Ataphonomicstudyofmodernpollenassemblagesfromdung ... Carrion_RPP.pdftum,Rhamnusoleoidessubsp.angustifolia,Pista-cialentiscus,Oleaeuropaeavar.sylvestris,Stipa tenacissima, Ephedra

A taphonomic study of modern pollen assemblages from dungand surface sediments in arid environments of Spain

Jose¤ S. Carrio¤n �

Departamento de Biolog|¤a Vegetal, Facultad de Biolog|¤a, Universidad de Murcia, 30100 Murcia, Spain

Received 12 June 2001; accepted 11 December 2001

Abstract

Coastal plant communities in arid southeastern Spain are characterized by insect-pollinated scrub species, whichfail to occur in Quaternary pollen sequences from valleys, marshlands and marine cores. We investigate pollen^vegetation relationships in samples from soil surfaces, animal dung, and sediments in depressions or basins that, intheory, should have pollen spectra that are comparable to those from sedimentary basins elsewhere. Pollen spectrafrom basins or depressions are very susceptible to long-distance wind and water transport. This can maskrepresentation of pollen from the surrounding insect-pollinated vegetation, as can over-representation of basin-marginhalophilous and hydrophilous pollen. Pollen spectra from biogenic materials of animal origin are the best analoguesof local and regional vegetation, and show the best analytical potential in terms of pollen concentration and taxondiversity. Pollination properties of the species studied indicate they will rarely be found in most conventional pollenrecords. It cannot be stressed too strongly that insight into Quaternary vegetation of arid regions demandscomplementary pollen analysis of coprolites, urine-cemented deposits, and cave sediments with preserved bioticremains, in addition to water-lain sediments. . 2002 Elsevier Science B.V. All rights reserved.

Keywords: palynology; paleoecology; pollen^vegetation relationships; Spain; Quaternary

1. Introduction

While pollen analysis is a well-grounded toolfor paleoecological studies in temperate regions,it poses several problems in arid environments,foremost among which are poor preservation,scarcity of suitable depositional basins, and rela-tively high rates of sedimentary accumulation(Horowitz, 1992). Pollen analyses in arid areasof southeastern Spain have tended to concentrate

on those sequences that were available rather thanon the more desirable ones. Those include adja-cent marine sediments (Parra, 1994; Targarona,1997); small peaty deposits in sub-coastal moun-tains (Riera et al., 1995); playa lakes (Burjachs etal., 1997); valley-bottom ¢lls in badland areas(Nogueras et al., 2000); paleo-lagoons and coastalmarshes (Yll et al., 1994; Pantaleo¤n-Cano, 1997);and prehistoric sites (Mariscal, 1992; Davis andMariscal, 1994; Carrio¤n et al., 1995, 1999a).The resulting paleoenvironmental picture is

fragmentary (Carrio¤n et al., 2000a). Worse, thereis a striking lack of the pollen of the most abun-dant and characteristic species of the coastal veg-

0034-6667 / 02 / $ ^ see front matter . 2002 Elsevier Science B.V. All rights reserved.PII: S 0 0 3 4 - 6 6 6 7 ( 0 2 ) 0 0 0 7 3 - 8

* Fax: +34-968-364995.E-mail address: [email protected] (J.S. Carrio¤n).

PALBO 2453 12-8-02

Review of Palaeobotany and Palynology 120 (2002) 217^232

www.elsevier.com/locate/revpalbo

mailto:[email protected]

http://www.elsevier.com/locate/revpalbo

etation today, if we except minor records from thearcheological sites of Perneras (Carrio¤n et al.,1995) and Almizaraque (Davis and Mariscal,1994). However it is unlikely that such speciesas Maytenus senegalensis subsp. europaea (Arto,Celastraceae), Periploca angustifolia (Cornical,Periplocaceae), Withania frutescens (Solanaceae),Ziziphus lotus (Azufaifo, Rhamnaceae), and Cali-cotome intermedia (Fabaceae), among others, arerecent arrivals to the region, or disappeared fromit during the Quaternary. These species todayform structurally and £oristically well-de¢nedcommunities and they point to an Ibero^NorthAfrican disjunction, linked to late-Miocene re-opening of the Strait of Gibraltar (Mota et al.,1997).This work was designed to test the hypothesis

that the reason why these species fail to occur inthe Quaternary pollen record of the region islargely taphonomical. Like other aridity-adaptedplants, most of them are zoophilous (in particular,insect-pollinated), and their pollen should be rarein the deposits most often used for pollen analyses

such as water-lain sediments. Studies of modernpollen samples of di¡erent kinds may shed lightonto this issue. It bears emphasis that, if theworking hypothesis were to be con¢rmed, weshould revise our notions about Quaternary veg-etation of the region.

