BlOTROPlCA 30(2): 179-189 1998

Facultative Ant- Interactions: Sugar Preferences of Introduced Pest Ant Species in South Florida’

Suzanne Koptur and Ni Truong Department of Biological Sciences, Florida International University, Miami, Florida 331 99, U.S.A.2 Fairchild Tropical Garden, Miami, Florida 33156, U.S.A.

ABSTRACT We observed nectar use by native and exotic ant species in nature, garden, and urban situations, and found ants utilizing floral and extrafloral nectar of a variety of species. We collected 31 plant (29 floral, 2 extrafloral) and used them in feeding preference tests against standard solutions of sugars (20 percent fructose, glucose, and sucrose, and their mixture), 10 trials for each nectar-ant comparison. We comparcd time-to-discovcry and total ant visits to each droplet using ANOVA, and found that both trial and solution contributed significantly to the variation in most experiments. Seven of the floral ncctars tested were significantly more attractive to certain ant species than the sugar solutions. Not only do ants use floral nectar, but it appears that some floral nectars contain compounds that are especially attractive to ants.

RESUMEN Se observ6 que varias especics de hormigas nativas y exoticas utilizaban nectar de florcs y nectar extrafloral dc varias cspecics de plantas en Aoraci6n, localizadas en jardincs, en el casco urbano y en areas naturalcs. Se colectaron 31 nectares dc plantas y su prcfcrencia fut comparada con soluciones de azucar (20 percicnto de fructosa, glucosa, y sucrosa, y sus rnczclas). Esta comparaci6n fuc realizada 10 vcces para cada combinacion de nectar y hormiga. Se utliz6 el Analisis de Varianza para comparar el tiempo quc lc tomaba a las hormigas en descubrir las solucioncs y el n6mero dc visitas a cada gota de nectar o solucibn. El analisis estadistico demonstrb que tanto el nhmero de replicaciones como la clase de soluci6n contribuia significativamcnte a la variacion dc la mayoria de 10s experi- mcntos. Sicte de 10s ncctarcs florales fueron mas significativos a algunas cspccics dc hormigas que las soluciones de azucar. Evidcntamentc, algunos nectarcs florales contcnian algunas substancias especialmente atraycntes a las hor- migas.

Key words: mts; feeding; Florida; nectar: preference.

ANTS ARE UBIQUITOUS AND PERFORM MANY IMPORTANT tile, ant-repellent substances in petal tissues of FUNc’ric” in the ecology of in tropical and plants, and demonstrated their existence in a sim- temperate areas (Huxley & Cutler 1991). Ants are ple experiment. often attracted to plants to collect exudates from The Bakers (1978) were the first to respond to extrafloral nectaries or from insects consuming the Janzen’s hypothesis, pointing out their observations plants. Flowers often contain nectar as the floral of chemicals in nectars such as alkaloids, phenolics, reward, and Janzen (1977) posed the question and nonprotein amino acids that may serve as de- “Why don’t ants visit flowers?”, hypothesizing that terrents to nectar theft. Feinsinger and Swarm floral nectar must contain ant repellents to protect (1978) tested the acceptability of nectar to ants of it for pollinators. Nearly one hundred years earlier, four common plant species in Trinidad, and found A. Kerner von Marilaun (1878) had described that responses of ants varied widely. Schubart and many ways in which plants can exclude nectar- Anderson (1978) offered freshly cut-open flowers thieving ants, including extrafloral nectaries, sticky of three plant species to ants in Brazil, and found stems, dense hairs, water traps, and distasteful sub- all to be readily consumed; though ants may have stances in floral tissue. Van der Pijl (1955) used the been excluded from flowers by structural barriers, term “myrmecophoby” for the deterrence of ants ant-repellent nectars were not indicated. Guerrant from flowers, hypothesizing the existence of vola- and Fiedler (1981) compared the acceptabiliry of floral nectars and floral tissue extracts to sugar so- lutions, and analyzed the plant substances chemi- Received 10 April 1997; revision accepted 29 Scptem- ber 1997. cally to detect deterrent substances; they found that * address for correspondence. floral nectars were attractive to ants, but floral tis-

