mineral dynamics in spanish moss, tillandsia usneoides l. (bromeliaceae), from central florida, usa

Science of the Total Environment 321(2004) 165–172

0048-9697/04/$ - see front matter� 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2003.09.001

Short communication

Mineral dynamics in Spanish moss,Tillandsia usneoides L.(Bromeliaceae), from Central Florida, USA

George J. Husk , John F. Weishampel *, William H. Schlesingera a, b

Department of Biology, University of Central Florida, Orlando, FL 32816-2368, USAa

Nicholas School of the Environment and Earth Sciences, Duke University, P.O. Box 90328, Durham, NC 27708, USAb

Accepted 3 September 2003

Abstract

Epiphytes absorb water and nutrients from the atmosphere through precipitation and dry deposition and from theirhosts through stemflow and throughfall. These commensals have been used as biological indicators or monitors ofair quality. To measure temporal changes in Spanish moss(Tillandsia usneoides) mineral concentrations, we revisitedsites in Central Florida where this epiphyte was collected and analyzed in 1973y1974. After 24–25 years, usingcomparable methods, concentrations of Ca, Mg, K and Cu decreased in the tissue samples while Fe increased. Thesedeclines in base cations corresponded to global atmospheric decreases. In the earlier study, patterns of elementalconcentrations in Spanish moss corresponded to the host tree categories primarily reflecting a P gradient that increasedfrom pine (Pinus spp.) to cypress(Taxodium spp.) to hardwood(e.g. Quercus spp.) hosts. Such host-specificassociations were mostly absent from the recent study, suggesting that epiphytic preferences based on the chemistryof phorophyte leachates have become less important in this region, perhaps, resulting from local(suburbanization) orregional(atmospheric composition) changes.� 2003 Elsevier B.V. All rights reserved.

Keywords: Aerosol; Central Florida; Epiphyte; Land use change; Phorophyte; Plant nutrition; Spanish moss

1. Introduction

Spanish moss(Tillandsia usneoides L.) andmany other species in this genus because of their‘atmospheric epiphytic’ nature have long beenthought of as potential bio-accumulators, indicatorsand monitors of airborne elements(Wherry andBuchanan, 1926; Wherry and Capen, 1928; Mc-

*Corresponding author. Tel.:q1-407-823-6634; fax:q1-407-823-5769.

E-mail address: [email protected](J.F. Weishampel).

Intyre and Berg, 1956; Martinez et al., 1971;Sheline et al., 1976; Benzing and Bermudes, 1991;Padaki et al., 1992; Gough et al., 1994; Brighignaet al., 1997; Calasans and Malm, 1997; Malm etal., 1998; Bi et al., 2001; Figueiredo et al., 2001;Amando Filho et al., 2002; Pignata et al., 2002).Air quality studies have shown that geographicvariations in mineral concentrations of Spanishmoss often correlate with the proximity to aerosolsources, e.g. the ocean, roads, mines, power plants,soils and urban centers. Mineral concentrations of

166 G.J. Husk et al. / Science of the Total Environment 321 (2004) 165–172

Fig. 1. Locations of sample sites designated by host species in the southwest corner of Orange County, Florida.

Spanish moss in situ are also a function of pho-rophyte or host species leakiness(Schlesinger andMarks, 1977; Callaway et al., 2002). ThoughTillandsia spp. have continually been proposed asa biomonitoring tool throughout the last century,the extent of repeated measures of the mineralconcentrations of these epiphytes has been limitedto -20 months and have shown fluctuations inelemental chemistry to be related to rainfall pat-terns(Gough et al., 1994; Brighigna et al., 1997;Husk, 1999). This study represents a 24–25-yearrevisit of Schlesinger(1976) who collected Span-ish moss from trees in Central Florida in 1973 and1974 and analyzed mineral concentrations in theirtissue relative to host tree species. Given changesover the 24–25-year period in air quality(associ-ated with improvements in per automobile emis-sions, but a large increase in vehicular traffic) andland use(transitioning from an agricultural to asuburban landscape) we expected mineral concen-tration differences in the epiphytic tissue.

