of N flowers). Robbers were about ten times more likely to rob inflorescences with only yellow
flowers than inflorescences containing yellow and red flowers (e.g. mean robbery per inflorescence was 3.71 and 0.334, respectively), even when reward and inflorescence size were similar. In addition, a ring of red flowers reduced robbery of yellow flowers by a factor of 37. Robbers may pollinate L. camara, as inflorescences visited only by robbers contained significantly more flowers with pollen on or near the pistil (1.64 flowers/infl) than inflorescences that were not visited (0.364 flowers/infl). L. camara plants may enhance fitness benefits by
prohibiting large-scale robbery from yellow flowers yet provide some reward for robbery in red flowers as a way to assure their pollination.
Las flores jvenes de Lantana camara son amarillas y se posicionan en el centro de la inflorescencia rodeadas por un anillo de flores rojas ms viejas. Este estudio investiga como los cambios en el color de las flores y la presencia/ausencia del nctar influye en la visitacin de los ladrones de nctar y los polinizadores de L. camara. Se encontr que una proporcin significativa de las flores rojas viejas contienen nctar (promedio = 20.3%). Adems, la mayora de las flores rojas y amarillas (alrededor del 90%) parecen tener polen y estar receptivas para la polinizacin. Ambos polinizadores y ladrones de nctar prefieren las flores amarillas. Los polinizadores toman el nctar de las flores amarillas en una mayor proporcin (por ejemplo 57%) en comparacin con las flores rojas (15%). Adems, los polinizadores no mostraron preferencia por las inflorescencias con el patrn de ojo de buey: las inflorescencias completamente amarillas que fueron del mismo tamao que el ojo de buey recibieron ms visitas de las mariposas (22 versus 10 por hora), y ms nctar fue removido (solo el 12.6% de las flores tenan nctar versus a un 37.5% de las flores N). Los ladrones fueron cerca de diez veces ms propensos a robar en inflorescencias con solo flores amarillas que en aquellas inflorescencias que contenan flores rojas y amarillas (por ejemplo, el promedio de robos por inflorescencia fue de 3.71 y 0.334 respectivamente), aun cuando la recompensa y el tamao de la inflorescencia fueron similares. Tambin, el anillo de flores rojas disminuye el robo en flores amarillas por un factor de 37. Los ladrones pueden polinizar a L. camara ya que las inflorescencias visitadas nicamente por ladrones contienen significativamente ms flores con polen en o cerca del pistilo (1.64 flores/inflorescencia) que las inflorescencias que no fueron visitadas (0.364 flores/inflorescencia). L. camara puede aumentar su xito reproductivo prohibiendo los robos a larga escala en las flores amarillas pero dando cierta recompensa a los robos en flores rojas como un mecanismo para asegurar la polinizacin de las mismas.
Text in English.
Pollination by insects
Costa Rica--Puntarenas--Monteverde Zone--Cerro Plano
Polinizado por insectos
Costa Rica--Puntarenas--Zona de Monteverde--Cerro Plano
Tropical Ecology Spring 2011
Monteverde Butterfly Garden
Ecologa Tropical Primavera 2011
Jardin de Mariposas Monteverde (Costa Rica)
t Monteverde Institute : Tropical Ecology
Floral Color Change and Maintenance of Old Flowers in Lantana camara (Verbenaceae ) Andrew M. Quinn Department of Biology, Pomona College ABSTRACT Young flowers of Lantana camara are yellow and positioned in the center of the inflorescence with a ring of older, red flowers around them. This study investigates how changes in flower color and presence/absence of nectar influence robbers and pollinators of L . camara . It was found that a significant pro portion of older, red flowers ( mean = 20 .3 percent ) contain nectar. In addition, the majority of yellow and red flowers (estimated 90 percent) appear to have pollen and are receptive to pollination . Both pollinators and robbers preferre d yellow flowers . Pollinators took nectar from a greater proportion of yellow flowers (e.g. 57 percent) versus red flowers (15 percent). In addition, pollinators showed no preference for inflorescences with the characteris all yel low inflorescences (AY) that were the same size as normal (N), , received more visits from butterflies (22 versus 10 per hour), and had more nectar removed ( only 12.6 percent of AY flowers had nectar versus 37.5 percent of N flowe rs ) . Robbers were about ten times more likely to rob inflorescences with only yellow flowers than inflorescences containing yellow and red flowers ( e . g . mean robbery per inflorescence was 3.71 and 0.334, respectively) , even when reward and inflorescence size