Nitrogen Xuxes from Treefrogs to Tank Epiphytic Bromeliads: an Isotopic and Physiological Approach

Oecologia (2010) 162:941–949 DOI 10.1007/s00442-009-1533-4

PLANT-ANIMAL INTERACTIONS - ORIGINAL PAPER

Nitrogen Xuxes from treefrogs to tank epiphytic bromeliads: an isotopic and physiological approach

Gustavo Q. Romero · Fausto Nomura · Ana Z. Gonçalves · Natacha Y. N. Dias · Helenice Mercier · Elenice de C. Conforto · Denise de C. Rossa-Feres

Received: 7 March 2009 / Accepted: 30 November 2009 / Published online: 19 December 2009 © Springer-Verlag 2009

Abstract Diverse invertebrate and vertebrate species live with frogs had higher stable N isotopic composition (15N) in association with plants of the large Neotropical family values than those without frogs. Similar results were Bromeliaceae. Although previous studies have assumed obtained from a controlled greenhouse experiment. Linear that debris of associated organisms improves plant nutri- mixing models showed that frog feces and dead termites tion, so far little evidence supports this assumption. In this used to simulate insects that eventually fall inside the bro- study we used isotopic (15N) and physiological methods to meliad tank contributed, respectively, 27.7% (§0.07 SE) investigate if the treefrog Scinax hayii, which uses the tank and 49.6% (§0.50 SE) of the total N of V. bituminosa. Net epiphytic bromeliad Vriesea bituminosa as a diurnal shelter, photosynthetic rate was higher in plants that received feces contributes to host plant nutrition. In the Weld, bromeliads and termites than in controls; however, this eVect was only detected in the rainy, but not in the dry season. These results demonstrate for the Wrst time that vertebrates contribute to bromeliad nutrition, and that this beneWt is sea- Communicated by Zoe Cardon. sonally restricted. Since amphibian–bromeliad associations G. Q. Romero (&) · E. C. Conforto · D. C. Rossa-Feres occur in diverse habitats in South and Central America, this Departamento de Zoologia e Botânica, IBILCE, mechanism for deriving nutrients may be important in Universidade Estadual Paulista (UNESP), bromeliad systems throughout the Neotropics. Rua Cristóvão Colombo, 2265, São José do Rio Preto 15054-000, SP, Brazil e-mail: [email protected] Keywords Bromeliad-frog interactions · Digestive mutualism · Nutrient provisioning · F. Nomura Tillandsioideae · Scinax hayii Departamento de Ecologia, ICB, Universidade Federal de Goiás (UFG), CP 131, Goiânia 7401-970, GO, Brazil Introduction A. Z. Gonçalves Pós-Graduação em Ecologia, Instituto de Biologia, Nutrient Xuxes across ecological compartments can exert Universidade Estadual de Campinas (UNICAMP), CP 6109, Campinas 13083-970, SP, Brazil profound direct and indirect eVects on the recipient system (Purtauf and Scheu 2005). Although transfer of nutrients N. Y. N. Dias from primary producers to higher trophic levels is well Pós-Graduação em Biologia Animal, established, inverse nutrient Xux, i.e., from animals to Universidade Estadual Paulista (UNESP), Rua Cristóvão Colombo 2265, plants, is a much less recognized process (e.g., Anderson São José do Rio Preto 15054-000, SP, Brazil and Polis 1999). Nutrient Xuxes from animals to plants can occur at multiple scales. For example, on a broad scale, sea H. Mercier bird guano improves the net primary productivity of plants Departamento de Botânica, Instituto de Biociências, Universidade de São Paulo (USP), Rua do Matão, on oceanic islands (Anderson and Polis 1999; Wait et al. 277, São Paulo 05508-900, SP, Brazil 2005), and resource pulses of cicadas (Magicicada spp.) in 123 942 Oecologia (2010) 162:941–949