2. Suitability of the study area for addressing theissue

The southeastern Iberian Peninsula, or Mur-cianÂlmerian phytogeographical province, ex-tends over ca. 13 000 km2, and represents one ofthe richest areas of European £ora, with impor-tant bioclimatic gradients and geological diversity(Sa¤nchez-Go¤mez et al., 1998) (Fig. 1). Climate isMediterranean and slightly continental, with an-nual values of 16^17‡C in the lower bioclimaticbelts. Extreme summer temperatures are common,with daily maxima of up to 46‡C. The territory islimited by the Mediterranean Sea and surroundedby mountain ranges acting as a natural shield

Fig. 1. Study area, location of sampling and reference pollen sites, and dominant vegetation types.

PALBO 2453 12-8-02

J.S. Carrio¤n / Review of Palaeobotany and Palynology 120 (2002) 217^232218

against precipitation. Yearly rainfall ranges from250 to 330 mm, though in mountainous areas itmay exceed 400 mm and in some coastal areas itfalls below 190 mm. Precipitation is irregular andoften causes devastating £oods of great erosivepower. Evapotranspiration is greater than any-where else in the Iberian Peninsula.More than half of the ca. 7500 Iberian vascular

plants occur in the small territory of this province.There is a majority of Mediterranean contingentspecies (65%), including Iberian, MurcianÂlmer-ian, Ibero^North African and IberoÎrano^Tura-nian endemics (Peinado et al., 1992). Coastalmountains and littoral depressions of this prov-ince are patchily occupied by sclerophyllousbrushwoods dominated by one or more of thefollowing species: Maytenus senegalensis subsp.europaea, Periploca angustifolia, Withania frutes-cens, Ziziphus lotus, Calicotome intermedia, Lau-naea arborescens, and several species of Genista.Other abundant species include Lycium intrica-tum, Rhamnus oleoides subsp. angustifolia, Pista-cia lentiscus, Olea europaea var. sylvestris, Stipatenacissima, Ephedra fragilis, Asparagus albus,Lavandula dentata, Rosmarinus o⁄cinalis, Euzo-modendron bourgaeanum, as well as a diversityof Cistaceae, Lamiaceae, and Asteraceae spe-cies.Brushwoods are sometimes accompanied by

sparse stands of Pinus halepensis and Tetraclinisarticulata. Open forests with P. halepensis andQuercus rotundifolia, and Quercus coccifera scruboccur in the more continental depressions and lessdry conditions of the mountainous hinterland(Fig. 1). Saline depressions or basins and intermit-tent water-courses, very common throughout theprovince, are colonized by Artemisia, Nerium ole-ander, Chenopodiaceae, Tamarix, Limonium, andPoaceae species.The MurcianÂlmerian bioprovince represents

an excellent model system to address a variety ofenvironmental issues because of its plant-speciesdiversity, its varied physiography with abundanceof ecotonal territories, its great risk of deserti¢ca-tion, and the antiquity of human pressure on thelandscape.A big problem with palynology in other arid

regions has been the di⁄culty of attaining speci¢c

and generic identi¢cations (Horowitz, 1992). Spe-ci¢c identi¢cation is possible for Maytenus, Peri-ploca, Withania, Ziziphus, Lycium, Calicotome,Osyris, Nerium, Arisarum, Olea, Pistacia, Ephe-dra, and Chamaerops (Table 1), which are mono-speci¢c in southern Spain, thereby enabling sepa-ration of regional and local sources. This makesour area suitable for studies of poorly knownpollen^vegetation relationships in arid environ-ments.

3. Materials and methods

3.1. Localities selected

Fifteen localities were selected that embody themain vegetation types with predominance of en-tomophilous species (Fig. 1). Localities 1, 3, 7, 10,and 14 fall within the area characterized by Peri-ploca angustifolia communities, often includingWithania frutescens and Maytenus europaeus,with Lycium intricatum and Asparagus albus incoastal blu¡s subjected to salt spray. Locality 2was selected because of the abundance of Genistacinerea subsp. murcica and Calicotome intermedia.Locality 5 is dominated by Launaea arborescens,Lavandula dentata, Chamaerops humilis, and sev-eral species of Helianthemum and Fumana. Local-ities 4, 8, and 12 are characterized by Ziziphuslotus and A. albus. Localities 6 and 9 were selectedbecause of the respective abundance of Launaeaarborescens and Genista spartioides subsp. reta-moides. Locality 11 is situated in the Sorbas gyp-sum karstic area dominated by the endemic com-munity of Genista ramosissima, Ononis tridentata,and Retama sphaerocarpa. Locality 13 is situatedin the Tabernas desert community of Euzomoden-dron bourgaeanum and Genista umbellata. Locality15 falls within the area dominated by commun-ities of Maytenus senegalensis and W. frutescens,with abundant Z. lotus and Phlomis purpureasubsp. almeriensis.