were similar . In addition, a ring of red flowers reduced robbery of yellow flowers by a factor of 37. Robbers may pollinate L. camara , as inflorescences visited only by robbers contained significantly more flowers with pollen on o r near the pistil (1.64 flowers /infl ) than inflorescences that were not visited (0.364 flowers /infl ). L. camara plants may enhance fitness benefits by prohibiting large scale robbery from yellow flowers yet provid e some reward for robbery in red flowers as a way to assure their pollination. RESUMEN Las flores jovenes de Lantana camara son amarillas y se posicionan en el centro de la inflorescencia rodeadas por un anillo de flores rojas mÃ¡s viejas. Este estudio investiga como los cambios en el color de l as flores y la presencia/ausencia de nÃ©ctar influencia la visitaciÃ³n por ladrones de nÃ©ctar y polinizadores de L. camara . Se encontrÃ³ una proporciÃ³n significativa de flores rojas viejas conteniendo nÃ©ctar (promedio = 20.3%). AdemÃ¡s, la mayorÃa de las flo res rojas y amarillas (alrededor del 90%) parecen tener polen y estar receptivas para la polinizaciÃ³n. Ambos polinizadores y ladrones de nÃ©ctar prefieren las flores amarillas. Los polinizadores toman nÃ©ctar en una mayor proporciÃ³n de las flores amarillas (p.e. 57%) contra las flores rojas (15%). AdemÃ¡s, los visitas por mariposas (22 contra 10 por hora), y mÃ¡s nÃ©ctar fue removido (12.6% de flores tienen nÃ©ctar contra 37.5%). Los ladrones fueron cerca de diez veces mÃ¡s propensos a robar en inflorescencias con solo flores amarillas que en aquellas inflorescen cias que contenÃan flores rojas y amarillas (p.e. promedio de robos por inflorescencia fue de 3.71 y 0.334 respectivamente), aÃºn cuando el premio y el tamaÃ±o es similar. TambiÃ©n, el anillo de flores rojas disminuye el robo en flores amarillas por un facto r de 37. Los ladrones pueden polinizar L. camara ya que las inflorescencias visitadas Ãºnicamente por ladrones contienen significativamente mÃ¡s flores con polen en o cerca del pistilo (1.64 flores/inflorescencia) que las inflorescencias que no fueron visit adas (0.364 flores/inflorescencia). L. camara puede aumentar su Ã©xito reproductivo prohibiendo robos a larga escala en las flores amarillas pero dando cierta recompensa a los robos en flores rojas como un mecanismo para asegura la polinizaciÃ³n de las mism as.
F LORAL TRAITS SHOULD MAXIMIZE THE LIKELIH OOD that visitors will pollinate while minimizing robbery (Weiss 1995, Maloof and Inouye 2000). In some plants, flowers remain on the inflorescence and change color as they age (Weiss 1991, Weiss 1995). This may increase attractiveness to pollinators (Weiss 1991, Weiss 1995) and/or protect younger, not yet pollinated flowers from nectar robbers (Barrows 1976, Maloof and Inouye 2000). R etention of post color change flowers around the outside of an infloresc ence may augment its attractiveness to pollinators through color cues or by increasing its size (Barrows 1976, Weiss 1991, Weiss 1995). Both of these possibilities have been considered in the tropical woody shrub, Lantana camara. This plant has yellow, orange, and red flowers (Barrows 1976). Yellow flowers are produced first in the center of the inflorescence and then subsequentl y darken to orange and then red , creating a characteristic pattern (Barrows 1976). These f lowers cannot self poll inate (Barrows 1976). S ince flowers that have changed color are assumed to have little or no nectar or pollen , color change s direct pollinators to more rewarding and viable younger flowers within an inflorescence (Barrows 1976, Gori 1989, Weiss 1991, Weis s 1995). In one survey of 97 butterfly visits, butterflies went to yellow over orange red fl owers 99 percent of the time (Ba rrows 1976). In addition, pollinators may be particularly attracted to the pattern on L. camara inflorescences. One s tudy found that pollinators prefer normal inflorescences (with the pattern) over inflorescences with only yellow flowers (Zurinskas 2003). However, this study did not control for inflorescence size: yellow inflorescences were smaller than norm al inflorescences (Zurinskas 2003). Size may be an important consideration; it has been found that butterflies ch o o se lar ger over smaller inflorescences independent of the level of nectar re ward offered or t he color of the inflorescence (Weiss 1991). Thu s, m ore visits to normal inflorescences could have been solely due to larger size. T he larger size of L. camara inflorescences also allows butterflies with longer body lengths to serve as pollinators, as compared with the smaller inflorescences of L. trifola (Schemske 