179 180 Koptur and Truong

Adenocalymna comosum Agave decipiens

time to discovery time to discovery

total ants total ants

la 10 15 20 25 30 0 5 Ib 0 10 20 30 40 50

Aloe saponaria Aph elandra jacobinioides

time to discovery time to discovery

totai ants totai ants

Ic 0 10 20 30 40 50 Id 0 5 10 15 20 25 30

Brownea sp. Brun felsia grandiflora

time to discovery time to discovery

total ants total ants

le 0 5 10 15 20 25 If 0 5 10 15 20 FIGURE 1. Results of feeding experiments with crazy ants, Paratrechina longicornis. Bars represent the means of ten trials for each experiment (variation not shown). The upper set of bars represents the time-to-discovery of the solution in minutes; the lower set of bars represents the total number of ants to visit that solution during the one hour that the experiment ran. EFN = extrafloral nectar. Solutions are abbreviated as: N = nectar, F = fructose, G = glucose, S = sucrose, and FGS = fructose, glucose, and sucrose. Significant differences shown by Analysis of Variance and Tukey post hoc tests are indicated by lower case letters, in which case bars with different letters are significantly different within that data set. Nectar Preferences of Pest Ants 181

Cnnum moorei Dipteryx panamensis (efn)

time to discovery time to discovery

total ants total ants

lg 0 5 10 15 20 25 30 35 Ih 0 10 20 30 40

Hamelia patens (squeezed) Hamelia patens (2)

time to discovery time to discovery

total ants total ants

li 0 5 10 15 20 25 30 li 0 10 20 30 40 50

lxora coccinea Justicia fulvicoma

time to discovery time to discovery

total ants total ants

Ik 0 5 10 15 20 25 30 11 0 5 10 15 20 5

~ ~ FIGURE 1. Continued.

sues showed highly variable palatability. Haber et consumed, and many of those enjoyed by ants were al. (1981) had observed ants taking floral nectar known to have phenolics and/or alkaloids. Many from many species of plants in the lowland dry nectars, therefore, are readily consumed by ants forest of Costa Rica, and tested many of these by when they are offered out of their flowers, if they placing nectar droplets on branches of and have been collected carefully so as to not contam- noting whether or not they were consumed by ants inate them with any repellents present in the floral within five minutes; all but two tested were readily parts. 182 Koptur and Truong

Kalanchoe gastonis-bonnieri Megaskepasma erythrochlamys

time to discovery time to discovery

total ants total ants

0 10 20 40 Im 0 10 20 30 40 In 30

Odontonema strictata Russelia sarmentusa

time to discovery time to discovery

total ants total ants

0 5 10 15 20 25 30 10 0 5 10 15 20 25 30 Ip FIGURE I. Continued.

In this study we document the use of nectars portant as pollinators since most have antifungal of a variety of cultivated tropical species by native metapleural secretions (.g.,myrmicacin) that kill and exotic ants, and investigate the attractiveness pollen (Beattie 1985). Our study relates primarily of the nectars to noncoevolved pest ant species in to the abilities of ants to obtain food, and to subtropical south Florida. In conducting the ex- whether or not the nectars tested might have non- periments we obtained information on the relative sugar constituents rendering them exceptionally at- preferences of these ants for the various sugar so- tractive to ants. lutions used in our controls as well as their different responses to the nectars tested. The facultative as- MATERIALS AND METHODS sociations observed suggested nectars that might be attractive to ants; we compared the acceptability of Observations were made over two years in the col- these nectars to standard sugar solutions in order lection of Fairchild Tropical Garden and, to a lesser to determine if any of these nectars were excep- extent, the University Park campus of Florida In- tionally attractive. Sugar solutions are nonvolatile, ternational University. We documented all ants us- and ants discover them by touch and taste; floral ing nectars, collecting ant specimens for determi- nectars may have other components that attract nation and later verification. ants from a distance, or affect their taste once they Several milliliters of nectar were collected into are tried. We do not presume that there are nec- a tiny centrifuge tube, and its concentration mea- essarily coevolved positive relationships between sured using a hand-held refractometer (Bellingham ants and floral nectars in natural communities. Un- & Stanley brand, reading percent sugar on a like the other social Hymenoptera, ants are usually weighdweight scale). Nectars were diluted with dis- visiting flowers as nectar robbers, and are unim- tilled water to a concentration of 20 percent, unless Nectar Preferences of Pest Ants 183