3.2. Sampling

Three categories of samples were studied paly-nologically. Each sample consisted of a collection

PALBO 2453 12-8-02

J.S. Carrio¤n / Review of Palaeobotany and Palynology 120 (2002) 217^232 219

of 10^25 subsamples which were placed in poly-thene bags and mixed before proceeding withmaceration. Subsamples were located as far aspossible from each other, and away from concen-trations of plants within about 2000 m2. In orderto minimize seasonal over-representation, sam-pling was made in early winter-time, when most£owers were not yet in blossom. Formaldehydewas added to the samples before the bags weresealed in order to deter bacterial activity.The ¢rst sample type is the ‘surface’ type. This

is minerogenic surface sediment that, supposedly,records airborne pollen. Subsamples were mainlya loose, yellow-bu¡ to brownish, ¢ne-grain dustfrom a depth of about 0^1 cm. We discardedsands and coarser clasts, cemented rocks, springdeposits, evaporites, and areas with nearby signsof bioturbation (e.g. earthworm or rat mixing,burrowing, etc.) and/or wind de£ation. We alsoavoided blackish sediments, because although in-deed usually containing pollen grains, they mayalso contain considerable quantities of organic

Table 1Taxonomic de¢nition of the main pollen types within the study area

Pollen type Taxa

Maytenus Maytenus senegalensis subsp. europaeaPeriploca Periploca angustifoliaWithania Withania frutescensZiziphus Ziziphus lotusCichorioideae Launaea arborescens, Sonchus tenerrimus (12), L. lanifera (10, 13)Lamiaceae Rosmarinus o⁄cinalis, Thymus hyemalis, Satureja obovata, Sideritis pusilla, S. osteoxyla (14), Lavandula

multi¢da, L. dentata, Ballota hirsuta, Teucrium freynii (5), T. lanigerum (7), T. turredanum (11),T. eriocephalum (15), Phlomis purpurea subsp. almeriensis (9, 10)

Cistaceae Helianthemum almeriense, H. syriacum, H. alypoides (11), H. squamatum (11), H. scopulorum (12),H. leptophyllum (15), Fumana ericoides, F. thymifolia, F. laevipes, F. hispidula (9), Cistus monspeliensis (5)

Asparagus Asparagus albus, A. horridusLycium Lycium intricatumRhamnus Rhamnus oleoides subsp. angustifolia, R. alaternus, R. lycioides (4, 9)Genista^Retama Genista cinerea subsp. murcica (2), G. umbellata (5, 15), G. spartioides subsp. retamoides (9),

G. ramosissima (11), Retama sphaerocarpa (11), Ulex parvi£orus (15)Calicotome Calicotome intermediaOsyris Osyris quadripartitaAsteroideae Carthamus arborescens, Asteriscus maritimus (1, 2, 15), Phagnalon saxatile (10), Santolina viscosa (8, 11),Nerium Nerium oleanderTamarix Tamarix canariensis, T. boveana (13)Euzomodendron Mainly Euzomodendron bourgeanumLimonium Limonium cossonianum, L. insigne, L. caesiumArisarum Arisarum simorrhinumArtemisia A. barrelieri, A. herba-alba, A. campestrisPoaceae+Lygeum Stipa tenacissima, Brachypodium retusum, Lygeum spartum, Dactylis glomerata, Saccharum ravennae (4, 8),

Arundo donax (4), Phragmites australisOlea Olea europaeaPistacia Pistacia lentiscusChamaerops Chamaerops humilisPinus Pinus halepensis, P. pineaChenopodiaceae Species of Salsola, Suaeda, Atriplex, Hammada, Anabasis, Salicornia, Arthrocnemum, SarcorniaEphedra Ephedra fragilisCupressaceae Tetraclinis articulata, Cupressus sempervirens, Juniperus turbinata (15)Quercus Quercus coccifera, Q. rotundifoliaCyperaceae Mainly Scirpus holoschoenus

Numbers in brackets refer to localities in which a particular species is the most likely pollen producer (Fig. 1).

PALBO 2453 12-8-02


substances, which may be di⁄cult to get rid o¡during the extraction processes.The second sample type is the ‘dung’ type. This

is a mixture of pellets/feces from di¡erent animals,mainly goat, sheep, rabbit, hare, and less com-monly wild boar. Both fresh and thoroughlydry excrements were collected in order to ensurethey had acted long enough as pollen traps torecord the whole set of £owering seasons.‘Dung’ samples largely represent the pollen con-sumed and transported by herbivores, especiallyby ingestion of £owers, leaves, and other plantmaterials.The third sample type is the ‘depression’ type.