17-year cycles improve growth and reproduction in temperate plants as diurnal shelters, diVerent degrees of specialization forest plants (Yang 2004). On a smaller scale, several can be noted. For instance, Aparasphenodon brunoi has a studies have reported nutrient Xuxes from ants to myrme- hyperossiWed helmet-like cranium that protects the species cophilous plants (Rico-Gray et al. 1989; Treseder et al. against predators in the bromeliad’s tanks (Teixeira et al. 1995; Sagers et al. 2000; Fischer et al. 2003; Solano and 2002), while Scinax hayii, also common in bromeliads, has Dejean 2004), from mutualistic Pameridea bugs (Miridae) no evident morphological modiWcations (Carvalho-e-Silva to the carnivorous host plant Roridula (Anderson and et al. 2000; F. Nomura, personal observation). Midgley 2002, 2003), and from the jumping spider Psecas In southeastern Brazil, the treefrog Scinax hayii chapoda to its Bromeliaceae host plant, Bromelia balansae (Hylidae) commonly uses the tank epiphytic bromeliad (Romero et al. 2006, 2008). Although diverse invertebrates Vriesea bituminosa (Tillandsioideae) as a diurnal shelter, and vertebrates live associated with plants of the large Neo- thus comprising a suitable system in which to test N Xuxes tropical family Bromeliaceae, little is known about whether and reciprocal beneWts in frog-bromeliad systems. In addi- and how these animals contribute to their host plants’ per- tion, insects that fall into bromeliad tanks might also con- formance. tribute to bromeliad nutrition. In this study we conducted Many species in the family Bromeliaceae typically live Weld observations and surveys, and performed a controlled in low-nutrient environments (rock outcrops, sandy soils, greenhouse experiment during two seasons (dry and wet) to tree canopies), but have their leaves organized in rosettes, address the following questions: an arrangement that forms a tank or phytotelmata, which 1. Does S. hayii contribute to V. bituminosa nutrition allows them to intercept and retain debris and water through its debris (feces)? (Benzing 2000). Minerals and water accumulated in the 2. Does the bromeliad absorb N derived from insects that rosette can be absorbed through specialized trichomes die inside its tanks? (Sakai and Sanford 1980; Benzing et al. 1985). Whereas 3. How much of the bromeliad’s N is derived from tree- plants of the subfamilies Pitcairnioideae and Bromelioi- frog feces and from dead insects? deae seem to be unable to absorb large organic mole- 4. How does N uptake from treefrogs and dead insects cules through their trichomes, plants of the subfamily inXuence the physiology of the bromeliad (e.g., gas Tillandsioideae (e.g., Vriesea) can take up complex exchanges, protein and chlorophyll concentrations)? organic compounds, such as amino acids (Endres and 5. Do these responses vary in diVerent seasons? Mercier 2003; Cambuí and Mercier 2006). Therefore, bromeliads of this subfamily could derive nutrients from predator feces and arthropod carcasses in their tanks Materials and methods (Romero et al. 2006). Many anuran species from South and Central America Study system use bromeliads during their life cycles, although the degree of specialization in the use of this microhabitat varies among Vriesea bituminosa (Tillandsioideae) is a tank epiphytic species (Peixoto 1995; Richardson 1999; Carvalho-e-Silva bromeliad that commonly inhabits trees in Brazilian rain- et al. 2000; Schineider and Teixeira 2001; Teixeira et al. forests, but can also occur in inselberg and cerrado vegeta- 2002; Haddad and Prado 2005). For example, some species tion (Romero 2006; Versieux and Wendt 2007; G. Q. are specialized, spending their entire life cycle in associa- Romero, personal observation). Its rosettes are large and tion with bromeliads, reproducing and feeding among the colorful, and accumulate large amounts of rainwater (range plant axils, and producing tadpoles with morphological and for four small bromeliads: 240–260 ml). Because of these behavioral adaptations [e.g., Syncope antenori (Krungel traits, it is preferred by landscape designers for use in pri- and Richter 1995); Phyllodytes luteolus (Giaretta 1996)]. vate gardens, and is frequently extracted from the wild for Other species use bromeliads during the reproductive this type of commercial use; thus it is an endangered spe- period, as calling or oviposition sites or as microhabitats for cies (Versieux and Wendt 2007). This plant’s rosette can tadpole development [e.g., Physalaemus spiniger (Haddad shelter several animal groups, including spiders, ants and and Pombal 1998); Aplastodiscus sibilatus (Cruz et al. frogs (Romero 2006; G. Q. Romero, personal observation; 2003); Dendrobates pumilio (Young 1979)]. At the other F. Nomura, unpublished data). extreme, some anurans use bromeliads only as diurnal shel- The treefrog Scinax hayii (Hylidae) frequently uses ters [e.g., Eleutherodactylus johnstonei (Ovaska 1991); V. bituminosa as a diurnal shelter. This is a common, Aparasphenodon brunoi (Teixeira et al. 2002); Dendropsophus medium-sized frog (about 43 mm total length) belonging to nahdereri and Scinax perereca (Conte and Rossa-Feres the Scinax ruber clade (Faivovich et al. 2005), which is 2006); and Scinax hayii (F. Nomura, personal observa- endemic to the Atlantic rainforest and occurs from Espírito tion)]. Even in those anuran species that only use bromeliad 123 Oecologia (2010) 162:941–949 943