xRuttyruspolia sp. Sanchezia speciosa

time to discovery time to discovery

total ants total ants

5 10 15 20 25 0 10 20 30 40 50 Iq 0 30 35 Ir

Sch o tia bra ch yp efa la

time to discovery

total ants

Is 0 10 20 30 40 50 FIGURE 1. Continued.

they were less than 20 percent sugar to begin with, with a different colony of ants). We did not replace in which case more dilute sugar solutions were pre- the droplets if they were used up before the end of pared. the trial (except for reruns of Hameliuputens, which For each nectar-ant comparison, we ran ten tri- initially was consumed very quickly; see below). als of the following design. Standard solutions of Trials that were interrupted by external factors (dis- 20 percent fructose (F), 20 percent glucose (G), 20 turbance by urban wildlife, inclement weather) percent sucrose (S), and a mixture of 20 percent were not included. fructose, glucose, and sucrose (FGS) were made For each droplet we figured time-to-discovery fresh each month and kept refrigerated, removing (time from initial placement to when it was first only small quantities to use for daily experiments. observed with ants) and the total number of ants We made circular Styrofoam trays (10 cm diameter) observed during the hour’s observation. Ants were with 20 uniformly spaced depressions around the counted at each 5 min interval and the total of edge, and in each depression placed a large droplet these observations was tallied for each droplet, al- (cu 0.1 ml) of one of the solutions or of the nectar though the same ant may have been counted more being tested. Each solution and nectar were rep- than once. resented by four droplets. We prepared a tray and Data were entered in files for each ant-nectar set it down near a colony of whatever ant species interaction, and analyzed using ANOVA, enabling we were testing (usually on concrete substrate) and us to see if either the solution or the trial had a observed it for 1 h, recording the number of ants significant effect. We used a Tukey post hoc test to on each droplet every 5 min. For each trial, the determine which solutions were significantly differ- tray was placed in a different location (presumably ent from each other. Solutions with the shortest 184 Koptur and Truong

TABLE I. Observations of nectar we by pest ant species in south Florida. EFN = extrajoral nectar.

White- Plant species (family) Crazy Carpenter Fire Ghost footed

Adenocalymna comosum (Bignoniaceae) X X X Agave decipiens (Agavaceae) X Aloe saponaria (Liliaceae) X Allamanda nerfolia (Apocynaceae) X Allamanda violacea (Apocynaceae) X Aphelandra jacobiniana (Acanthaceae) X Brownea sp. (Fabaceae) X X Brunflsia grandiflora (Solanaceae) X Calliandra haematocephala (Fabaceae) X Callistemon lilacinus (Myrtaceae) X Callistemon viminolus (Myrtaceae) X Clerodendrum myricoides () X X Clerodendrum speciosissimum (Verbenaceae) X X Clerodendrum wallichii (Verbenaceae) obovata (Verbenaceae) Dipteryx palaamensis (Fabaceae) EFN X Hamelia patens (Rubiaceae) x X Hibiscus tiliaceus (Malvaceae) EFN X X X Ixora coccinea (Rubiaceae) X Justicia fi*lvicoma (Acanthaceae) Kalanchoe gastonis-bonnieri (Crassulaceae) Megaskepasma erythrochlamys (Acanthaceae) Odontonema srrictata (Acanthaceae) X Russelia equisetqormis (Scrophulariaceae) X X Russelia sarmentusa (Scrophulariaceae) x xRutyruspolia sp. (Acanthaceae) X X Sanchezia speciosa (Acanthaceae) X X Schotia brachypetala (Fabaceae) X X Solandra grandijlora (Solanaceae) X Tecoma stans (Bignoniaceae) X Thunberfia erecta (Acanthaceae) X