This is sediment from depressions or basins asso-ciated with marshes, wadis and arroyos. We oftenused the characteristic soft mud cakes depositedat the end of the last £ood. This sediment typeseems to correspond to pollen deposition in la-goons, marshes, salt-pans, and playa-lake systems(Horowitz, 1992), which have been a good sourceof long continuous sequences in the area (e.g.Pantaleo¤n-Cano, 1997). In addition, all samplesincluded one or several subsamples obtained bydredging the top layer of underwater sedimentsin bodies of perennial water. This was done inorder to overcome possible biasses attributableto di¡erential pollen preservation in samples ofdry mud cakes which are exposed to temporarydrying out. We avoided coarse clasts, and cal-cretes.

3.3. Laboratory methods

Maceration was undertaken within 2 monthsafter sample collection. A weight of 12^42 g ofdry sediment per surface sample was processedin the laboratory. This amount is thus greaterthan in samples from forested territories of theregion, where 1 or 2 g of sediment can be enough(Navarro et al., 2000). For dung samples, we pro-cessed 2^8 g of dry material. Smaller samples maycontain enough pollen to give an acceptable spec-trum, but they may be inadequate for drawing¢rm conclusions. This increased sample size re-sults from mixing dung fragments from di¡erentanimals and is intended to overcome most of thevariability that might re£ect a single pellet or a

single animal source. Finally, in the case of de-pression samples, we macerated a weight of 28^60g of dry sediment.The laboratory preparation techniques follow

conventional methods in palynology using HCl,HF, KOH and mineral separation with a heavyliquid (ZnCl2). At the beginning of the pollen pro-cessing, ¢ve Lycopodium clavatum spore tablets(ca. 12 542 spores per tablet) were added to eachsample in order to facilitate concentration calcu-lations (grains per gram dry weight). All slideswere mounted in glycerin jelly and stained withsafranin.

3.4. Pollen identi¢cation and counting

Identi¢cation and counting were performed us-ing the pollen reference collection of Murcia Uni-versity’s Palynology Laboratory. Non-vascularcryptogam spores (mainly Riccia, Glomus chlamy-dospores, and Zygnemataceae zygospores andaplanospores), and other non-pollen microfossils(mainly Pseudoschizaea) were not considered inthis study. No reworked palynomorphs fromunderlying Paleozoic and Mesozoic strata wereobserved.Several pollen identi¢cations merit explanation.

Lycium pollen was separated from other Solana-ceae types based on its characteristically smallapocolpium and striate apertural membrane.Withania pollen shows a peculiar foveolate pat-tern, striate markings being seldom visible acrossthe mesocolpium. Ziziphus was discriminatedfrom Rhamnus on the basis of its less clearlymarked reticulum and less pronounced di¡erencesin lumina-width towards the center of the meso-colpium. Calicotome pollen is clearly di¡erentfrom Genista type, displaying thicker reticulummuri, a rectangular shape in equatorial view,and more £attened mesocolpia. Asparagus pollenis relatively small when contrasted against thecommonest size-pattern for the Liliaceae, and ischaracteristically elliptical in polar view. Cha-maerops, being the only natural palm species inour area (where it is abundant nowadays), wasdistinguished from Phoenix by its larger and lessisodiametric pollen. Considering local vegetation,Launaea arborescens and Carthamus arborescens

PALBO 2453 12-8-02


Fig.2.Percentagepollendiagramandaveragecovervaluesofanemophiloustaxa.

PALBO 2453 12-8-02


must be the main pollen contributors to the Ci-chorioideae and Asteroideae groups (Table 1).The Lamiaceae type includes undi¡erentiatedhexa- and tricolpate grains as well as pollen ofTeucrium and Sideritis. Brassicaceae pollen isonly abundant in locality 13 so it must belongmainly to Euzomodendron bourgaeanum. Pollentypes relative to ‘other anemophilous’ (Fig. 2),and ‘other zoophilous’ forms (Fig. 3) are includedin Table 2. Nomenclature of plant taxa followsSa¤nchez-Go¤mez et al. (1998) and Mota et al.(1997).

3.5. Pollen diagrams

Three pollen spectra per locality are displayedin pollen percentage diagrams of anemophilous(Fig. 2) and zoophilous taxa (Fig. 3). Pollen per-centages are based on pollen sums of at least 200pollen grains, excluding aquatic pollen andspores. Mean pollen percentages of surface,dung, and depression samples are expressed asgraphs in Fig. 4. Pollen taxa with relative frequen-cies steadily below 2% have been excluded frompollen diagrams (Table 2).