1976). This is because older flowers are a landing platform that allows larger butterflies to holdfast as they forage on the interior flowers (Schemske 1976). While the retention of older flowers may create a physical barrier to robbers, t he impact of floral color change on these organisms is relatively unknown. Barrows (1976) reported, after examining a limited sample of flowers, that a ring of older orange and red flowers on L. camara inflorescences reduce d nectar robbing of the interi or yellow flowers from 71 to 23 percent . N ectar robbers may show a preference for younger pre change flowers, as do pollinators (Weiss, 1991) . These robber interactions are important to consider as robbers may decrease the fitness of the plant, though th ey occasionally pollinate as well (Kendell and Smith 1976, Maloof and Inouye 2000, Richardson 2004) . A review of 18 studies found that negative, neutral, and positive net effect s of robbery w ere equal ly likely (Maloof and Inouye 2000). I examined floral color changes in L. camara inflorescences in regard to p ollinator and robber behavior. Many of the hypotheses for how these changes impact pollinators and robbers are based on assumptions (older flowers have no nectar), conflations (size with pattern), a limited sample (reduction of robbery in yellow flowers), or remain largely untested (robbers as pollinators). I intended to resolve these shortcomings by considering whether: 1) older flowers have no nectar and are reproductively unviable , 2) older flowers block yellow flowers from robbery , 3) floral color change is a cue to r obbers as well as pollinators, 4 ) coloration attracts pollin ators independent of size, and 5 ) robbers can serve as pollinators. Answering these questions is essential in fully understanding the functionality of color changing
flowers in L. camara : how these changes relate to pollinator and robber attraction, and whether the robbers are really only robbers. METHODS The study was conducted in Monteverde, Costa Rica, in April 2011. Study sites were a small field located across from the Monteverde Butterfly Garden that contained approximately 50 small L. camara plants and a garden at the Cloud Forest Sc hool that contained sev eral large L. camara plants. Both sites were located in Pacific slope Premontane Moist habitats. In the following subheadings, study sites ( Field and Garden ) are indicated in parenthesis. P LANT PHENOLOGY : COLOR , REWARD , AND REPROD UCTIVE POTENTIAL ( F IELD ) . To investigate the timing of changes in color and reward in L. camara , seven actively flowering inflorescences with unopened flowers were covered with a mesh bag (day 1) to exclude robbers and pollinators. The number of flowers of each type (unopened, yellow, orange, red) was recorded the next day (day 2) at 0630, 09 30, 1230, 1530, and 1830, and then at 1230 for the next four days. To examine the reproductive phenology of L. camara , five inflorescences that had not be en previously touched w ere collected and examined under a dissecting microscope. Flowers were dissected using a razorblade. The presence of pollen (both on the anthers and near the pistil) was recorded. In addition, the appearance of the pistil whether it was moist, erect, and generally healthy looking, was noted . N ECTAR CONSUMPTION AN D VISITATION ( FIELD ) . T o see whether orange and red flowers contain nectar and to assess whether pollinators prefer yellow flowers over orange and red, 20 inflorescences were bagged on 14 plant s one evening. In addition, 20 inflorescences on the same 14 plants were marked with a piece of masking tape but otherwise left untouched. Care was taken so that every bagged inflorescence had an unbagged companion on the same plant. All inflorescences were normal (N inflorescences), meaning they contained yellow, orange, and red flowers. In addition, inflorescences were of similar size and condition. The next day these inflorescences were examined at 0630 (dawn), 0930 , 1230, 1530, and 1830 (dusk). This experiment was conducted throughout the day because the schedule of nectar production in L. camara is unknown and to assure that nectar not taken was not reabsorbed throughout the day. At each of the five time periods , one bagged a nd unbagged inflorescence on four plants (total of 16 inflorescences) was examined in detail . The number of flowers of each color as well as the number of flowers with nectar were recorded for each inflorescence. The presence of nectar was detected by ge ntly removing the flower from its calyx and firmly pressing the base of the flower between two fingers. If a visible drop appeared, the