Santo to Santa Catarina states, southeastern Brazil. This the frogs ate prior to capture, the Wrst feces produced were species lives in primary and secondary lowland and mon- discarded, and the remaining feces were randomly applied tane forests, along forest edges, in secondary vegetation, in the experimental bromeliads (treatment 1 below). Ter- and even inside houses. Usually it occurs on low vegetation mites were also collected, dried, weighed and stored to be near streams and ponds (Carvalho-e-Silva and Carvalho- used in the experiment. The experiment had three treat- e-Silva 2004). It breeds in pools along streams and ponds ments: six bromeliads received frog feces inside the tank, between September and March, and often shelters in bro- six bromeliads received termites, and six control bromel- meliad rosettes during the day (F. Nomura, personal obser- iads received no organic matter. For the experimental treat- vation). It commonly returns to the same bromeliad after a ments, either one feces acorn (mean § 1 SD of the dry night of reproduction (F. Nomura, unpublished data). Indi- weight: 0.152 g § 0.011; n = 50) or the same amount of dry viduals are usually found alone in the bromeliads, but weight of termites was deposited in the central tank of bro- sometimes a single bromeliad can shelter up to three indi- meliads at 2-day intervals over 103 days, from 25 January viduals of S. hayii (F. Nomura, personal observation). to 17 May 2006. Experimental bromeliads were cultured in the green- Nutrient Xuxes from frogs to plants: tests using isotopic house, and thus had no previous contact with animals. They methods were of the same cohort, and were cultured in pots (20 cm width, 15 cm height) containing vegetal earth maintained in Field surveys shade, under a 50% sunlight regime to simulate their natural environments. The size of the experimental bromeliads We Wrst investigated the isotopic signature of 15N (natural was similar (leaf length: 25 cm) and corresponded to those abundance) of Weld V. bituminosa bromeliads that were in nature which support only one frog. Bromeliads were either inhabited by treefrogs or uninhabited. Bromeliads of watered through an automatic irrigation system by a Wne similar size, at heights varying from 0.85 to 8.0 m, were spray using three sprinklers, each with a capacity of 8 l/h, randomly inspected and the presence/absence of frogs was which worked for 15 min every 2 h. recorded. These epiphytic bromeliads inhabited diverse At the end of the experiment (17th May), we collected host species and likely accumulated litter fallen from vari- two new leaves (Wrst node) of each experimental bromeliad able host tree types. Of the 31 bromeliads inspected, ten for isotopic determinations (15N). had frogs. For each bromeliad inspected we collected one randomly selected leaf from the median layer (third–fourth Isotopic analyses node) for isotopic analysis. The survey was done in December 2006 near a natural lake at a humid mountain summit The bromeliad leaves from both the Weld and the green- (1,291 m a.s.l.) in the Atlantic rainforest (i.e., ombrophitic house experiments were washed for at least 3 min in a cur- dense high-montane forest) near Atibaia city (23º10ЈS; rent of water and scrubbed by hand to eliminate 46º31ЈW), São Paulo State, in southeastern Brazil. The contamination (organic particles or mites). Bromeliad local climate shows a distinct dry/cold (May–September) leaves were oven-dried for 30 h at 65°C, ground to a Wne and a wet–warm (October–April) season. Mean annual powder in a ball mill, and transferred to airtight containers. rainfall is 128 mm and mean annual temperature is 20°C Frog feces and muscle tissue, as well as samples of termites (CIIAGRO 2008, http://www.ciiagro.sp.gov.br). used in the experiment (n = 5) were frozen and dried for isotopic determinations. Stable isotope analyses were done Experimental design by the Stable Isotope Facility at the University of California at Davis. Stable isotope ratios of N and C, as well as N con- The N Xux from frogs and insects to V. bituminosa was centration (g of total N/mg of dried plant tissue), were experimentally tested in a greenhouse at the Universidade determined by continuous Xow isotope ratio mass spec- Estadual Paulista, São José do Rio Preto city. The feces trometer (20-20 mass spectrometer; PDZ Europa, Sand- added to the plants were obtained from frogs of S. hayii fed bach, England) after sample combustion to CO2 and N2 at with workers of the termite Nasutitermes sp. (Termitidae) 1,000°C by an on-line elemental analyzer (PDZ Europa collected in a Weld at the University. To obtain feces, 11 ANCA-GSL). Isotope ratios of C for leaves of Weld and frogs (adults) were collected in the Weld (Atibaia city). greenhouse bromeliads were determined to characterize the W They were kept in ve 65 £ 40 £ 50-cm terraria in the lab- photosynthetic pathway (i.e., C3 or CAM) of V. bituminosa. oratory and fed termites ad libitum for 64 days. The frog Natural 15N and 13C abundances were calculated using 15 feces were collected daily, dried in a vacuum oven, N=(Rsample/Rstandard –1)£ 1,000 (‰ vs. At-air) and 13 weighed and then stored individually in polypropylene C=(Rsample/Rstandard –1)£ 1,000 (‰ vs. Vienna-Pee tubes. To avoid or minimize any eVect of imprint of food Dee belemnite), where 15N is the stable N isotopic 123 944 Oecologia (2010) 162:941–949 composition, R is the ratio of mass 15N/14N and 13C/12C, the supernatant (15 l) was mixed with 185 l of the respectively. Comassie Brilliant Blue G-250 dye solution, obtained from 100 mg of the dye dissolved in 95% ethanol with 100 ml of Physiological responses 85% phosphoric acid. After 5 min, absorbance was read in a spectrophotometer at 595 nm. A calibration curve of Gas exchange analyses protein concentration was made with bovine serum albumin. Net photosynthetic rate per leaf unit area (A), transpiration rate (E), stomatal conductance (gs), intercellular CO2 con- Statistical analyses V centration (Ci), and vapor pressure di erence between the plant and the environment (w) were measured on intact, 15N and N concentration values were compared between attached leaves of all experimental bromeliads in the green- Weld bromeliads with and without S. hayii using analysis of house using an ADC LCA-4 infra-red portable gas analyzer covariance (ANCOVA), with presence/absence of frogs as system and PLCB-4 leaf chamber (ADC, Hoddesdon, UK). a Wxed factor (one level) and distance of the bromeliads The chamber was positioned at the central region of mature from the lake and height of the bromeliads on trees as the leaves from the median rosette nodes (third–fourth node). covariates; the dependent variable presented homogeneous Gas exchange data were recorded as soon as readings variance (Cochran’s test; P = 0.815). Data on 15N and N became stable, usually 60–120 s after leaf insertion into the concentration values of experimental bromeliads were chamber. Three to Wve replications were carried out for compared among treatments using ANOVA, with treatment each experimental rosette, but only mean values were used as a Wxed eVect (two levels). Data on 15N presented homo- in the statistical analyses. Measurements were done on geneous variance (Cochran’s test; P = 0.485); data on N clear days at two experimental periods, one at the end of concentration were log transformed to equalize the vari- rainy season (12 April 2007) and one in the dry season (17 ances. Fisher least square diVerence (LSD) post-hoc test May 2007), from 8:00 to 10:30 a.m. During the measure- was used for pairwise comparisons. ments air temperature and relative air humidity were To determine the fraction of N (% Ndf feces or termites) that 31.2 § 0.1°C and 48.3 § 0.4% (SE) on 12 April, and experimental bromeliads derived from soil, termites and 31.5 § 0.3°C and 38.9 § 0.7% on 17 May. Mean photo- frog feces (mixture), we used linear mixing models devel- synthetic photon Xux density on 12 April and 17 May was oped by Phillips and Gregg (2001). Through sensitivity 1,548 § 17.7 and 1,131 § 50.2 mol m¡2 s¡1 (SE), respec- analysis, this model assesses the relative importance of the tively. isotopic signature diVerence between two sources (e.g., soil and frog feces, or soil and termite carcasses), the SD of iso- Chlorophyll and protein analyses topic signatures in the sources and mixed populations, sample size, analytical SD, and evenness of the source Immediately after the leaf gas exchange measurements on populations to determine the SE of source proportion esti- 17 May, leaves were collected for biochemical analyses. mates (Phillips and Gregg 2001). Mean 15N values of Concentrations of chlorophyll a and b, carotenoid, and total leaves of V. bituminosa that received termites or feces from soluble protein were determined from the same bromeliad frogs that fed on termites were used as the signature for leaf. Total chlorophyll (a + b) was obtained from fresh source mixture, mean 15N values of leaves of the control leaves (1 g), by cutting them into little pieces, freezing in plants were used as the soil end-member, and mean 15N liquid N and homogenizing at ca. 4°C in 7 ml of cold 80% values for termites or feces from frogs that fed on termites acetone (v:v, acetone:water). The homogenized product were used as the animal end-member. Fractionation for was Wltered through a paper funnel previously sprayed with plant absorption of animal debris is largely unknown. How- 2 ml of cold 80% acetone, and the residue was washed 3 ever, an unpublished study (A. Z. Gonçalves, et al.) times with 4 ml of cold 80% acetone; the volume was detected only an insigniWcant fractionation in N Xuxes from topped up to 20 ml with acetone at 80%. The absorbance Drosophila Xies and spider feces (guanine) to the three bro- was measured in a spectrophotometer (Ultrospec 3000; meliad species B. balansae, Aechmea distichantha and Cambridge, England) at 470, 647, and 663 nm and then cal- Ananas comosus; Thus, fractionation of N was not calcu- culated through equations developed by Lichtenthaler lated here. (1987). Data on A, E, gs, Ci, and w were log10 or log10 (n +1) Total soluble protein was determined according to Brad- transformed for variance homogenizations and then com- ford (1976). Fresh leaves (1 g) were cut into little pieces, pared using repeated measures ANOVA, with treatment as frozen in liquid N and homogenized in ca. 3 ml of ultra- a Wxed eVect (two levels) and sampling dates (12 April and pure water. After centrifugation at 12,000 rpm for 10 min, 17 May 2007) as the repetition factor. Separate ANOVAs 123 Oecologia (2010) 162:941–949 945 were also performed to test the inXuence of feces and ter- feces and termites (ANOVA/LSD; P = 0.339; Fig. 1c). mites for speciWc sampling dates. Data on chlorophyll a, Similar results were also obtained for N concentration chlorophyll b, chlorophyll a + b, carotenoid and soluble (F2, 15 =3.95, P = 0.042; pairwise comparisons, feces vs. protein concentrations ( g/g fresh leaf mass) were log10 control, P = 0.032; termites vs. control, P = 0.025; feces vs. transformed and compared among treatments using one- termites, P =0.897; Fig.1d). way ANOVA. When necessary, Fisher LSD post-hoc test The 15N values of frog feces, termites and frog tissue was used for paired comparisons. were 6.15 § 1.1 (SE), 3.22 § 0.12 and 0.88 § 0.65 (n =5), All ANOVAs were run using type III sums of squares. respectively. Using the linear mixing model we estimated All statistical analyses were performed using GLM. The that frog feces and dead termites contributed 27.7% (§0.07 mean values (§1 SE) presented in the Wgures, tables, and SE) and 49.6% (§0.50 SE), respectively, to the total N of text, were computed directly from untransformed data. V. bituminosa. The feces and termite bodies, including soft and hard (chitin) parts, were decomposed after 4–6 days of application inside the bromeliad tanks. Results V. bituminosa had 13C values varying from –26.01 to