Table 2Pollen and spore types with percentages steadily below 2%, which form the basis for the sums of ‘other zoophilous’ and ‘otheranemophilous’ taxa in the pollen diagrams (Figs. 2^4)

Zoophilous Anemophilous and ferns

Agave Lotus Alnus (DP)Anchusa Malvaceae Asplenium (SU)Anthyllis Messembryanthemum (DP) Betula (DP)Aristolochia (DU) Myosotis (DU) Castanea (DP)Asphodelus Myrtus CasuarinaBoraginaceae Neatostema (DU) Cedrus (DP)Bupleurum Nicotiana Corylus (DP)Campanula Ononis Cosentinia (SU)Capparis Opuntia Equisetum (DP)Caryophyllaceae Oxalis EucalyptusCentaurea Papaver Fraxinus (DP)Centaurium (DP) Paronychia Juncus (DP)Ceratonia (DU) Polygala (DU) MercurialisChaenorrhinum (DU) Polygonum (DP) PhoenixConvolvulus Primulaceae PlantagoCoris Prunus Populus (DP)Coronilla Psoralea RumexCynoglossum Ranunculaceae SanguisorbaDorycnium (DU) Reseda (DU) SchinusEcballium Rubiaceae SelaginellaEchium (DU) Rubus ThalictrumEricaceae Ruta Typha (DP)Erodium Salix Ulmus (DP)Euphorbia Sarcocapnos UrticaceaeFagonia (DU) ScrophulariaFicus SedumFumaria SmilaxGladiolus SolanumHedera ThymelaeaLathyrus (DU) Verbascum (DU)Linum Viola (DU)Lithodora ZygophyllumLonicera

Types exclusively found in surface samples (SU), dung samples (DU), and depression samples (DP).

PALBO 2453 12-8-02


Fig.3.Percentagepollendiagramandaveragecovervaluesofzoophiloustaxa.

PALBO 2453 12-8-02


3.6. Estimates of plant-species abundance

To estimate local taxon cover, average valuesthat sum to 100% are used for the abundance-coverage index (Braun-Blanquet, 1964) in phyto-sociological studies carried out over broadly sim-ilar areas of about 2000 m2 at localities selectedhere (Alcaraz et al., 1989; Peinado et al., 1992;Mota et al., 1997). In order to make them com-parable with pollen taxa percentages, vegetationdata of higher taxonomic ranks were eventuallyused instead of the original species informationat the phytosociological tables. We are aware,of course, that sampling methodology here ishardly the best that could be wished for, but thereare no other ecological studies that provide morereliable estimates of plant-species abundance forthe area. In any case, we consider that, sincethis phytosociological approach permits usingpercentages as units of coverage index, givesweight to characteristic species that are not par-ticularly abundant, and establishes, albeit roughly,the plant-species abundance over relatively largeareas, it serves quite well our overall purpose. Inorder to facilitate pollen^vegetation comparison,cover values for species and species groups aregraphed within the pollen diagrams (Figs. 2and 3).

3.7. Reference pollen sites

Mean pollen percentages of selected taxa fromthe following pollen records (Fig. 1) have beenrepresented in the pollen diagrams (Figs. 2 and3):(i) La Plata Cave (Mazarro¤n, Murcia), a karstic

cave 1 km from locality 5 of this study. Averagevalues derive from 10 pollen samples of £oor-sur-face sediment (Prieto and Carrio¤n, 1999; Navar-ro, 2000);(ii) SU-8103, a late glacial to early Holocene

pollen sequence from an o¡shore marine core be-side the Murcian coast (Parra, 1994);(iii) Perneras Cave (Lorca, Murcia), a pollen

sequence from calcreted Middle Paleolithic beds(Carrio¤n et al., 1995). The site is near Los Ce-peros (locality 6);(iv) Antas (Almer|¤a), a Holocene pollen se-

quence obtained in the lower course of the RiverAntas, Vera Basin (Pantaleo¤n-Cano, 1997), veryclose to locality 8;(v) Almizaraque (Almer|¤a), average values of

pollen samples from ¢ve open sections in Chalco-lithic settlement areas along major streams andnear estuaries associated to the Vera Basin, within10 km of the coast (Mariscal, 1992; Davis andMariscal, 1994);

Fig. 4. Synthetic diagram representing mean pollen percentages of selected taxa in surface, dung, and depression samples.

PALBO 2453 12-8-02


(vi) El Cautivo (Tabernas, Almer|¤a), a Holo-cene pollen sequence from a valley-bottom depos-it in the Tabernas badland area (Nogueras et al.,2000), relatively near locality 13;(vii) San Rafael (Campo de Dal|¤as, Almer|¤a), a

late Pleistocene and Holocene pollen record froma coastal marshland system in the vicinities oflocality 15 (Pantaleo¤n-Cano, 1997);(viii) Roquetas de Mar (Campo de Dal|¤as, Al-

mer|¤a), a Holocene pollen record recovered fromnearby San Rafael (Yll et al., 1994; Pantaleo¤n-Cano, 1997);(ix) 11-P, a Holocene pollen sequence from an

o¡shore marine core of the Albora¤n sea at Al-mer|¤a (Targarona, 1997).

3.8. Detrended correspondence analysis (DCA)

DCA was carried out on a matrix of a data-base including cover values, pollen percentagesof surface, dung, and depression samples, andaverage pollen percentages of selected taxa inregional reference pollen sequences (Fig. 7).DCA options (STATISTICA version 3.05) in-cluded downweighting of rare species, weights ap-plied to columns in sequential order, and axisrescaling.