flower was considered to offer a nectar reward. This method follows Barrows (1976). To test whether butterflies prefe r inflorescences with coloration independent of size, I compared nectar in N inflorescences to all yellow inflorescences (AY inflorescences) of the same size. AY inflorescences were found on two plants in the field, and observation revealed th at they contained approximately the same distribution of nectar as N inflorescences (i.e. flowers on the edge were less likely to contain nectar). Thus, these inflorescences appeared to be identical to N inflorescences, with the exception that all flowers were yellow. F our AY inflorescences were bagged along with four inf lorescences from normal plants in close proximity. Four additional AY and N inflorescences on the sa me plants were marked but
otherwise untouched . The next day at 1230, the number of fl owers and the number with nectar on each inflorescence were recorded. To compliment the above experiment , visitation for N and AY inflorescences was compared. Ten AY and 10 N inflorescences in close proximity were watched for an hour. A visit was a bu tterfly landing on an inflorescence. R OBBERY ( FIELD AND C LOUD F OREST S CHOOL ) . T o see whether old flowers block robbery and color change serves as a cue for robbers, 32 N inflo rescences and 15 AY inflorescences were marked on 17 plants across from the butterfly garden on a sunny morning at 0630. O range and red flowers were removed on 15 of the 32 N inflorescences to create 15 yellow inflorescences (Y inflorescences). This left 17 N inflorescences. All inflorescence s were examined for robbery and robbed flowers were removed . Robbery was readily apparent as a small hole at the base of the corolla. After robbery was removed , N and AY inflorescences contained 20 30 flowers and Y inflorescences contained 8 15 flowers. That evening, at 1600, the same inflorescences were checked for robbery tha t had accumulated that day. The amount of robbery and the color of the flower robbed were recorded. This same comparison was repeated at the Garden . At 0630, approximately 40 N i nflorescences (the actual number varied between 18 and 24) on three plants with seemingly equal amount of robbery were marked. From half, orange and red flowers were removed to create 20 Y inflorescences. Robbed flowers were removed. Flower ranges on in florescences were the same as above . At 1600 , these inflorescences were examined for robbery in the same way as before. To erase possible effects of differential nectar quantity in flowers of different color/age, the previous exp eriment was repeated with 10 N, 10 Y, and 10 added nectar (AN) inflorescences on the same three plants . AN inflorescences were created by adding nectar to orange and red flowers on normal inflorescences with a syringe. Nectar added was a 30 percent sugar solution, which matched the sugar content of L. camara flowers as previously measured with a refractometer. To isolate the effect of size on robber preference , the experiment was repeated with 10 Y, and 10 red inflorescences (R inflorescences ) on each of the three plants. R inflorescences contained only red flowers and were created by removing enough flowers on inflorescences with only red flowers to have a range of 8 12 flowers. T he range of the Y inflorescences was reduced from the usu al 8 15 flowers to 8 12 flowers to match that of R inflorescences . Finally, to assess robbery on the AY inflorescence s while controlling for per plant differences , nine AY inflore scences were picked from across the butterfly garden, placed in water, and brought to the Cl oud Forest School. These were placed in small plastic vials fille d with water and taped to Plant 1 in the garden (see results) at 0630. At the same time, 11 N and 10 Y inflorescences (usual flower ranges) were marked on the same plant. Robbed flowers were removed, and all inflorescences were examined at 1600 with the same methods as always. This experiment was conducted because I had previously observed (unexpectedly) that AY inflorescences received a very small amount of robbery, and wished to invest igate whether this was an inherent property of the inflorescences. P OLLINATION (C LOUD F OREST S CHOOL ) . At 0630, 11 normal inflorescences on the usual three plants in the Cloud Forest School were checked for robbery, robbed flowers were removed, and the inf lorescences were bag ged so that neither pollinators nor robbers could reach them. At the