2. Methods

The 24 sample sites were located within a;400-km area in Orange County, Florida, south-2

west of Orlando(Fig. 1). Since 1973y1974, landuse in this area has changed from primarily ruralwith citrus agriculture, to suburban, residential andcommercial. To identify the sites, a map fromSchlesinger(1976) was co-registered with currentGIS coverages, i.e. roads and lakes, of OrangeCounty, available from the Florida Department ofEnvironmental Protection. Geographic coordinatesof the original sample sites were obtained fromthese GIS datalayers and were loaded into a GPSunit that had a navigation accuracy of 1 m.Spanish moss collection occurred from February

to March 1998; procedures closely followedSchlesinger(1976) who sampled in 1973 and1974. For each site representing one of threephorophyte categories,Taxodium spp. (six sites),Quercus spp.(six sites), andPinus spp.(12 sites),

167G.J. Husk et al. / Science of the Total Environment 321 (2004) 165–172

i.e. P. clausa Engelm.,P. elliottii Engelm. andP.palustris Mill., approximately 50 g of Spanishmoss was collected from the lower limbs of fivetrees in the dominant phorophyte category usingan expandable 15-m, fiberglass pole. The hostspecies that were sampled at each site in 1973y1974 were resampled in 1998 with one exception;Spanish moss was collected fromMyrica ceriferaL. instead ofQuercus spp. at one site because theoriginal host species was absent from the surround-ing area. Land cover changes over the 24–25-yearperiod prevented host tree locations containingSpanish moss from exactly matching the geograph-ic coordinates from the original 1973y1974 sites.Urban development in the area had eliminated orseverely reduced the extent of eight of the previousforested sites. When a 1973y1974 forested site hadbeen cleared, the nearest location with the samehost trees present was sampled. Geographic coor-dinates of these sites were recorded. The averagedifference between 1973y1974 and 1998 locationswas 272.0 m(minimum differences56.6 m, max-imum differences818.4 m). An examination ofhistoric aerial photographs illustrated the land cov-er changes and suggested that some of the posi-tioning errors were due to inaccuracies ingeoreferencing the original map; thus, the averagedistance difference is actually smaller.Analyses of the 120 samples were performed at

DB Environmental Laboratories in Rockledge,Florida. This laboratory is state-certified for all ofthe analyses conducted. Though this study attempt-ed to replicate the 1973y1974 study, tissue samplesfrom the earlier study were not available to cross-calibrate the laboratory analyses. Thus, differencesin laboratory equipment and procedures are apossible source of comparison error. Furthermore,only element mean and standard deviation valuesfor each site remained from the 1973y1974 study.This precluded an analysis of locality relatedchanges in mineral concentrations. Tissue sampleswere dried in forced air ovens, then ground in aWiley mill, and passed through a 100-mesh screento further homogenize. We digested subsamples ofthe plant tissue in nitric acid, hydrochloric acidand hydrogen peroxide(Method�SW 3050; USEPA, 1986) to extract seven of the minerals(Ca,

Cu, Fe, K, Mg, Mn and Zn). We used flameatomic absorption spectrometry(FLAAS) to quan-tify six of these minerals(Ca, Fe, K, Mg, Mn andZn), and a transversely heated graphite furnaceatomic absorption spectrometer with Zeeman back-ground correction was used to measure Cu due toboth the low levels present and the poor detectionlimit for Cu using the FLAAS. Dried Spanishmoss tissue was also digested for total P usingmethod COE 3-227(Plumb, 1981). The digestswere analyzed for total P following EPA 365.1(US EPA, 1979). Lastly, total N was analyzedfrom the dried tissue using an elemental analyzer(NA1500 NCS by Carlo Erba Instruments). Toassure data quality and integrity, two known plantsamples(pine needles and tomato leaves) fromthe US National Institute for Standards and Testingwere digested along with each batch. This ensuredthat a complete digestion of the plant samplesoccurred. Element concentrations of standards fellwithin the expected ranges.Mineral concentrations of the five samples at

each site were converted to a weight percent basedon dry wt. and then averaged for each site. Aseries of unpairedt-tests were performed to deter-mine the significance of differences in mineralmeans from the sites in 1973y1974 and 1998.Ordination using principal components analysiswas performed on the 1973y1974 and 1998 aver-ages to examine relationships among phorophyteand mineral concentrations and to identify trendsin the overall patterns of mineral dynamics. Tobetter understand patterns among different phoro-phyte categories, ANOVA tests with a Bonferronicorrection were performed on the three host groupsbased on the elemental concentrations found in thecommensal plant.