–30.29‰ (n = 31), corresponding to the range of the C3 W 15 V In the eld, bromeliads with frogs had higher N values bromeliads. A per leaf unit area, , gs, Ci, and w did not di er than those without frogs (ANCOVA: F1, 25 =5.31, among the treatments in the repeated measures ANOVAs P = 0.029; Fig. 1a). In contrast, the N concentration in the (Table 1). However, separate one-way ANOVAs for season leaf tissues did not diVer between bromeliads with and showed that A, E, and w diVered among treatments without frogs (ANCOVA; F1, 23 = 0.46, P = 0.50; Fig. 1b). (Fig. 2) during the rainy season (A, F2,15 =4.13, P = 0.037; The covariates Distance from the margin and Height of the E, F2,15 =4.26, P =0.034; w, F2,15 =3.74, P = 0.048), but bromeliad, as well as the interaction terms between these not at the beginning of the dry season (Fig. 2). A and E covariates and treatment (presence/absence of frogs) did were higher in plants that received feces and termites than not aVect the results of 15N (P ¸ 0.36) or N concentration in the control plants during the rainy season (Fig. 2). 15 (P ¸ 0.65). In the greenhouse experiment, N values of Though E, gs, and Ci increased from the wet (12 April) to V bromeliad leaves di ered among treatments (F2, 15 =10.5, the dry season (17 May), the A of bromeliads that received P =0.001). 15N values were higher after treatment with feces and termites decreased to levels of control plants in frog feces and termites compared to control plants this period (Fig. 2; Table 1). (ANOVA/least signiWcant diVerence; feces versus control, Concentrations of chlorophyll a, chlorophyll b, chloro- P < 0.001; termites vs. control, P = 0.004; Fig. 1c), and phyll a + b, and carotenoids tended to be higher in the frog there was no statistical diVerence between treatments with feces treatment, but the values did not diVer signiWcantly