same time, 14 normal inflorescences were subjected to the same treatment, and then the opening of the flower was s ealed with a small drop of glue to prevent pollinato rs (but not robbers) from accessing them. Fourteen additional inflorescences were subjected to the same treatment but left as is. Later that day, at 1530, all inflorescences were collected and examined for pollen deposition around the pistil under a diss ecting microscope. Pollen was easily seen once flowers had been removed from the base of the inflorescence, leaving behind the pistil. The number of flowers with pollen deposited around or on the pistil per inflorescence was recorded. RESULTS P LANT PHENOLOGY : COLOR , PO SITION , REWARD , AND REPRODUCTIVE POT ENTIAL . N ew flowers began opening at dawn and the majority (92 percent of 84 flowers examined) were open by 0800 (figure 1). All of the 84 yellow flowers that opened on day 2 stayed yellow the whole day, only darkening a bit towards the evening . By the next morning, the 84 flowers were orange. Flowers lasted 4 days total (~50 percent of flowers that opened on day 2 were missing on the morning of day 6) . FIGURE 1. Lantana color changes over time . These two pictures depict the same inflorescence separated by 24 hours. New flowers opened in the morning and turned orange by the next day. Flowers lasted approximately 4 days. Examination of five inflorescences under the dissecting microscope revealed t hat most flowers ( a n estimated 95 percent) , including red ones, had viable looking pistils: these were moist and erec t, a nd showed no signs of decay . Only red flowers very near the outs ide of the inflorescence had pistils t hat looked withered ( ~ 10 percent of red flowers) . Furthermore, t he maj ority of flowers, including red ones, had pollen on the anthers ( ~ 90 percent) , although observation revealed that pollen in yellow flowers was the most abundant. M ore flowers closer to the edge of the inflorescence had pollen deposited around the pistil (17.2 percent of 58 red orange flowers versus 4.2 percent of 24 yellow flowers), yet many flowers even on the edge (82.8 percent) did not appear to be pollinated. N ECTAR CONSUMPTION AN D VISITATION . I observed Anartia and Danaus butterflies as the primary pollinators of the L. camara plants under study. Trigona bees were observed as robbers. Patterns of nectar differed between bagged and unbagged fl owers (figure 2). N ectar in bagged yellow flowers peaked in the middle of the day (1230 ) and all examined had nectar (figure 2a ) . In contrast, bagged yellow flowers were least likely to have nectar at 1230 (only 43 percent of yellow flowers had nectar, figure 2b). Thus the proportion of flowers with nectar dropped by 57
percent as compared to the bagged flowers. O range flowers in both the bagged and unbagged condition w ere most likely to have nectar earlier in the day, at 0930 (69 a nd 38 percent of flowers had nectar, respectively) after which the proportion of orange flowers with nectar sharply dropped off ( 35 and 10 percent had nectar at 1230) . There are no data for orange flowers after 1230 because these had darkened so much by t he time that they could no longer be distinguished from red flowers. About 20 percent of bagged red flowers contained nectar. Nectar in red flowers peaked during the evening in both conditions (40 percent bagged had nectar and 10 percent unbagged had nec tar) , and in the un bagged condition they had the least nectar at 1230 (0 percent had nectar a drop of 15 percent when compared to bagged flowers). It appears that yellow and red flowers replenish their nectar stores throughout the day. Groups differed in proportion of flowers with nectar between N and AY inflorescences (Chi squared, X 2 = 5.71, df = 1, p = 0.017 , figure 3 ). Bagged N inflorescences had slightly more nectar at 1230 than bagged AY inflorescences (46 and 40 percent respectively) , but this dif ference was not statistically significant (Chi s quared, X 2 = 1 .02, df = 1, p = 0.312 ). On the other hand, unbagged N inflorescences had nearly three times as much nectar as unbagged AY inflorescences ( 37.5 versus 12.6 percent; Chi squared, X 2 = 16.9, df = 1, p < 0.0001). Butterflies visited AY inflorescences nearly twice as often as N inflorescences: visitatio n was 22 and 10 in one hour , respectively (Chi squared, X 2 = 4.5, df = 1, p = 0.034). FIGURE 2. Proportion of yellow, orange, and red L. camar a flowers for bagged (a) and unbagged (b ) inflorescences with nectar at five times of day. Proportions for yellow flowers are shown in yellow (top line), while those for orange (middle line) and red (bottom line) are shown in orange and red, respectively. Four inflorescences for each time period were examined ( 20 30 flowers per inflorescence).