average time-to-discovery and the largest total ants and fire ants were each recorded taking 11 number of ants were interpreted as being the most nectars. Less frequently encountered were ghost acceptable to the ants being tested. ants (in five nectar-producing species) and white- footed ants (only observed in two species). RESULTS FEEDINGPREFERENCE TEs'rs.--The number Of nec- hiANT vxm.-Five of the major pest ant spe- tar-ant preferences investigated reflects the relative cies in the Miami area were observed taking nectars abundance of the pest ant species and a similar and used in our feeding preference tests. These spe- pattern to the observed nectar use above. We tested cies (in alphabetical order of their common names) 18 nectars with crazy ants, 8 with fire ants, 6 with were: carpenter ants [ Camponotus abdominalis var. carpenter ants, 3 with ghost ants, and 2 with white- floridanus (Buckley)]; crazy ants [Paratrechina lon- footed ants. Since we performed ten trials of each gicornis (Latreille)]; fire ants (Solenopsis invicta Bur- ant-nectar interaction, it was easiest to test many en); ghost ants [ Tdpinoma melanocepbalum (Fabri- nectars with the more common ants. cius)]; and white-footed ants [ Technomyrmex al- Using horizontal bar graphs provides an easy bipes (F. Smith)]. way to visualize the results of the feeding preference tests. Each bar represents the mean of ten trials, NECTARUSE BY PEST mTs.-We observed ants tak- and so has some variation around it (not shown in ing nectar of 31 different cultivated plant species the figures). Analysis of variance and a Tukey post (2 bearing extrafloral nectaries and 29 bearing floral hoc test revealed if there were significant differences nectaries). Crazy ants utilized the nectar of 17 spe- among the solutions being tested, and these are cies of cultivated plants (Table l), while carpenter represented by lower-case letters in the figures. We Nectar Preferences of Pest Ants 185

A /lama nda n eri folia Allamanda violacea

time to discovery

L IiriI

,a 2a 0 5 10 15 20 25 30 35 2b 0 10 20 30 40 50

Clero dendrum myricoides Clerode ndrum spe ciosissimum

time to discovery time to discovery a.b

8", total ants 1a.b

2c 0 5 10 15 20 25 30 35 2d 0 10 20 30 40

Clero de n drum wallic hii Russelia equisetiformis

time to1 discovery time to discovery

total ants total ants

2e 0 5 10 15 20 25 30 2f 0 5 10 15 20 25 30 35 40 FIGURE 2. Results of feeding experiments with carpenter ants, Camponotus abdominalis floridmaus. For explanacion see Figure 1 legend. have grouped the tests by ant species and by al- Significant preferences by crazy ants (Fig. 1) phabetical order of the plant species within these were shown for nectar of Agave decipiens, Hamelia groups. Certain nectars were tested with more than patens, and kklanchoe gastonis-bonnieri. For Agave one ant species, but not all possible nectar-ant tests decipiens (Fig. 1b), time-to-discovery for nectar was were performed because we did not always obtain not significantly shorter than that of fructose, but enough nectar for all desired tests before the plant the total number of ants was greatest. Kahncboe in question went out of bloom, and at some times gastonis-bonnieri nectar time-to-discovery was low could not find the desired ants for experimentation. (Fig. lm), but not significantly different from fruc- 186 Koptur and Truong

Adenocalymna comosum Callian dra ha ema tocep h ala

time to discovery time to discovery

total ants total ants Fb mS i: a 0% 3a 0 5 10 15 20 25 30 35 40 3b 0 10 20 30 40 50 60

Clero dendrum myricoides Hamelia patens (squeezed)

time to discovery time to discovery

total ants total ants

3c 0 10 20 30 40 50 60 3d 0 5 10 15 20 25 30 35

Hamelia patens (2) Odontone ma stricta ta

time to discovery time to discovery

total ants total ants

3e 0 10 20 30 40 50 60 3f 0 10 20 30 40 50 60

FIGURE 3. Results of feeding experirnencs with fire ants, Solenopsis invicta. For explanation see Figure 1 legend.