Fig. 1 Mean a stable N isotopic 0.4 a 35 b composition (15N) and b total N 30 concentration (15N+14N) values 0.2 in leaf tissues from Weld-grown 25 0 bromeliads with or without 15 20 frogs, and mean c N and d -0.2 total N concentration in leaf 15 tissues from greenhouse-grown -0.4 10 bromeliads that received either no supplementation (control), -0.6 5 termites, or the feces of frogs -0.8 that were fed termites. DiVerent 0 Frogs absent Frogs present Frogs absent Frogs present letters indicate signiWcant diVer- N (‰) ences [P < 0.05; ANOVA/Fish- 15 3 30 V δ c b d er least square di erence (LSD) b b post hoc test; =0.05]. Error 2.5 25 b bars indicate § 1 SE a 2 20 Total N concentration (µg/mg) 1.5 a 15

1 10

0.5 5

0 0 Control Termites added Feces added Control Termites added Feces added

123 946 Oecologia (2010) 162:941–949

Table 1 Repeated measures ANOVAs examining the eVect of termite N-rich compound (urea; Lehninger et al. 1993; Duellman and frog feces on bromeliad physiological variables: net photosyn- and Trueb 1994), in our experiment, we suggest that the thetic rate per leaf area unit (A), transpiration rate (E), stomatal conduc- contribution of frogs to bromeliad nutrition might be even g C tance ( s), intercellular CO2 concentration ( i), and vapor pressure W diVerence between the plant and the environment (w) higher in the eld. This anuran species primarily uses bromeliads as a diurnal shelter, and should excrete an amount Source of variation df MS FP of their feces and urine in reproductive and foraging sites A when outside bromeliads (i.e., lakes, ponds, streams). In Treatment 2 0.03163 1.88 0.187 contrast, several other amphibian species are more special- Error 15 0.01685 ized in bromeliads and use these microhabitats throughout Time 1 0.04629 7.61 0.015 their life cycle as reproductive, foraging, calling, and egg- Time £ Treatment 2 0.02038 3.35 0.062 laying sites, and as microhabitat for tadpole development Error 15 0.00608 (Young 1979; Krungel and Richter 1995; Giaretta 1996; E Haddad and Pombal 1998; Cruz et al. 2003). In these frog- Treatment 2 0.061346 3.06 0.077 bromeliad systems, bromeliads likely derive more nutrients from frogs. This should be a suitable theme for future Error 15 0.020074 research. Time 1 0.100103 7.80 0.014 Whereas frog feces contributed 27.7% of the total N of Time £ Treatment 2 0.008458 0.69 0.532 the Tillandsioideae species V. bituminosa, dead termites Error 15 0.012839 used to simulate insects that fall into bromeliad tanks con- g s tributed 49.6% of the bromeliad N. Contrasting results were Treatment 2 0.000039 1.51 0.252 obtained by Romero et al. (2006), which showed that a ter- Error 15 0.000025 restrial bromeliad Bromelia balansae (Bromelioideae) Time 1 0.000495 27.48 <0.001 derived only 3% of N from insect carcasses, whereas it Time £ Treatment 2 0.000007 0.39 0.685 derived 15% from spider feces. It is well established that Error 15 0.000018 terrestrial bromeliads from the subfamily Bromelioideae C i without extensive phytotelmata (e.g., Bromelia, Ananas) Treatment 2 0.0018 0.39 0.685 bear less specialized absorptive leaf trichomes and are Error 15 0.0045 assumed to depend mostly on their roots for soil inorganic Time 1 0.2384 91.48 <0.001 nutrient acquisition (e.g., Endres and Mercier 2003). In Time £ Treatment 2 0.0022 0.86 0.443 contrast, tank-bromeliads with epiphytic habits from the Error 15 0.0026 subfamily Tillandsioideae (e.g., Vriesea) have specialized w foliar trichomes, thus are better adapted to use larger Treatment 2 0.04204 3.33 0.064 organic molecules containing N (Owen and Thomson 1988; Error 15 0.01264 Endres and Mercier 2001). Our Wndings support the Time 1 0.08800 8.91 0.009 hypothesis of Romero et al. (2006), which suggested that Time £ Treatment 2 0.00999 1.01 0.387 tank-bromeliads may beneWt even more from animal nutri- Error 15 0.00988 ent input than do terrestrial bromeliads. V The sample treatments were performed on two sampling dates (12 The contrasting di erence between feces and termite April and 17 May 2007) contribution to bromeliad nutrition (27.7 vs. 49.6%) was unexpected and can be partially explained by the fact that urine was not used in the experiments, as mentioned above. among the treatments (Table 2). In contrast, total soluble In the digestion process, an amount of termite N could have protein concentration diVered among the treatments been sequestered in frog urine, or assimilated by the preda- (F2, 15 =14.39, P < 0.001), and was higher for bromel- tor. Even so, we believe that Weld bromeliads should not iads that received termites and the control than those that take up a large quantity of N derived from insect carcass. received feces (Table 2). Only small numbers of insects that fall inside the bromeliad tank die; many survive by climbing out of the tank onto the rosette leaves (G. Q. Romero, unpublished data). In con- Discussion trast to carnivorous bromeliads, which attract insects by releasing nectar-like odors from extra-Xoral nectaries in the Our results clearly demonstrated a N Xux from the associ- rosettes [e.g., Brocchinia reducta, Pticairnioideae (Givnish ated treefrog S. hayii to the tank epiphytic bromeliad et al. 1984)], V. bituminosa has no apparent mechanism of V. bituminosa. Since we did not use frog urine, which is a insect attraction (G. Q. Romero, personal observation). 123 Oecologia (2010) 162:941–949 947