FIGURE 3. Proportion of L. camara flowers with nectar in bagged and unbagged inflorescences on normal inflorescences (N) and all yellow inflorescences (AY ). All flowers on each inflorescence were counted. Groups differed in proportion with nectar (Chi squared test) . There was no difference between bagged N and AY inflorescences, but the difference between unbagged inflorescences was significant (Chi squared test) . From left to ri ght, n = 117, 104, 108, and 103 flowers. R OBBERY . Of the N, Y, and AY inflorescences examined for robbery in the butterfly garden, Y experienced the most robbery per inflorescence (mean Â± 1 standard error = 2.7 Â± 0.958 flowers per in florescence ), followed by N inflorescences (1.7 Â± 1.08), and finally by AY inflorescences, which experienced almost no robbery (0.13 Â± 0.13). However , these differences were not significant (One Way ANOVA, F = 2.33, df = 2, p = 0.110) because one plant of the 17 sampled dominated robbery: 93.8 percent and 90.2 percent of the total robbery on N and AY inflorescences, respectively, took place on this plant. I had previously observed that this plant received a dis proportionate amount of robbery, and that AY inflorescences were robbed only on extremely rare occasions (most days they had no robbery at all). Y inflorescences at the Cloud Forest School experienced about ten times more robbery per inflorescence than N inflorescences ( 3.71 Â± 0.323 versus 0.334 Â± 0.340) and this difference was significant (Two Way ANOVA, F = 51.6, df = 1, p < 0.0001, figure 4). There was also a significant difference between plants (Two Way ANOVA, F = 8.18, df = 2, p = 0.0005), but this was because Y inflorescences on plan t 1 had much more robbery than those on plants 2 and 3 (6.39 Â± 0.864 versus 2.42 Â± 0.662 and 2.33 Â± 0.744 , respectively). In addition, yellow flowers were 37 times more likely to be robbed on Y inflorescences than those on N inflorescences: 37 percent of yellow flowers were robbed on Y inflorescences while only 1 percent of yellow flowers were robbed on normal inflorescences (n = 792). 0 0.1 0.2 0.3 0.4 0.5 N AY Proportion of Flowers with Nectar Inflorescence type Bagged Unbagged
FIGURE 4. Robbery per inflorescence for normal (N) and yellow (Y) inflorescences on three L. camara pl ants. Error bar s represent one standard error. Y inflorescences were created by removing orange and red flowers from N inflorescences. Y inflorescences experienced significantly mor e robbery than N inflorescences, and those on plant 1 had significantly more robbery tha n Y in florescences on plants 2 and 3 , p < 0.05 ) . From left to right n = 20, 24, 21, 18, 18, and 24 inflorescences ). R OBBERY CONTROLLING FOR REWA RD . Robbery per inflorescence was significantly different between Y, N, and AN inflorescences (Two Way ANOVA, F = 80.6, df = 2, p < 0.0001 , figure 5 ). Once again, Y inflorescences had significantly more robbery than other groups , with 6.67 Â± 0.47 instances per inflorescence. AN inflorescences had about twice as much robbery as N inflorescences (1.8 3 Â± 0.34 versus 0.900 Â± 0.227); however, this difference was not significant ( (Two Way ANOVA, F = 0.870, df = 2, p = 0.423). FIGURE 5. Robbery per inflorescence between normal (N), yellow (Y), and added nectar (AN) i nflorescences on three L. camara plants. Error bars represent one standard error. Y inflorescences were created by removing orange and red flowers from N inflorescences. A N 0 1 2 3 4 5 6 7 8 Plant 1 Plant 2 Plant 3 Robbery/inflorescence Plant number N Y 0 1 2 3 4 5 6 7 8 9 Plant 1 Plant 2 Plant 3 Robbery/inflorescence Plant number N Y AN
inflorescences were created by adding nectar (30% sugar) to orange and red flowers on N inflorescences . Robbery per inflorescence was significantly different between inflorescence types , but not between plants (Two Way ANOVA). Differences between N and AN , p > 0.05 ). n = 10 inflorescences for each bar. R OBBERY CONTROLLING FOR SIZE . Robbery per inflorescence was significantly different between R and Y inflorescences (One Way ANOVA, F = 184.4, df = 1, p < 0.0001 , figure 6 ). Y inflorescences had a large amount of robbery (8.13 Â± 0.415) while R had very little slightly less than the usual for N inflorescences (0.227 Â± 0.415). There was no significant difference between plants (Two Way ANOVA, F = 0.97 3, df = 2, p = 0.384). FIGURE 6. Robbery per inflorescence between red (R) and yellow (Y) inflorescences on three L. camara plants. Error bars represent one standard error. Y inflorescences were created by removing orange and red flowers from N inflo rescences, and R inflorescences were created by removing flowers from all red inflorescences . Both Y and R inflorescences had 8 12 flowers. Robbery per inflorescence was significantly different between inflorescence types but not between plants (Two Way ANOVA). n = 11 inflorescences for the first bar, and 10 for the rest. R OBBERY AY INFLORESCENCES . Robbery per inflorescence for Y, N, and AY inflorescences at the Cloud Forest School (AY had been transplanted from the field near the butterfly garden) di ffered between groups (One Way ANOVA, F = 90.1, df = 2, p < 0.0001). Y inflorescences had 5.6 Â± 0.830 instances of robbery per inflorescence , while N and AY inflorescences had much less: 0.81 Â± 0.791 and 0.00 Â± 0.875, respe ctively. It is noteworthy that once again AY inflorescences experienced less robbery than N inflorescences; however, this difference was not P OLLINATION . Inflorescences that both pollinators and robbers could access, that only robbers could acc ess, and that neither pollinators nor robbers could access had significantly different amounts of pollen deposited near or on the pistil (One Way ANOVA, F = 15.0, df = 2, p < 0.0001, figure 7) . Inflorescences that both pollinators and robbers could access had about twice as many pistils with pollen on them than those that only robbers could access (3.07 Â± 0.450 0 2 4 6 8 10 12 Plant 1 Plant 2 Plant 3 Robbery/inflorescence Plant number R Y