4. Results

Anemophilous pollen taxa (Artemisia, Poaceae,Olea, Pinus, Chenopodiaceae, Ephedra, Cupressa-ceae, Quercus, Cyperaceae) are over-representedin surface and especially depression samples(Fig. 2), plausibly due to to the fact that theirpollen is produced in large quantities, and dis-persed over large distances by wind. In particular,Quercus and Pinus occur relatively abundantly inthe absence of local producers. Minor anemophi-lous types such as Alnus, Betula, Castanea, Ce-drus, Corylus, and Fraxinus, not present in theregional £ora, were only recorded in depressionsamples (Table 2).Dung pollen spectra give the best re£ection of

coastal vegetation in terms of occurrence of minorpollen taxa that are crucial for characterizing lo-cal £oristic assemblages. This would be the case

with Maytenus at localities 1 and 12, Periploca inlocalities 2, 5, 6, and 12, Withania and Ziziphus atlocality 14, and Osyris at locality 3 (Fig. 3). Dungsamples record sometimes zoophilous pollen typeswhere the producing species is absent locally.Thus, Calicotome pollen occur at localities 1, 5^12, 14, and 15, Maytenus at localities 6, 7, and 13,Periploca at localities 13 and 15, Ziziphus at local-ities 3, 10, 11, and 13, and Osyris at localities 4, 7,12, and 14 (Fig. 3).When zoophilous vs. anemophilous pollen per-

centages are considered together, dung sampleso¡er the most abundant records for the ¢rstgroup (Figs. 4^6). Likewise, they display thebest pollen-analytical potential in terms of highnumber of taxa, relatively high total pollen con-centration, and low frequencies of indeterminablepalynomorphs (Fig. 5). Finally, dung pollen spec-tra alone contain pollen of minor zoophilous taxasuch as Aristolochia, Ceratonia, Chaenorrhinum,Dorycnium, Echium, Fagonia, Lathyrus, Myosotis,Neatostema, Polygala, Reseda, Verbascum and Vi-ola (Table 2).Surface samples, excluding those from depres-

sions, may show relatively high amounts of par-ticular zoophilous pollen types, but only at local-ities where the producing species show high cover.Thus, Periploca pollen is only relatively abundantat localities 7 and 10, Maytenus at localities 3, 14,and 15, Withania at locality 15, Ziziphus at local-ities 4, 8, and 12, Asparagus at localities 4 and 12,Genista at localities 9, 11, and 13, Calicotome atlocality 2, Osyris at localities 6 and 10, Euzomo-dendron at locality 13, and Arisarum at locality 12(Fig. 3). Clearly, pollen of zoophilous scrub isshed almost directly into the ground and showsscant lateral travel. Since it is not dispersed bywind over large distances, this pollen is poorlyrepresented in the regional airborne pollen depo-sition.Zoophilous scrub species characteristic of the

present-day coastal vegetation are poorly repre-sented or totally absent in depression pollen sam-ples (Fig. 3). Not a single pollen grain was foundof Periploca at localities 1, 2, 5, 6, 8, 12, and 14,of Withania at localities 1, 3, 4, 5, and 14, ofAsparagus at localities 1, 10, 12, and 14, of Lyci-um at locality 12, of Rhamnus at localities 3 and

PALBO 2453 12-8-02


12, or of Osyris at locality 10. Very low amountswere found of Euzomodendron pollen at locality13, Maytenus at localities 13 and 15, Periploca atlocalities 3, 7, and 10, Withania at locality 15,Genista at localities 9 and 11, and Calicotome atlocality 2. The abundance of Nerium, Tamarix,and Limonium (Fig. 3), and exclusive occurrenceof Centaurium, Ulmus, Populus, Typha, Juncus,Polygonum, Messembryanthemum, and Equisetumin depression samples (Table 2), may be related to

local presence and/or water transport from otherbeds of the same depositional system.DCA shows the distribution of the samples

along the ¢rst two axes, AU1 and AU2, whichaccount for 49% and 26% of the total variation(Fig. 7). It is suggested that this distribution re-sults rather from sample type than in£uencedfrom locality. Dung samples are relatively closelylocated to cover samples, while depression sam-ples group rather with continental pollen sequen-

Fig. 5. Curves of variation of the number of zoophilous taxa, total number of taxa, indeterminable percentages, and total pollenconcentration.

PALBO 2453 12-8-02


ces of Antas, San Rafael, Roquetas de Mar, andEl Cautivo. Surface samples occupy intermediatecoordinates, although several of them fall close totheir respective cover samples (e.g. 5, 6, 9, 13).Marine sequences SU-8103 and 11-P display iso-latedly the highest values of AU1. Archeologicaland cave pollen records were not grouped. Perne-ras falls within a group made up of depressionsamples. La Plata falls close to the surface sam-ples that show highest values of AU2. Almizara-que falls relatively close to cover samples 7, 3, 10,and 14. According the ¢rst DCA axis, dung andsurface samples are separated from the coverageand depression samples. It is con¢rmed that evendung samples do better re£ect local vegetationthan depression samples in which anemophiloustaxa can result in a biassed representation of thevegetation.