Fig. 2 Values of net photosyn- 6 a thetic rate per leaf area unit ) Control A a E b -1

( ; ), transpiration rate ( ; ), s 5 a stomatal conductance (g ; c), -2 Termites added s a m intercellular CO2 concentration 2 4 (Ci; d), and vapor pressure Feces added V di erence between the plant and b the environment ( w; e) of mol CO 3 Vriesea bituminosa rosettes that µ received termites, frog feces or A ( neither (control), measured on 2 two sampling dates (12 April April, 12 May, 17 and 17 May 2007). DiVerent W 2.1 b 0.07 c

letters indicate signi cant ) ) -1 diVerences (P < 0.05; ANOVA/ -1

s 0.06 s

1.7 -2

Fisher LSD post hoc test; -2 = 0.05). Error bars a

Om 0.05

a 2 § O m indicate 1 SE 2 1.3 0.04 b 0.9 0.03 gs (mol H E (mmol H 0.5 0.02 April, 12 May, 17 April, 12 May, 17

300 14

) d e -1

s 260 12 -2 m

2 a 220 10

w ab ∆ 180 8

mol CO mol b µ 140 6 [Ci] ( 100 4 April, 12 May, 17 April, 12 May, 17

Table 2 Chlorophyll a, chlorophyll b, chlorophyll a + b, carotenoid and total soluble protein concentrations (g/g fresh leaf mass) of bromeliads grown under three treatments (control, termites added and feces added) Parameters Control Treatments

Termites added Feces added

Chlorophyll a 337.0 (13.9) a 352.7 (29.9) a 397.2 (46.7) a Chlorophyll b 159.3 (7.7) a 153.6 (18.1) a 193.4 (30.1) a Chlorophyll a + b 496.4 (20.5) a 506.3 (35.5) a 590.6 (76.3) a Carotenoids 35,651.5 (2368.7) a 37,585.4 (3243.4) a 41,460.8 (4442.8) a Total soluble protein 479.0 (41.1) a 442.3 (42.8) a 240.8 (29.5) b Each value represents the mean of six replicates, and SEs are indicated in parentheses. DiVerent letters indicate signiWcant diVerences (P < 0.05; ANOVA/Fisher least square diVerence post hoc test; =0.05)

