International Association for Ecology
Roles of Nectar Robbers in Reproduction of the Tropical Treelet Quassia amara (Simaroubaceae) Author(s): David W. Roubik, N. Michele Holbrook, German Parra V. Source: Oecologia, Vol. 66, No. 2 (1985), pp. 161-167 Published by: Springer in cooperation with International Association for Ecology Stable URL: http://www.jstor.org/stable/4217608 Accessed: 29/11/2010 14:35 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=springer. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact
[email protected].
Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and extend access to Oecologia.
http://www.jstor.org
Oecologia
Oecologia (Berlin) (1985) 66:161-167
© Springer-Verlag 1985
Roles of nectar robbers in reproduction of the tropical treelet Quassia amara (Simaroubaceae) David W. Roubik, N. Michele Holbrook*, and GermanParra V.** SmithsonianTropical Research Institute, APDO 2072 Balboa, Panama
Summary. Different subsets of mainland nectarivores visited Quassia amara (Simaroubaceae), a self-compatible, predominately bird-pollinated treelet, at three islands and the mainland in Panama. Factors correlatedwith reproductive success, defined as seed to ovule ratio, included the species pollinating and robbing flowers, visitor activity, pollinator response to nectar robbing, and internal regulation of fruit production. The absence of robbers and former pollinators on an island separated from the mainland during the holocene was associated with shifts in flower size, nectar production, and 3-4fold increases in population reproductive success and pollinator efficiency (= seeds produced per visit). Exclusion of robbers at three sites resulted in seed production 4-12 times greater than control flowers, at which robbers accounted for 52-98% of all visits. Although 36% of buds and over 83% of all flowers were robbed, this had no direct influence on the recorded 36-61 % respectiveabortion rates of buds and flowers. Opportunistic avian robbers appeared where normal robbers were absent; three avian robbersextensivelyused floral perforations made by Trigona bees, and all ancillary pollinators also robbed. Selection pressures from nectar robbers are discussed that may relate to plant reproductivefitness.
Many ecological and evolutionary relationships, certainly those involving flowering plants and their visitors, are complicated by population changes, the fluctuations of climate and weather, and extinction (e.g. Feinsinger 1983; Janzen 1983; Howe 1984). Although pollination of flowers presumably requires a degree of coevolution between lineages of flowering plants and their mutualists, the type and abundance of pollinators and non-pollinating visitors may vary dramaticallyboth in time and between localities. One manner in which floral speciation is thought to occur results from change of primary pollinator (Hagerup 1951; Grant and Grant 1965; Heinrich and Raven 1972; West-Eberhard 1983; Crepet 1983), but much less attention is given to adaptivechanges due to the type and frequencyof visitation by predominately non-pollinating organisms. Thieves, which can be though to include occasional pollinators and Offprintrequeststo: D.W. Roubik * Presentaddress:Universidadde Los Andes, Merida,Venezuela ** Presentaddress: INCIVA, APDO 2056, Cali, Colombia
all visitors which to greater or lesser extent legitimately remove floral resources without pollinating, and robbers, which damage floral parts and take resources whthouteffecting pollination (Inouye 1980), often visit flowers ostensibly adapted for a different class of visitor (Heithaus et al. 1982; Haber and Frankie 1982; Roubik 1982; Cruden et al. 1983). When the activities of all types of flower visitors are considered, the adaptive value of traits such as breeding system, morphology and mode of pollen or nectar presentation may be clarified. Other characteristics of flowering, such as the number of fruit and flowers produced, the number aborted, and timing and synchrony of anthesis within populations have extremelyvaried biotic and abiotic correlates (Janzen 1978; Reich and Borchert 1984; Bawa and Webb 1984; Heithaus et al. 1982; Willson 1983; Stephenson 1984). If some correlates are ignored, the ecology, natural selection and evolution within plant-pollinator assemblagesmay be misunderstoodor even obscured. Assessing relative plant reproductive success requires a basic judgement which, as attested by the foregoing studies, must remain uncertain-ascertainingthe extent to which ovules and pollen representpotential gamete contribution to future generations when all biotic and abiotic factors are at an optimuim. This study considers the ecological importanceof pollinating and non-pollinating visitors of a tropical forest understorey treelet, Quassia