3. Results

The mean weight percent based on dry wt. forthe nine minerals found in Spanish moss for thecentral Florida sites in 1973y1974 and 1998 arelisted in Table 1. Averages over the 24 sites yieldedsignificant changes from 1973y1974 to 1998 forfive of the nine elements analyzed. K, Mg, Mnand Cu exhibited significant(P-0.01) decreases


Table 1Nutrient content based on dry wt. from the 24 sample sites

Mineral 1973y1974 1998

Mean S.D. Mean S.D.

N (%) 1.060 0.232 1.131 0.620K (%)* 0.504 0.097 0.255 0.086Ca (%) 0.465 0.120 0.416 0.240Mg (%)* 0.207 0.031 0.136 0.024P (ppm) 399.6 110.3 414.4 171.4Fe (ppm)* 292.3 137.3 574.6 309.2Mn (ppm)* 136.8 56.8 52.8 30.8Zn (ppm) 65.4 16.2 73.5 28.3Cu (ppm)* 20.7 8.3 7.1 2.3

Indicates significantly different means(P-0.01).*

Table 2Eigen vectors associated with the principal component analysisof mineral composition of Spanish moss tissue from the 24sample sites

Mineral 1973y1974 1998

PC1 PC2 PC1 PC2

N 0.372 0.262 0.478 y0.288K y0.154 0.544 0.534 y0.277Ca y0.088 0.000 0.505 0.368Mg 0.397 y0.234 0.279 y0.424P y0.080 0.658 0.156 0.031Fe 0.453 0.035 0.267 0.555Mn y0.203 0.250 0.213 0.049Zn 0.460 0.090 y0.087 y0.315Cu 0.456 0.274 0.081 0.338

% Variance 31.7 18.4 23.9 18.9

The two sampling periods were analyzed separately.

Fig. 2. Ordinations of Orange County, Florida sites and host affiliates based on mineral concentrations in(a) 1973y1974(Schlesinger,1976) and(b) 1998(Husk, 1999) Spanish moss tissue samples. Groupings of host categories are shown by dotted envelopes. The1973y1974 groupings follow Schlesinger and Marks(1977).

from those measured in 1973y1974. Conversely,Fe showed a significant(P-0.01) increase. N,Ca, P and Zn showed no significant temporaldifference.The segregation in mineral concentrations

among phorophyte species found in 1973y1974was absent in 1998(Fig. 2). The first two principalcomponents combined explained 50.1 and 42.8%of the variance for the 1973y1974 and 1998 data,respectively. The ordination in Schlesinger andMarks (1977) showed the mineral composition ofSpanish moss samples to cluster for different hostspecies primarily along a P gradient. Thoughcomparable in amount to the 1973y1974 samples,

P in 1998 had relatively low loadings on the firsttwo principal components. In general, the principalcomponents loadings were very different betweenthe two sample periods(Table 2). For the 1973y1974 period, five elements, K, Mg, P, Fe and Mn,had significantly different concentrations amongthe three host categories(Table 3). In 1998, onlyFe yielded significantly different levels.When mineral concentrations from the two sam-

pling periods were analyzed together, the first twocomponents explained 51.6% of the variance.


Table 3Average element concentrations based on the percent dry wt. among different host trees. Standard deviations are in parentheses

Element Sampling period Host trees

Taxodium spp. Mixed Hardwood Pinus spp.

N (%) 1973y1974 1.171(0.103) 1.060(0.173) 1.004(0.289)1998 0.978(0.126) 1.057(0.077) 1.271(0.884)

K (%)* 1973y1974 0.468 (0.096)a,b 0.591 (0.101)a 0.478 (0.073)b

1998 0.210(0.056) 0.312(0.105) 0.261(0.073)

Ca (%) 1973y1974 0.405(0.060) 0.498(0.172) 0.478(0.110)1998 0.483(0.384) 0.346(0.087) 0.438(0.209)

Mg (%)* 1973y1974 0.225 (0.038)a 0.182 (0.026)b 0.211 (0.021)a,b

1998 0.131(0.018) 0.126(0.021) 0.146(0.022)