5. Discussion

5.1. Conventional pollen spectra are in£uenced bylong-distance transport

Pollen spectra of the category ‘depression sam-ple’ contain very little pollen produced by the

surrounding zoophilous vegetation. Rather, theyare dominated by Pinus, Quercus, deciduous trees,Artemisia, and Chenopodiaceae, with minor oc-currences of pollen from hydrophilous and halo-philous communities (Nerium, Tamarix, and Li-monium) (Figs. 2^4, Table 2). A major problemis that most of the analyzed localities have at leastsome connection with water-courses and intermit-tent £oods, thus a distinction should be madebetween the di¡erent contributions of pollenbrought by water and wind. Establishing pollensources for these taxa and other like Poaceae,Lygeum, Olea, Lamiaceae, Cichorioideae, Asteroi-deae, and Ephedra is di⁄cult because these taxaare widely represented in both local and regionalenvironments.The category ‘surface samples’ gives a better

picture of the surrounding vegetation, with signif-icant pollen amounts of Withania, Ziziphus, Lam-iaceae, Cistaceae, Genista type, and other zoophil-ous taxa (Fig. 4), but they are still biassedtowards anemophilous pollen taxa (Fig. 6).Other palynological investigations in arid areas

coincide with this study in ¢nding that pollenspectra may be greatly prone to over-representa-tion of extra-local pollen. Long-distance transportof pollen by winds, rivers, and even sea currents,

Fig. 6. Relative abundance of zoophilous and anemophilous taxa in regional vegetation, and pollen samples from recent and fos-sil sites.

PALBO 2453 12-8-02


which only has a subordinate e¡ect on pollenspectra from richly vegetated regions, is of primeimportance in arid areas (Horowitz, 1992). Pollen-rain studies in the Sahara have shown that largepopulations of anemophilous plants (Chenopodia-ceae, Poaceae, Artemisia, Olea, Cyperaceae), evendistant ones, are the principal components of pol-len spectra, while trees and shrubs of tropical andequatorial regions (e.g. Combretaceae, Acacia,Maerua, Lannea, Ziziphus, Celtis) release littlepollen into the air and thus are under-represented(Van Campo, 1975; Schulz, 1980; Ritchie, 1987).On the other hand, pollen originating nowadaysin the Sahara Desert and Nile Delta, very com-mon in silt in Israel due to dust storms, can makeup 20^30% of recent spectra near Jerusalem (Hor-owitz et al., 1975). Modern lowland pollen spectrafrom the Sudanian Sahel (El Ghazali and Moore,1998) is principally made up of the anemophiloustaxa Poaceae, Aerva and Typha, and althoughnumbers of anemophilous elements are far lessthan zoophilous ones (eight out of 26 taxa), theirtotal pollen percentage is considerably greater.Modern surface samples from the arid southwest-ern USA generally record less than 40 pollen

types of which only ¢ve, namely Pinus, Juniperus,Poaceae, Chenopodiaceae, and Asteraceae, mayaccount for 90% of the pollen counts. In contrast,Cactaceae, and the dominant Sonoran desert spe-cies Larrea tridentata, are under-represented insurface pollen samples (Hall, 1985).O¡shore and delta sediments are also prone to

pollen brought from great distances by rivers andsea currents, which can mask dry vegetation ofthe nearby coastal territories. In modern sedi-ments of the Dead Sea, desert plant pollen wasconsiderably outnumbered by pollen of Mediter-ranean vegetation and carried via the River Jor-dan (Rossignol, 1969). In the Bay of Elat insouthern Israel, modern pollen spectra containedup to 80% of pollen derived from sea-bottommuds o¡ Lebanon and northern Israel, not tomention pollen carried by sea currents £owingacross 25^30‡ of latitude (Horowitz, 1992). Inthe central Negev, wadi sediments contained upto 77% Chenopodiaceae, and 19% Poaceae^Cy-peraceae, while dust collected from the surfaceon nearby highlands showed Poaceae^Cyperaceae91%, Chenopodiaceae 6%, and Fabaceae 2%(Horowitz, 1979).

Fig. 7. DCA including cover values, pollen percentages of surface, dung, and depression samples, and average pollen percentagesof selected taxa in regional reference pollen sequences.