versus 1.62 Â± 0.269 ), while those that neither could access had significantly less pollen deposited on them (0.36 Â± 0.2 03 ). FIGURE 7. access them. Error bars represent one standard error. The three conditions were significantly different from each other (One Way ANOVA). From left to right n = 11, 14, and 14 inflorescences. D ISCUSSION In contrast with the results of Barrows (1976), it was found that both orange and red flowers contain nectar. It appears that the plant reabsorbs nectar from yellow and then orange flowers as they turn red and then maintains those s maller nectar quantities in red flowers by placing nectar in them throughout the day . The plant invests the most heavily in yellow flowers, which also receive more visits than the other flowers (as evidence d by the larger drop in proportion with nectar fo r yellow unbagged flowers). This supports the idea that pollinators use the color yellow as a cue for rewarding flowers (Barrows 1976, Weiss 1995). All flowers had nectar removed, which means each type of flower was visited by pollinators and/or robbers throughout the day. Examination of flowers under a dissecting microscope revealed that most flowers, incl uding orange and red ones, were capable of p ollinating and being pollinated . This conflicts with previous claims by both Barrows (1976) and Weiss (19 91, 1995) that red flowers contain spent male and female parts . In addition, although more red flowers than yellow appeared pollinated, the majority were not. These results, as well as those of the previous paragraph, imply that there is more to these ol der, red flowers than previously thought. L. camara plants are still investing energy (nectar and reproductive capability) in mostly unpollinated, older flowers. coloration does not attract pollinators when size is controlled . More nec tar was taken from AY versus N inflorescences , even though they contained approximately the same number of flowers. Visitat ion data substantiates this , as butterflies were more likely to visit the AY inflorescences. This conflicts with Zurinskas (2003) and s uggests that there is no functional significance to the pattern in terms of pollinator attraction. 0 1 2 3 4 None Robbers Robbers & Poll Pistils with pollen/inflorecence Inflorescence type
The presence of orange and red flowers around the outside of the inflorescence severely decre ases robbery of yellow flowers. In fact, I found t hat orange and red flowers are much more effective at this than originally reported by Barrows (1976), who found that their presence reduced robbery by a little over a factor of three. I found, after examining a sample several times that robbery was reduced by a factor of 37. This occurs because orange and red flowers physically block the corollas of yellow flowers from access by robbers. Furthermore, the presence of older flowers decreases total robbery per inflorescence by about 1 0 times: t hese flowers do not just block yellow flowers from getting robbed, they dissuade large amount s of robbery overall . The most likely explanation for this is that robbers have lea rned that older flowers often do not contain nectar, and therefore avoid inflorescences containing them in order to forage more optimally. It may be that this occurs due to associative learning, as has been proposed for pollinators (Weiss 1991). Post cha nge flowers may be serving an important adaptive function if large scale robbery reduces fitness of L. camara . Like pollinators, robbers appear to use color as a cue for identifying rewarding flowers. Y inflorescences were much more likely to be robbed t han inflorescences containing yellow, orange , and red flowers, even when presence of nectar and inflorescence size were controlled . I t appears that robbers are hijacking color cues that are meant to aid pollinator efficiency for their own purposes , and th is behavior is likely at net cost to the plant (once again, this is assuming that large scale robbery of yellow flowers reduces fitness of L. camara ) . My results provide some evidence that robbers are serving as pollinators of L. camara . Inflorescences t hat only robbers could access had significantly more flowers with pollen deposited on or near the pistil than those that neither pollinators nor robbers could reach . However , these data must be interpreted with caution , as robbers could simply be knocking pollen from th e same flower onto the stigma . L. camara cannot self pollinate (Barrows 