However, it has a tank Wlled by rain water that can be used larger organic compounds not completely digested by the as diurnal shelter against predators and/or desiccation by frogs. Mineralization of N compounds could be further associated insectivorous predators, like amphibians. By for- accelerated by microorganisms living in the tanks of Vriesea aging outside the bromeliads, frogs can concentrate a con- bromeliads (see Inselsbacher et al. 2007). siderable amount of N-rich compounds (feces and urine) Bromeliads from the Weld that sheltered frogs had higher inside the tank bromeliad when returning to their diurnal 15N values than those without frogs. However, while Weld shelter, in an analogous way to carnivorous plants. The bromeliads in the absence of frogs had similar total N con- water from the tank may thus help the plant digest and use centration compared to those with frogs, control plants 123 948 Oecologia (2010) 162:941–949 from the experiment had smaller N concentrations than most relevant abiotic constraint for growth and vegetative those that received termites or feces. These results indicate functions is water shortage, while other factors, such as that Weld bromeliads take up a certain amount of N from nutrient availability, are generally of less importance frogs (data on 15N), but in the absence of these vertebrates (Laube and Zotz 2003). Our results suggest that even in the they do not become nutritionally depleted. In the Weld, presence of N-rich compounds derived from amphibian these bromeliads likely take up organic compounds or min- debris, the plants did not beneWt during moisture-limited erals derived from litter, which accumulate in large periods. amounts in their tank and is quickly decomposed by detriti- As previously thought, frogs associated with epiphytic vores (see Ngai and Srivastava 2006). Similarly, Romero bromeliads can contribute to host plant nutrition and perfor- et al. (2008) showed that B. balansae from open grasslands mance (Endres and Mercier 2001). In addition, our results were associated with the spider Psecas chapoda and had suggest that tank epiphytic bromeliads may derive more N higher 15N values compared to forest bromeliads; how- from associated animals than terrestrial bromeliads, which ever, forest bromeliads had higher N concentrations. There- lack tanks (see Romero et al. 2006). Although N derived fore, bromeliads in the Weld seem to use diVerent sources of from frog feces improved the net photosynthetic rate of N. The inXuence of variable N sources (i.e., animal or vege- V. bituminosa, this type of beneWt provided by frogs was tal) on bromeliad nutrition and growth deserves more atten- temporally restricted and seems to be related to water stress tion, and is a suitable theme for future research. during the dry season. This is the Wrst study to show nutri- Studies have independently reported that several ent provisioning in a frog-bromeliad system. Because several amphibian species live in association with bromeliads amphibians live associated with Bromeliaceae in South and (see “Introduction”), and that urea supply, a soluble com- Central America, this nutrient transfer phenomenon may be pound rich in amphibian excretes (Lehninger et al. 1993; common throughout the Neotropics. + Duellman and Trueb 1994), can increase free-NH4 and total amino acids in tissues of Vriesea gigantea (Endres and Acknowledgments The authors thank J. Purcell for comments on Mercier 2001). In this study we show that N derived from the manuscript and language corrections, and J. C. Souza for help with data collection in the Weld. C. A. Cambuí and R. P. Andreoli helped feces of bromeliad-dwelling frogs in fact improves the net with the protein and gas exchange analyses, respectively, and photosynthetic rate of V. bituminosa. Intriguingly, bromel- G. Martinelli identiWed the bromeliad. G. Q. Romero was supported by iads that received frog feces had soluble proteins decreased research grants from Fundação de Amparo à Pesquisa do Estado de São by a half. This result suggests that V. bituminosa is able to Paulo (FAPESP; 04/13658-5 and 05/51421-0). F. Nomura was supported V by research grants from Coordenação de Aperfeiçoamento de Pessoal store the N di erently depending on the source accumu- de Nível Superior (CAPES; 3300415-3). lated in the tank. Bromeliads in which the only source was soil or soil plus termite may have stored N as soluble proteins, while bromeliads that received feces may accumulate References N as amino acids (e.g., asparagine), which are not detected by the Bradford method (see “Materials and methods”). Anderson B, Midgley JJ (2002) It takes two to tango but three is a tangle: mutualists and cheaters on the carnivorous plant Roridula. Endres and Mercier (2003) showed that V. gigantea grown Oecologia 132:369–373 + with NH4 , glutamine or glycine accumulated large Anderson B, Midgley JJ (2003) Digestive mutualism, an alternate amounts of asparagine (Asn). Accumulation of N sources pathway in plant carnivory. Oikos 102:221–224 as Asn may be a suitable way to improve growth in epi- Anderson WB, Polis GA (1999) Nutrient Xuxes from water to land: seabirds aVect plant nutrient status on Gulf of California Islands. phytic bromeliads, once transamination reactions easily Oecologia 118:324–332 transform Asn to other amino acids (Endres and Mercier Benzing DH (2000) Bromeliaceae: proWle of an adaptive radiation. 2003). Cambridge University Press, Cambridge Several bromeliad species are tank epiphytes, and there- Benzing DH, Givnish TJ, Bermudes D (1985) Absorptive trichomes in Brocchinia reducta (Bromeliaceae) and their evolutionary and fore have no contact with the pedosphere and thus need to systematic signiWcance. Syst Bot 10:81–91 obtain N from other sources. Many epiphytic bromeliads Bradford MM (1976) Rapid and sensitive method for quantitation of live in association with amphibians, like V. bituminosa, microgram quantities of protein utilizing principle of protein-dye thus frog-supplied N could provide a major beneWt to these binding. Anal Biochem 72:248–254 W Cambuí CA, Mercier H (2006) Organic nitrogen nutrition in plants. plants. However, the bene t provided by frogs observed Curr Top Plant Biol 7:13–17 here was temporally restricted, i.e., photosynthetic rate was Carvalho-e-Silva SP, Carvalho-e-Silva AMT (2004) Scinax hayii. high in the wet season, but decreased to control levels in the IUCN 2008. 2008 IUCN red list of threatened species. http:// dry season. Decreased photosynthetic rate under drought www.iucnredlist.org. Downloaded 22 December 2008 Carvalho-e-Silva SP, Izecksohn E, Carvalho-e-Silva AMPT (2000) conditions has been reported for other bromeliad species Diversidade e ecologia de anfíbios em restingas do sudeste brasile- (reviewed in Martin 1994). It is well established that, for iro. In: Esteves FA, Lacerda LD (eds) Ecologia de restingas e lagoas vascular epiphytes (including Vriesea sanguinolenta), the costeiras. NUPEM/UFRJ, Macaé, Rio de Janeiro, pp 89–97 123 Oecologia (2010) 162:941–949 949