amara (Simaroubaceae).Descriptive and experimentalstudies of visitation and fruit production were used to examine visitor impact and discover selective pressureswhich help account for featuresof flowering, nectar and pollen presentation and fruiting by Quassia. Comparative studies were carried out in a variety of settings - a natural, original mainland community, one recently formed natural island community, and two disruptedplantpollinator communities on manmade islands. Here we document and interpret (1) the impact of nectar robbers; (2) bud, flower and fruit survivorship; (3) pollinator efficiency and (4) unique features within an island population of Quassiawhich has been isolated from the mainland since the holocene. Methods and materials Plants and study sites Quassia amara is an evergreen, understorey treelet four to eight meters high, common in young forest and forest edges
162
I1
Quassia amara is abundant throughout the young forest (80-120 years in age, Croat 1978; Foster and Brokaw 1982). We studied treelets growing in three discrete areas, separated by a maximum of 1 km. (2) A second manmadeisland was Scheelea Island (SI), located approximately 5 km N of BCI. This 7 hectare island has vegetation similar to young forest on BCI. (3) The mainland study site was near the Pipeline Road (PR), 8 km N Gamboa within Parque Nacional Soberania. This area is approximately 10 km E of BCI. It is comprised of forest edge 43 years old along a road and trails, and 1,300 square km of secondary and mature forest. Study trees were from four discrete areas along ravines or streams adjacent to, but not along, the Pipeline Road, and were separated by no more than 4 km. (4) The last site was on Taboga Island (TI) on the northernfacing slope in second growth deciduous tropical dry forest. Here we studied the treelets growing in four discrete sites, separated by less than 1 km. This island lies 12 km S of the mainland, approximately 37 km SE of BCI, in the Bay of Panamfaand has been separatedfrom the mainland some 12,000 years (Bartlett and Barghoorn 1973). Observation offlower visitors
Fig. 1. A female hummingbird,Damophilajulie, removing nectar from Quassiaamarathrough a hole made by the social bee Trigona
Note floraltissuedepositedby robbingbeesnearthe fulviventris. middleof flowers,abortedbuds, and a flowerwith more than one perforation.The insertshows a one day old flowerat the leftanda two day old flower from southern Mexico to northern South America (Croat 1978). In central Panama, flowering is concentrated in the end of the rainy season (October to January), although scattered flowering occurs in any month from August to March (Croat 1978, D. Roubik, personal observation). Here the trees are found in tropical lowland moist forest and also in dry, deciduous forest of Taboga Island, in the Pacific Ocean (Croat 1978). The gynoecium of Q. amara consists of five, very rarely six, bi-ovulate pistils. However, we saw no case of only one seed developing within a locule; apparentlypollen fertilizesboth ovules simultaneously.Fertilized locules are easily counted without destructive sampling. Newly opened flowers had a broad gap between the petal tips and had full, yellow anthers, while two day old flowers were closed at the tip and had shed the anthers (Fig. 1). The flowers are deep pink and their average length at dehiscence varies between 4 and 4.1 cm on the mainland (range= 3.7-5.1 cm, N= 161, data taken at four sites at Atlantic and Pacific coasts and central Panamfa)and is 3.0 cm (range 2.5-3.4 cm) on Taboga Island. Flowers are sympetalous and tubular, typical for hummingbird-pollinatedflowers (Fig. 1). They are generally open for two days (Fig. 1) and each inflorescence presents 1-4 open flowers at once. The flowers are fragile and sometimes fall from the rachis at the slightest touch. The floral parts and sap of the plant contain a bitter tannin chemical, quasin (Croat 1978). Our studies were performedat four sites in central Panama'during late 1980 and early 1981. Two island sites were in Gatun Lake, which was formed in 1914 during construction of the PanamfaCanal. (1) The first of these was Barro Colorado Island (BCI), a forest preserve containing 1560 hectares of semi-deciduoustropical moist forest (Holdridge et al. 1971). On the northern side of this former hilltop,