1975) , so this action would not enhance the fitness of the plant. Trigona bees , the only robbers seen in this study, are often associated with negative effects on seed set (Maloof and Inouye 2000). One study of the tropical treelet , Quassia amara , found that exclusion of robbers resulted in 4 12 times greater seed production than in control flowers (Roubik et al. 1985) . With all this in mind, I propose that large scale robbery of L. camara inflorescences, especially the most active and rewarding yellow flowers, may decrease the fitness of the plant most likely through a reduction in pollinator visitation (Maloof and In ouye 2000). This makes sense in light of the fact that a ring of red flowers is so effective at reducing robbery , and may have arisen due to selective pressure from robbers. Another possibility is that these flowers are the outcome of pressure to enlarge the landing platform or to increase the attractiveness of the inflorescence with larger size , as has been suggested by Schemske (1976) and Weiss (1991), but it seems like an outstanding coincidence that they would also be so successful at reducing robbery at the same time. It is likely that all these forces have contributed to the maintenance of older flowers. However, g iven that robbers are capable of pollinating, it makes sense to maintain limited nectar quantities and reproductive capabilities in oran ge and red flowers. In this way, the plant is balancing fitness benefits from pollinators and robbers, or at least mak ing the most of a bad situation: pollinator interactions are not greatly disrupted , since robbers cannot reach the yellow flowers and ar e dissuaded from (likely detrimental) large scale robbery , yet the plant still receives some pollination from robbers at a low cost. This hypothesis fits well with the results and explains why post change flowers contain nectar and are reproductively activ e. Of course, this could be solely for the pollinator benefit, but the results of previous studies suggest that pollinators visit older flowers very rarely: about 1 percent of the time (Barrows 1976) . T hese
limited pollinator interactions with older fl owers may not be beneficial enough to explain the . A future study examining see d set in N and Y inflorescences with differential robber visitation would be helpful in testing my hypothesis and resolving this question . Finally, i t appears that there is genetic variation in the population of L. camara , as AY inflorescences discovered during the course of the study contained no orange or red flowers, yet had the same distribution and amount of nectar as N inflorescences. Despite th ese similarities, AY inflorescences appeared to be highly resistant to robbery, even when attached to a plant at the Cloud Forest School garden, where robbery was high. The mechanism and selective forces associated with this resistance remain unknown . Th is study has shed light on how color changes influence the beha vior of pollinators and robbers. The results suggest that these changes serve a dual function of directing pollinators to rewarding flowers and reducing robbery with the end of maximizing poll ination from legitimate pollinators and possibly robbers. F indings for L. camara may apply to the many species with similar floral changes; however, idiosyncrasies due to the identity of the legitimate pollinator and the robbers are likely . Exploring these idiosyncrasies with L. camara in mind may be the next step in fully understanding the significance of floral color changes to plant fitness. ACKNOWLEDGEMENTS I thank Alan Masters for his support in the development, execution, and analy sis of this study. Emily Shelly and Laura Hurley were particularly helpful in their review of the written work. Those at the Cl oud Forest School in Monteverde who g ranted access to the study site are appreciated. LITERATURE CITED B ARROWS , E. M . 1976. Nectar robbing and pollination of Lantana camara (Verbenaceae). Biotropica. 8 : 132 135. G ORI , D. F. 1989. Floral color change in Lupinus argenteus (Fabaceae): why should plants advertise the location of unrewarding flowers to pollinators? Evolutio n. 43 : 870 881. K ENDELL D. A., AND B. D. S MITH . 1976. The pollinating efficiency of honeybee and bumblebee visits to flowers of the runner bean Phaseolus coccineus L. Jo urnal of applied ecology. 13 : 749 752. M ALOOF , J.E., AND D.W. I NOUYE . 2000. Are nectar robbers cheaters or mutualists? Ecology. 81: 2651 2661. R ICHARDSON , S. C . 2004. Are nectar robbers mutualists or antagonists? Oecologia. 139: 246 254. R OUBIK , D. W., N. M. H OLBROOK , AND P. V. G ERMAN . 1985. Roles of nectar rob bers in reproduction of the tropical treelet Quassia amara (Simaroubaceae). Oecologia. 66: 161 167. S CHEMSKE , D. M . 1976. Specificity in Lantana camara and L. trifola (Verbenaceae). Biotropica. 8: 260 264. W EISS , M. R . 1991. Floral colour changes as cues for pollinators. Nature. 354: 227 229. . 1995. Floral color change: a widespread functional convergence. American Journal of Botony. 82 : 167 185. Z URINSKAS , M. 2003. The importance of the bulls eye coloration pattern on the rate of inflorescence visitation of Lantana camara . CIEE summer tropical ecology and conservation.