tose or from the FGS mixture, although the total within 15 or 20 min in all cases). Analyzing results number of ants was significantly greater than for in the normal manner could not take this extreme all the other solutions. Hamelia patens nectar pro- preference into account, so we ran the experiments vided us with an interesting insight: our initial tri- again, replacing the nectar drops when they were als showed ants to have a great preference for the completely consumed. The first retest round nectar over the sugars (and the nectar disappeared showed the nectar to actually appear repellent (Fig. Nectar Preferences of Pest Ants 187

Russe lia equise tifo rmis Sanchezia speciosa

time to discovery time to discovery

total ants total ants

3g 0 10 20 30 40 50 3h 0 10 20 30 40 50

Thunbergia erecfa

time to discovery

FOG total ants

3i 0 10 20 30 40 50 60

~ FIGURE 3. Continued. li), but it was discovered that the nectar had been a preference for more concentrated sugar solutions, contaminated by squeezing the corollas to expedi- fire ants did not exhibit a preference for sucrose in ently collect larger quantities of nectar; retesting our experiments, but rather for fructose, in cases with carefully collected nectar did not reveal the where there was a difference among the sugars marked preference shown by the ants in our earliest (Figs. 3c, 3d, 3f). trials (Fig. lj), though the nectar was preferred over Ghost ants showed significant preferences for all the other solutions except for fructose. The nec- nectars of Bauhinia acuminata and Schotia bra- tar of Crinum moorei (Fig. lg) had the greatest chypetala (Fig. 4).Though the time-to-discovery time-to-discovery, and significantly fewer ants than of the Bauhinia nectar was not significantly dif- any of the sugar solutions, suggesting that this nec- ferent from that of the sucrose solution, the total tar has some repellent properties. In some of the number of ants was significantly different (Fig. experiments (Figs. lj, lp), fructose was preferred 4a). For Schotia, time-to-discovery of the nectar over the other sugar solutions. No other marked was not significantly shorter than that of the preferences were demonstrated. FGS mixture, but the total number of ants was Of the six nectars tested with carpenter ants significantly greater than all the sugar solutions (Fig. 2), none were preferred substantially over the (Fig. 4c). No preferences for any of the sugar sugar standards. No preferences among the sugar solutions were shown in the three experiments standards were exhibited in our tests. conducted. Fire ants showed a preference for the nectar of Only two species were tested with white-footed Calliandra haematocephala; this nectar had the ants, and neither had exceptionally attractive nectar shortest time-to-discovery and the largest total (Fig. 5). In these two experiments, fructose and the number of ants (Fig. 3b). Though previous studies nectars were more attractive than the other sugar using sucrose solutions have shown fire ants to have solutions. 188 Koptur and Truong

Bauhinia acumina ta Callistemon viminolus

time to discovery time to discovery

total ants total ants

4a 0 10 20 30 40 50 60 70 4b 0 10 20 30 40 50

Scho tia bra chype tala

total ants

4c 0 10 20 30 D FIGURE 4. Results of feeding experiments with ghost ants, Tapinoma melanacephalum. For explanation, see Figute 1 legend.