Visitor activity at flowers of Q. amara was monitored on BCI during December 8-18 in 1980 and at all four sites during January9-31, 1981. Initial observations at BCI were made hourly from 06.00 to 18.00 h, but all subsequent observationswere made from 08.00 to 18.00, since early morning pollinator visits were extremely rare. Observationswere made during 10-min intervals at 25, 18, 24 and 21 trees for BCI, PR, SI and TI, respectively, for a total of 98 h. The total time that flowers were watched for visitors during hourly intervalsaveraged200, 102, 124 and 148 min at these respective sites. Five or six plants were monitored during each hour on a given day. Allbirds foraging nectar were recorded and identified to species and sex when possible. A visit was scored as legitimate when the birds entered the nectar channel by inserting the beak into the corolla, or illegitimate when it pierced or damaged the flower, or used an existing hole made by a robber. Bee visitors were also scored for abundance and type of visit. For Trigona,the total bees on flowers was counted at the beginning and the end of a 10 min observation. During additional observations, individual bees were followed 10-15 min to determine their rate of visitation. Total time spent by bees making a hole in flowers was recorded. The number of visits made by other bees was scored in the same manner used for the birds. Nectar dynamics
To determine amounts of sugar available to visitors and nectar volume, the nectar quantity and sugar content (% total dissolved solids, weight/total weight) were measured from flowers bagged with nylon netting and open flowers at all sites. Such measurements were made by destructive sampling of flowers during two daily intervals: between 08.00 and 10.30, and from 12.00 to 17.00. Both one- and two-day old bagged and unbagged flowers were sampled from each site; unbagged flowers were both robbed and intact. Samples came from a total of 626 flowers on 17 trees on TI, 16 on BCI, 12 on SI and 5 on PR; none of the
163
flowers was in prolonged direct sunlight. On BCI the nectar of 35 buds was sampled one day before anthesis. Flower survivorship
Factors affecting survival of buds in all stages, flowers and fruit were studied by following temporal patterns of floral maturation, abortion and robbing damage for two cohorts of flower buds at BCI. One cohort of 592 buds from 18 trees was studied from 6-21 December; a second cohort of 100 large buds (>3.5 cm in length, 1-3 days before anthesis) from 10 trees was studied from 13 December to 10 January. The corolla length, condition of stigma and anthers, number and location of robbing holes and the number and size of developing seeds were noted for each. This information was recorded daily. Two other methods were used to study bud and flower abortion. During study of the 100-bud cohort, 10 fallen flowers and buds from each of 5 trees were collected every 2-3 days and scored for robber damage, for a total of 494 flowers and buds. In addition, all old fallen flowers and buds were removed beneath 5 trees and, in the next four days, all of 469 buds and flowers aborting were gathered to determine the proportion aborting as buds and robber damage among all aborted floral units. Reproductive dynamics, robber exclusion and pollination efficiency
To estimate average production of seeds and fruit during the entire flowering season, a census of total inflorescences, flowers, fruit and seeds was conducted during 1980-81 on 37, 28, 22 and 19 plants at BCI, PR, SI and TI, respectively. Each treeletwas scored for inflorescenceand flower number when flowering was nearly finished and fruit on trees were well developed, past the stage when abortion occured. From 112 to 734 fruit were scored for seed (locule) set at each site. Total potential flower production was estimated by multiplying total inflorescences held on a tree during the flowering period by the average total number of buds per inflorescence.Data for average bud number were taken on BCI (N= 16 trees, 210 inflorescences)and TI (N= 23, 60). We hand pollinated flowers to assess self compatibility and stigmatic receptivity.Bagged day-one flowers were pollinated using anthers from early day-two flowers on the same plant; day-one pollen was placed on day-two stigmas on the same plant, and selfings of day-one with day-one flowers from the same plant also were performed. Emasculated as well as unmanipulated flowers were bagged to check for apomixis and self-pollination. At least 10 flowers were used for each treatment.Fruit and seeds were recorded at least 7 days later, when they had attained roughly half their mature size. Indirect tests of robber impact on pollinator activity and pollination were made by making the base of flowers inaccessibleto robbers. Large flower buds were fitted with small, open-ended muslin pockets in a way that offered protection from robbers but allowed legitimate pollination. Robbers were thus excluded from 36-277 flowers on 4-15 treelets at BCI, PR and SI. As controls, 184 flowers on 6 of the experimental plants at PR and BCI were marked at the same time other flowers were fitted with pockets, left uncovered, and later scored for fruit and seeds.