P (ppm)* 1973y1974 423.3 (43.7)a,b 551.7 (38.2)b 311.7 (50.4)a

1998 456.8(245.3) 508.8(189.5) 343.9(80.4)

Fe (ppm)* 1973y1974 439.0 (169.8)a 212.7 (36.5)b 258.8 (96.4)b

1998 833.4 (479.6)a 399.3 (119.2)b 532.9 (183.5)a,b

Mn (ppm)* 1973y1974 95.5 (33.9)a 187.7 (84.4)b 132.1 (24.8)a,b

1998 34.0(12.9) 65.0 (54.6) 56.0 (16.5)

Zn (ppm) 1973y1974 69.8(21.0) 59.8 (11.7) 65.9 (16.0)1998 58.7(16.6) 62.3 (24.9) 125.8(135.9)

Cu (ppm) 1973y1974 23.6(7.3) 19.7 (6.0) 19.7 (9.9)1998 6.8(3.6) 6.7 (2.1) 7.3 (1.7)

Different superscripts indicate significantly(P-0.05) different concentrations within a sampling period.*

Because the data from the two sampling periodswere combined, it was not expected that thesepatterns would match those when ordinated sepa-rately. It is apparent that there was a shift alongthe first principal component from 1973y1974 to1998 (Fig. 3). The segregation between samplingperiods and loadings(Table 4) reflect the signifi-cant differences found in Table 1.

4. Discussion

Given that there have been considerable changesin the population size and land use patterns ofCentral Florida since 1973(Kautz, 1998) andpresumed associated changes in air quality, it isnot surprising that differences in Spanish mossmineral concentrations were found. Furthermore,because short-term fluxes inTillandsia spp. min-eral concentrations may result from recent rainfallevents(Brighigna et al., 1997; Husk, 1999), vari-ations were expected.The values for Ca, K, Mg and P fall within the

range reported by Gough et al.(1994) in South

Carolina. The general decreases in base cation(Ca, Mg and K) concentrations from 1973y1974to 1998 followed global atmospheric declines doc-umented by Hedin et al.(1994). One of theNational Atmospheric Deposition ProgramyNational Trends Network monitoring stations usedfor that study is in Florida north of Orange County.Though nitrate concentrations in Florida rainfallhave increased over this time period(Brezonik etal., 1980; Madsen et al., 1992; Grimshaw andDolske, 2002), there was no difference in N in theepiphyte tissue. Florida precipitation is moderatelyacidic which could lead to leaching of elementsfrom this host–epiphyte–soil system. Though atthe time of the Schlesinger(1976) study, CentralFlorida was experiencing a marked increase inH deposition(Brezonik et al., 1980), the generalq

trend over the last couple of decades in region hasbeen a slight increase in pH(Madsen et al., 1992;Madsen and Dreschel, 1996).In other geographic regions at local scales,

elevated levels of Fe in plant tissues have beenassociated with proximity to roads(Loranger et


Table 4Eigen vectors associated with the principal components anal-ysis of mineral composition of Spanish moss tissue from the24 sample sites

Mineral PC1 PC2

N 0.000 y0.470K y0.491 y0.168Ca y0.039 y0.683Mg y0.465 y0.046P y0.007 y0.261Fe 0.336 y0.452Mn y0.464 0.027Zn 0.048 0.071Cu y0.459 y0.062% Variance 34.6 17.0

The two sampling periods were analyzed together.

Fig. 3. Trajectories of Spanish moss samples in ordination space based on mineral concentrations. White and black markers representsamples from 1973y1974 to 1998, respectively.