Conte CE, Rossa-Feres DC (2006) Diversidade e ocorrência temporal Owen TP Jr, Thomson WW (1988) Sites of leucine, arginine, and gly- da anurofauna (Amphibia, Anura) em São José dos Pinhais, cine accumulation in the absorptive trichomes of a carnivorous Paraná, Brasil. Rev Bras Zool 23:162–175 bromeliad. J Ultrastruct Mol Struct Res 101:215–223 Cruz CAG, Pimenta BVS, Silvano DL (2003) Duas novas espécies Peixoto OL (1995) Associação de anuros a bromeliáceas na Mata pertencentes ao complexo de Hyla albosignata Lutz & Lutz, Atlântica. Rev Univ Rural Ser Ciênc Vida 17:75–83 1938, do leste do Brasil (Amphibia, Anura, Hylidae). Bol Mus Phillips DL, Gregg JW (2001) Uncertainty in source partitioning using Nacl Zool 503:1–13 stable isotopes. Oecologia 127:171–179 Duellman WE, Trueb L (1994) Biology of amphibians. Johns Hopkins Purtauf T, Scheu S (2005) Nutrient and resource dynamics and food University Press, Baltimore webs: understanding the mutual relationships between the Endres L, Mercier H (2001) Ammonium and urea as nitrogen sources dynamics of food webs, resources, and nutrients. In: Ruiter P, for bromeliads. J Plant Physiol 158:205–212 Wolters V, Moore JC (eds) Dynamic food webs: multispecies Endres L, Mercier H (2003) Amino acid uptake and proWle in bromel- assemblages, ecosystem development, and environmental change. iads with diVerent habitats cultivated in vitro. Plant Physiol Academic Press, MA Biochem 41:181–187 Richardson BA (1999) The bromeliad microcosm and the assessment Faivovich J, Haddad CFB, Garcia PCA, Frost DR, Campbell JA, of faunal diversity in a neotropical forest. Biotropica 31:321–336 Wheeler WC (2005) Systematic review of the frog family Hyli- Rico-Gray V, Barber JT, Thien LB, Ellgaard EG, Toney JJ (1989) An dae, with special reference to Hylinae: phylogenetic analysis and unusual animal-plant interaction: feeding of Schomburgkia taxonomic revision. Bull Am Mus Nat Hist 294:1–240 tibicinis (Orchidaceae) by ants. Am J Bot 76:603–608 Fischer RC, Wanek W, Richter A, Mayer V (2003) Do ants feed Romero GQ (2006) Geographic range, habitats, and host plants of bro- plants? A 15N labeling study of nitrogen Xuxes from ants to plants meliad-living jumping spiders (Salticidae). Biotropica 38:522–530 in the mutualism of Pheidole and Pipier. J Ecol 91:126–134 Romero GQ, Mazzafera P, Vasconcellos-Neto J, Trivelin PCO (2006) Giaretta AA (1996) Reproductive specializations of the bromeliad Bromeliad-living spiders improve host plant nutrition and growth. hylid frog Phyllodytes luteolus. J Herpetol 30:96–97 Ecology 87:803–808 Givnish TJ, Burkhardt EL, Happel RE, Weintraub JD (1984) Carni- Romero GQ, Vasconcellos-Neto J, Trivelin PCO (2008) Spatial varia- vory in the bromeliad Brocchinia reducta, with a cost/beneWt tion in the strength of mutualism between a jumping spider and a model for the general restriction of carnivorous plants to sunny, terrestrial bromeliad: evidence from the stable isotope 15N. Acta moist, nutrient-poor habitats. Am Nat 124:479–497 Oecol 33:380–386 Haddad CFB, Pombal JP (1998) Redescription of Physalaemus spiniger Sagers CL, Ginger SM, Evans RD (2000) Carbon and nitrogen isotopes (Anura: Leptodactylidae) and description of two new reproductive trace nutrient exchange in an ant-plant mutualism. Oecologia modes. J Herpetol 32:557–565 123:582–586 Haddad CFB, Prado CPA (2005) Reproductive modes in frogs and Sakai WS, Sanford WG (1980) Ultrastructure of the water-absorbing their unexpected diversity in the Atlantic forest of Brazil. Biosci- trichomes of pineapple (Ananas comosus, Bromeliaceae). Ann ence 55:207–217 Bot 46:7–11 Inselsbacher E, Cambui CA, Richter A, Stange CF, Mercier H, Schineider JAP, Teixeira RL (2001) Relacionamento entre anfíbios Wanek W (2007) Microbial activities and foliar uptake of anuros e bromélias da restinga de Regência, Linhares, Espírito nitrogen in the epiphytic bromeliad Vriesea gigantea. New Santo, Brasil. Iheringia 91:41–48 Phytol 175:311–320 Solano PJ, Dejean A (2004) Ant-fed plants: comparison between three Krungel P, Richter S (1995) Syncope antenori—a bromeliad breeding geophytic myrmecophytes. Biol J Linn Soc 83:433–439 frog with free-swimming, non-feeding tadpoles (Anura: Micro- Teixeira RL, Schineider JAP, Almeida GI (2002) The occurrence of hylidae). Copeia 1995:955–963 amphibians in bromeliads from a southeastern Brazilian Restinga Laube S, Zotz G (2003) Which abiotic factors limit vegetative growth habitat, with special reference to Aparasphenodon brunoi (Anura, in a vascular epiphyte? Funct Ecol 17:598–604 Hylidae). Braz J Biol 62:263–268 Lehninger AL, Nelson DL, Cox MM (1993) Amino acid oxidation and Treseder KK, Davidson DW, Ehleringer JR (1995) Absorption of ant- the production of urea. In: Lehninger AL, Nelson DL, Cox MM provided carbon and nitrogen by a tropical epiphyte. Nature (eds) Principles of biochemistry, 2nd edn. Worth, New York, 357:137–139 pp 503–541 Versieux LM, Wendt T (2007) Bromeliaceae diversity and conservation Lichtenthaler HK (1987) Chlorophylls and carotenoids: pigments of in Minas Gerais state, Brazil. Biodivers Conserv 16:2989–3009 photosynthetic biomembranes. Methods Enzymol 148:350–382 Wait DA, Aubrey DP, Anderson WB (2005) Seabird guano inXuences Martin GE (1994) Physiological ecology of the Bromeliaceae. Bot Rev on desert islands: soil chemistry and herbaceous species richness 60:1–82 and productivity. J Arid Environ 60:681–695 Ngai JT, Srivastava DS (2006) Predators accelerate nutrient cycling in Yang LH (2004) Periodical cicadas as resource pulses in North Amer- a bromeliad ecosystem. Science 314:963 ican forests. Science 306:1565–1567 Ovaska K (1991) Reproductive phenology, population structure, and Young AM (1979) Arboreal movement and tadpole-carrying behavior habitat use of the frog Eleutherodactylus johnstonei in Barbados, of Dendrobates pumilio Schmidt (Dendrobatidae) in northeastern West Indies. J Herpetol 25:424–430 Costa Rica. Biotropica 11:238–239

123 COPYRIGHT INFORMATION

Title: Nitrogen fluxes from treefrogs to tank epiphytic bromeliads: an isotopic and physiological approach Source: Oecologia 162 no4 Ap 2010 p. 941-9 ISSN: 0029-8549 DOI: 10.1007/s00442-009-1533-4 Publisher: Springer SBM B.V. Gustav Mahlerlaan 10, P.O. Box 283, 1000 EA Amsterdam, The Netherlands

The magazine publisher is the copyright holder of this article and it is reproduced with permission. Further reproduction of this article in violation of the copyright is prohibited. To contact the publisher: http://www.springerlink.com/content/1432-1939/

This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub- licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.