Statistics and computations
Since this study deals primarilywith population phenomena involving individualplants and possible coincident selective pressures,reproductivesuccess and pollinator effectiveness are discussed in terms of population averages. At all sites the potential population reproductivesuccess (RS) was calculated by considering the rachis number on each census tree, which does not change during flowering,and multiplying this by total average bud number per rachis. This product was multipled by 5, the number of locules that can be produced from each flower. The total recorded locules on a tree at a final census was divided by the product above, and the average of these quotients is considered population RS. Unless otherwise indicated, all statistical comparisons of RS at different sites were made using non-parametric Friedman 2-way ANOVA and Scheffe Multiple Comparison tests, where rows representindividual treelets, and columns representeach of the four study sites. Pollinator efficiency was computed from the observed hourly visitation rates during the period in which pollinators visited flowers and stigmas were receptive. The average flower visitation rates for each observation hour were computed as the ratio of recorded visits to the product of total flowers observed and total observation time. These rates were summed for all hours of pollinator visitation during stigmatic receptivity. This period was found to be 14 h (from 10.00 to 17.00 during two days). Thus, the index of pollinator efficiency considered estimated total visits to flowers divided by total seed set. Results In the presence of perforations made by robbers, all flower visitors but the primary pollinator were at least occasional secondary nectar robbers. Primary robbers made perforations used by secondary robbers, as shown in Fig. 1. The names and roles of flower visitors are given in Table 1. The social bee Trigona (Trigona) f. fulviventris (Apidae:
Meliponinae)was the only robber that made a circularhole, very distinctive from the slit made by the specialized robbing Little Hermit hummingbird, Phaethornis longuemareus
(Trochilidae:Phaethorninae).The primarypollinator at PR and BCI was the Long Tailed Hermit, P. superciliosus, whereas one of the primary nectar robbers, P. longuemareus, was not seen at flowers on BCI. These three species are at present the principalvisitors of Quassia on the mainland, yet Phaethornisappears to be gradually disappearing from BCI communities (Willis and Eisenmann 1979). All of the other visitors recorded in Table 1 are known from the BCI and PR areas, although none were seen at flowers there, including the solitary bee Ceratinalaeta (Anthophoridae: Xylocopinae), which was one of the principal pollinators at Taboga Island (Willis and Eisenmann 1979, D. Roubik, personal observation). Avian robbers exhibited a high degree of behavioral flexibility. Only the bananaquit, Coerebaflaveola (Coerebidae), which robbed by tearingoff a petal, and the hummingbird, Damophilajulie, which exclusively used bee perforations, showed consistent behavior. In contrast, P. longuemareusslit corollas or used bee perforations; the hummingbird, Thaluraniacolombica,used bee perforationsor visited legitimately as a pollinator (13% of visits), and the hummingbird, Lepidopyga coeruleogularis,slit corollas, used
164 Table 1. Communitycomparisonsof flower visitors, pollinator efficiencyand reproductivesuccess of Quassiaamarain Panama Community description
Site & RSa
Pollinators
Robbers
Pollinatorvisitsb
Seeds per visit
Original
PR.