al., 1996; Janssen et al., 1997); thus, the increasein road density and vehicular traffic in this area ofCentral Florida could explain the increased con-centration in Spanish moss. However, if this werethe source other elements(e.g. K, Mn, Zn andCu), associated with roads should have increasedas well as found in Loranger et al.(1996) andJanssen et al.(1997). Florida soils represent anoth-er possible source of Fe, but they are especiallyprone to leaching and their metal content is lowerthan most soils in the US. The soils of this areaof Orange County are largely spodosols that evenhave lower metal contents than other Florida soils(Ma et al., 1997). Across larger scales, atmos-pheric Fe inputs into Florida environs correspondto African dust transportation events, but anincrease in the magnitude of these events is notconsistent with the observed patterns from 1974to 1996(Prospero, 1999).A change of biological note is the apparent

weakening or breakdown of the relationshipsbetween phorophyte identity and the mineral con-centrations of the Spanish moss grown on that hostby Schlesinger and Marks(1977). Two axes ofeach the ordinations accounted for approximately

half of the variance. The remaining variance prob-ably relates to local differences in land use, soilproperties, tree host age and health and, perhaps,recent rainfall events. The 1973y1974 segregationof hardwood-, cypress- and pine-dominated standsdue to mineral concentrations in epiphyte tissuewas largely related to differences in P. Though the1998 average levels of P concentrations for a hostcategory were consistent with the 1973y1974 data,


i.e. hardwood)cypress)pine, the differencesunlike those from 1973y1974 were not significant.Hence, P differences were not critical in ordinatingphorophyte groups by epiphyte mineral concentra-tion. Moreover, concentration differences in K, Mgand Mn also became insignificant. Only significantdifferences of Fe associated with the host categoryremained. These changes suggest that either theleachate from the host species has been modifiedor that its contribution to the overall mineralcomposition of Spanish moss has diminished.Stemflow collections from the different hostswould permit analyses of the leachate composi-tions. Regardless, species-specific preferences(Callaway et al., 2002) for hosts that leach largeramounts of nutrients, primarily P, enabling morerapid epiphyte growth(Schlesinger and Marks,1977) may not be as important as they had beenin the past. These new patterns of elementalcomposition in Spanish moss tissue representchanges in nutrient cycling(Vitousek, 1994) per-haps resulting from local land use changes causedby suburbanization andyor broader regional chang-es in atmospheric chemistry.

Acknowledgments

We wish to thank Tom DeBusk and ForrestDierberg from DB Environmental Laboratories foradvice and permission to use their facilities; JasonGodin for assistance with sampling and locatingsites; and James Bennett, Jack Stout, Peter Weis-hampel and Henry Whittier for helpful commentson the manuscript.

References

Amando Filho GM, Andrade LR, Farina M, Malm O. Hglocalisation inTillandsia usneoides L. (Bromeliaceae) anatmospheric biomonitor. Atmos Environ 2002;36:881–887.

Benzing DH, Bermudes D. Epiphytic bromeliads as air qualitymonitors in South Florida. Selbyana 1991;12:46–53.

Bi M, Austin E, Smith BW, Winefordner JD. Elementalanalysis of Spanish moss using laser ablation inductivelycoupled plasma mass spectrometry. Anal Chim Acta2001;435:309–318.

Brezonik PL, Edgerton ES, Hendry CD. Acid precipitationand sulfate deposition in Florida. Science 1980;208:1027–1029.

Brighigna L, Ravanelli M, Minelli A, Ercoli L. The use of anepiphyte(Tillandsia caput-medusae morren) as bioindicatorof air pollution in Costa Rica. Sci Total Environ1997;198:175–180.

Calasans CF, Malm O. Elemental mercury contaminationsurvey in a chlor-alkali plant by the use of transplantedSpanish moss,Tillandsia usneoides (L.). Sci Total Environ1997;208:165–177.

Callaway RM, Reinhart KO, Moore GW, Moore DJ, PenningsSC. Epiphyte host preferences and host traits: mechanismsfor species specific interactions. Oecologia 2002;132:221–230.

Figueiredo AMG, Saiki M, Ticianelli RB, Domingos M, AlvesES, Markert B. Determination of trace elements inTillandsiausneoides by neutro activation analysis for environmentalbiomonitoring. J Radioanal Nucl Chem 2001;249:391–395.

Gough LP, Severson RC, Jackson LL. Baseline element con-centrations in soils and plants, Bull Island, Cape RomainNational Wildlife Refuge, South Carolina, USA. Water AirSoil Pollut 1994;74:1–17.

Grimshaw HJ, Dolske DA. Rainfall concentrations and wetatmospheric deposition of phosphorus and other constituentsin Florida, USA. Water Air Soil Pollut 2002;137:117–140.