DISCUSSION gies that exclude crawling visitors from flowers, such as narrow tubes constricted at the mouth It appears that, in most cases, the nectars we (Brunfelsia grandtflora, Clerodendrum spp., Ixora studied were accepted as well as sugar solutions coccinea). Many of the species we studied have by ants. In only one case (Crinum moorei) was long, tubular corollas with wide openings (Aloe sa- the nectar less attractive than the sugars; and in ponaria, Kahncboe gastonis-bonnieri, ,Justicia Jitlvi- numerous cases the nectar was more attractive coma, Russelia sarmentusa, Tbunbergia erecta) and than sugars to one or more pest ant species. The we often found ants taking nectar from the intact floral nectars used in our experiments are pro- flowers. Though nectar is a valuable commodity duced by plants with very showy flowers, used to produced by plants for the purpose of rewarding beautify tropical landscapes. The nectars of these pollinators, an overabundance may often be pro- flowers have presumably coevolved with polli- duced as insurance to attract visitors in the face of nators, not ants, and their use by ants (especially few available pollinators or serious competition pest species) is facultative. Our observations con- from other plant species. The relatively small cur with those of Guerrant and Fiedler (1981) amount lost to ants may not warrant protective that floral nectars are not, in general, ant repel- measures in species that make an abundance of lent; however, there are some floral tissues that nectar; in some cases the plants may even benefit contain repellent substances (e.g., Hamelia pat- from the presence of ants utilizing their floral nec- ens), and when the corollas are damaged while tar, as when they use extrafloral nectar (Keeler collecting the nectar, the nectar is tainted with 1989, Koptur 1992). Such potential facultative the repellent substance. protective relationships warrant future investiga- Few of the species we studied have morpholo- tion. Nectar Preferences of Pest Ants 189

Brownea sp. Clero den drum myricoide s

time to discovery time to discovery

, , ,, I, , I ,,,, ,,,,,, ,, , ,,,, , , , , ,' , , b,c , ,

total ants h:b total ants

DFGS 5a 0 10 20 30 40 50 5b 0 10 20 30 40 50 FIGURE 5. Results of feeding experiments with white-footed ants, Echnomyrmex albipes. For explanation, see Figure 1 legend.

ACKNOWLEDGMENTS cussion and facilitated support of this research by the Clo- rox Company. The manuscript was improved by com- We would like to thank Maria Cristina RodrigueL and ments from Paulo Oliveira and two anonymous reviewers, Emiliano Granado for help in the laboratory, and Mark and Jorge Pefia aided in Spanish translation. This paper Deyrup for initial determination of the ant specimens. is contribution 001 to the Program in Tropical Biology Don Bieman and Jules Silverman piovided valuable dis- at Florida International University.

LITERATURE CITED

BAKER,H. G., AND I. BAKER.1978. Ants and flowers. Biotropica lO(1): 80. BriA'rTiF., A. J. 1985. The evolutionary ecology of ant-plant mutualisms. Cambridge University Press, Cambridge, England. FEINSINCXR,I!, AND L. A. SWARM.1978. How common are ant-repellent nectars? Biotropica 10(3): 238-239. GUERRANT,E. O., JR., AND P. L. FIEDLER.1981. Flower defenses against nectar-pilfcragc by ants. Biotropica (Repro- ductive Botany Supplement): 25-33. HAWK,W. A., G. W. FRANKIB,H. G. BAKER,I. BAKMZ,AND S. K~I~IUR.1981. Ants likc flower nectar. Biotropica 13(3): 211-214. Huxi EY, C. R., ANI) D. F. CUILER.1391. Ant-plant interactions. Oxford University Press, Oxford, England. 601 pp. JANZEN,D. H. 1977. Why don't ants visit flowers? Biotropica 9(4): 252. KEELER,K. H. 1989. Ant-plant interactions. In W. G. Abrahamson (Ed.). Plant-animal interactions, pp. 207-242. McGraw-Hill, New York, New York. KERNF,RVON MARILAUN,A. 1878. Flowers and their unbidden guests. Translated into English and edited by W. Ogle. C. Kegan Paul & Co., London, England. 164 pp. KOP.I.UR,S. 1992. Interactions benveen insects and plants mediated by extrafloral nectaries. In E. I3crnays (Ed.). Insect/ plant interactions. Volume 4, pp. 85-132. CRC Press, Boca Raton, Florida. Scirun~~r,13. 0. R., AND A. B. ANDERSON.1978. Why don't ants visit flowers? A reply to D. H.Janzen. Riotropica lO(4): 58-61. VANDFR PIJI.,L. 1955. Some remarks on myrmecophytes. Phytomorphology 5: 190-200.