0.05
Phaethornis superciliosus hb
48%
6.3
0.04
Artificial, PollinatorDepauperate,
BCI.
0.02
P. superciliosus Thalurania colombica hb
Trigonafulviventris b P. longuemareus hb T. fulviventris T. colombica Damophila julie hb
2%
0.2
0.55
Artificial, Ancillary visitors
SI.
0.01
Lepidopyga coeruleogularis hb Coerebaflaveola hc L. coeruleogularis hb
20%
2.2
0.02
Recently isolated
TI.
0.14
100%
4.5
0.17
Ancillaryvisitors
Amaziliaedwardhb Ceratina laeta b
a RS = Average plant reproductivesuccess, total seeds divided by total ovules b
The first value gives the percentage of visits not involving primary or secondary robbing; the second value represents the total pollinator visits calculated over the period of stigmatic receptivity,based on observed average flower visitation rates (see text). hb= hummingbird; b = bee, hc = honeycreeper
Table 2. Average numberof fruits and seeds (2 per locule) set per flower at robber-excludedand control flowers of Quassiaamara Site
BCI PR SI
Control flowersa
Robber-excludedflowers Locules
Fruits
Locules per fruit
Locules
0.35 1.36 0.60
0.12 0.42 0.20
2.85 2.67 3.00
0.09 0.27
Locules per fruit
Fruits 0.09 0.24 0.05
0.04 0.05
0.03 0.08 0.02
2.85 3.27
3.12 3.93 2.21
a The first values are for uncovered flowers observed simultaneously at the same plants on which robbers were excluded; the second
values are from total population censuses
flowers opened by the bananaquit, or made 40% legitimate pollinator visits. Robbers appeared to inspect flowers for holes, and some species (Table 1) functioned as robbers and pollinators while visiting a flower. Bees did not defend flowers from birds and were responsible for most robbing where birds also robbed. On BCI 78% of robbing visits were by T.Julviventris, while 54% of robbing visits were by this species at PR. Bees were aggressive toward foragers from rival colonies but did not arrive in groups and initiated hovering displays singly with conspecifics (Roubik 1982). Bees visited an average of 4.2 flowers in 10 min, visited all perforations they encountered, perforated up to three times on a flower, and spent about 10 min making a perforation. They foraged individually at flowers and their rate of visitation was not statistically variable through the day (P>0.05, Model 1 ANOVA for three-hour observation blocks, 06.00-18.00 h). Ants were not observed robbing but were occasional thieves and included Ectatomma, Leptogenys, and Paraponera. Only one insect gathered pollen from the anthers, the small meliponine bee Trigona (Plebeia) franki, during which it did not contact the exserted stigmas. There was no general agreement between the efficiency of pollinators and reproductive success of Quassia at the study sites. Table I provides estimates of population reproductive success, assuming that five locules could be filled for each flower bud, and the average single visit efficiencies of pollinators. Since the results of bagging experiments showed no fruit production from emasculated or bagged flowers, autogamy or apomixis were unlikely, thus a pollin-
ator visit was required to produce a fruit. On the average, pollinators at BCI were 13 times more efficient than those at PR, but population reproductive success (RS) at the former site was only 40% that of PR. This indicates that although pollinations by Phaethornis superciliosus or Thalurania colombica were relatively rare on BCI, when a bird did visit a flower it transferred more pollen that germinated to fertilize ovules than did pollinating birds at the PR site. In contrast, in the absence of robbers at the TI site, pollinators were four times more efficient and population RS was three times greater. The lowest RS and pollinator efficiency were recorded at SI (Table 1). Statistical comparisons of treelet population RS at the four sites showed substantial differences between the four sites, between the mainland and artificial island sites, and between most pairs of sites (P