Hedin LO, Granat L, Likens GE, Buishand TA, Galloway JN,Butler TJ, Rodhe H. Steep declines in atmospheric basecations in regions of Europe and North America. Nature1994;367:351–354.

Husk GJ. Spatio-temporal dynamics of Spanish moss(Tilland-sia usneoides L.; Bromeliaceae) biogeochemistry. M.S. The-sis. Orlando, FL: University of Central Florida; 1999. p. 95.

Janssen NAH, van Mansom DFM, van der Jagt K, HarssemaH, Hoek G. Mass concentration and elemental compositionof airborne particulate matter at street and backgroundlocations. Atmos Environ 1997;31:1185–1193.

Kautz RS. Land use and land cover trends in Florida 1936–1995. Florida Scientist 1998;61:171–187.

Loranger S, Tetrault M, Kennedy G, Zayed J. Manganese and´trace elements in urban snow near an expressway. EnvironPollut 1996;92:203–211.

Ma LQ, Tan F, Harris WG. Concentrations and distributionsof eleven metals in Florida soils. J Environ Qual1997;26:769–775.

Madsen BC, Dreschel TW. Acid rain monitoring in Floridafrom 1978 to the present and evaluation of trends inrainwater composition. Abstracts of Papers of the AmericanChemical Society 1996;212:50-ENVR Part 1.

Madsen BC, Kheoh T, Hinkle CR, Dreschel T. Precipitationchemistry in East Central Florida from 1978 to 1987. WaterAir Soil Pollut 1992;65:7–21.

Malm O, de Freitas Fonseca M, Hissnauer Miguel P,Rodrigues Bastos W, Neves Pinto F. Use of epiphyte plantsas biomonitors to map atmospheric mercury in a gold tradecenter city, Amazon, Brazil. Sci Total Environ 1998;213:57–64.

Martinez JD, Nathany M, Dharmarajan V. Spanish moss, asensor for lead. Nature 1971;233:564–565.


McIntyre RT, Berg EW. Mineral content of Spanish moss.Ecology 1956;37:605–606.

Padaki PM, McWilliams EL, James WD. Use of Spanish mossas an atmospheric monitor for trace elements. J RadioanalNucl Chem 1992;161:147–157.

Pignata ML, Gudino GL, Wannaz ED, Pla RR, Gonzalez CM,˜ ´ ´Carreras HA, Orellana L. Atmospheric quality and distri-bution of heavy metals in Argentina employingTillandsiacapillaris as a biomonitor. Environ Pollut 2002;120:59–68.

Plumb RH. Procedure for handling and chemical analysis ofsediment and water samples. Technical Report EPAyCE-81-1. US EPA and Corps of Engineers, US Army EngineersWaterways Experimental Station, Vicksburg, MS; 1981.

Prospero JM. Long-term measurements of the transport ofAfrican mineral dust to the southeastern United States:implications for regional air quality. J Geophys Res Atmos1999;104:15917–15927.

Schlesinger WH. Biochemical Limits on Two Levels of PlantCommunity Organization in the Cypress Forest of Okefe-nokee Swamp. Ph.D. Dissertation, Ithaca, NY: Cornell Univ.;1976.

Schlesinger WH, Marks PL. Mineral cycling and the niche ofSpanish moss,Tillandsia usneoides L. Am J Bot1977;64:1254–1262.

Sheline J, Akselsson R, Winchester JW. Trace element simi-larity groups in North Florida Spanish moss: evidence fordirect uptake of aerosol particles. J Geophys Res1976;81:1047–1050.

US EPA. Methods for chemical analysis of water and wastes.Technical Report EA-600 4-79-020. Cincinnati, Ohio: Envi-ronmental Monitoring and Support Laboratory, US EPA;1979.

US EPA. Test methods for evaluating solid waste, physicalychemical methods, Publication SW-846, 3rd ed.; 1986.

Vitousek PM. Beyond global warming: ecology and globalchange. Ecology 1994;75:1861–1876.

Wherry ET, Buchanan R. Composition of the ash of Spanish-moss. Ecology 1926;7:303–306.

Wherry ET, Capen RG. Mineral constituents of Spanish-mossand ballmoss. Ecology 1928;9:501–504.

mineral dynamics in spanish moss, tillandsia usneoides l. (bromeliaceae), from central florida, usa

Documents