PALBO 2453 12-8-02


5.2. Zoophilous pollen types are absent in marshand o¡shore regional pollen sequences

It is hardly surprising that pollen of Maytenus,Periploca, Withania, Ziziphus, Lycium, Calico-tome, and Osyris was absent from Antas, San Ra-fael, El Cautivo, and Roquetas de Mar, and SU-8103 and 11-P o¡shore cores (Fig. 3), that pinepollen dominates marine pollen spectra, and Che-nopodiaceae prevail in marsh pollen sequences(Figs. 2, 6 and 7). Like depression samples, thesesediments have connection with water-courses andintermittent £oods. Yet it is worth wonderingwhether several of these studies could have paidinsu⁄cient attention to the less numerous pollengrains such as Ziziphus, which, albeit with lowrepresentation, occur in North African pollen dia-grams of Ait Blal, Ksabi, Moulouya (Ballouche,1986), and Tizi n-Inouzane in the Central Atlas(Reille, 1976), and several southern African sites(Scott, 1982). Minor zoophilous types carefullyidenti¢ed in marine cores have provided insightinto past vegetation characteristics along conti-nental northwestern Africa (Agwu and Beug,1982). Furthermore, surface samples in our studyshow that these pollen grains do occur in sedi-ment, even though, for whatever reason, theyonly travel very short distances and comprisebut trivial shares of pollen spectra recoveredfrom depression deposits.

5.3. The value of biogenic deposits of animal originin paleo-vegetation studies

Dung pollen spectra represent the vegetationcomposition of the local area, and also carry im-portant information about regional vegetation.Dung producers may have ingested plant materialover a large part of the study area. It must also beregarded that each leaf surface eaten by a herbi-vore may contain regional pollen rain. For successof dung pollen spectra to represent surroundingvegetation, the mixing of fecal material from dif-ferent animals may have been crucial. Moe (1983)studied pollen contained within sheep’s feces inNorway and concluded that pollen analysis offeces provide virtually no information about thequantitative composition of the vegetation of the

sheep’s environment, but do provide valuable in-formation about their diet.Other studies support a di¡erent viewpoint.

Thompson et al. (1980) analyzed pollen from cop-rolites of the Shasta ground sloth and compared itwith plant remains in the same material. Theyconcluded that the animal diet was not in£uentialin the dung pollen spectra. Scott and Cooremans(1992) compared pollen spectra in recent birdguano, hyrax (Procavia) and dassie-rat (Petromus)midden samples with those at surface samplesfrom various parts of South Africa. Dung sampleswere not strongly in£uenced by dietary preferen-ces and re£ected local vegetation, including rela-tively high values for Aizoaceae, Liliaceae, Acan-thaceae, Asteraceae, Acacia, and Euphorbia.Similarly to these studies, packrat (Neotoma) mid-den pollen has been shown to represent both localand regional desert vegetation in the southwesternUSA, recording under-represented and/or raretypes such as Cereus, Ferocactus, Mammilaria,Opuntia, Larrea, Yucca, Ceanothus, etc. (e.g. Da-vis and Anderson, 1987). Pollen assemblages fromfresh cow dung samples are in good agreementwith composition of the present-day vegetationin the neighborhood of sampling sites, thus sub-stantiating a paleoecological interpretation of pol-len spectra obtained from Iron Age consolidatedcow dung in Zimbabwe, Botswana, and SouthAfrica (Carrio¤n et al., 2001).Deposits of faunal origin often accumulate in

archeological and prehistoric sites in caves orunder rockshelters. Under dry sedimentation con-ditions, pollen and other biogenic materials maypreserve well, thus enhancing the value of paleo-botanical analyses in these environments (Davis,1990; Navarro et al., 2000). Thus, pollen assem-blages of Pleistocene beds in Perneras Cave andLa Plata Cave (Figs. 2 and 3) include minor oc-currences of Periploca, Withania, Lycium, Aspara-gus, Genista, Osyris, Nerium, Tamarix, and otherzoophilous taxa, and abundant pollen of Lamia-ceae, Brassicaceae, Asteraceae, and Cistaceae(Carrio¤n et al., 1995; Prieto and Carrio¤n, 1999;Navarro, 2000).It can be concluded that dung accumulations

may represent relatively unbiassed pollen traps.Admittedly, pollen sequences obtained from con-

PALBO 2453 12-8-02


ventional continental and marine deposits shouldbe combined with studies of coprolites and otherbiogenic accumulations in order to achieve a morerealistic picture of past vegetation characteristicsof this and other arid territories.

Acknowledgements

This research was funded by the Spanish proj-ects ARDACHO (Ref. BOS2000-0149) and PI-17/00739/FS/01 of the Ministerio de Ciencia y Tec-nolog|¤a and Fundacio¤n Se¤neca of the Murcia re-gional government respectively. Fernando Ojedaprovided help with multivariate analysis. MichaelWalker, Paleo-Anthropology Laboratory at Mur-cia University, kindly re-wrote the English text.Ton‹i Andrade and Cristina Navarro helped withrespectively laboratory and statistical essays at thebeginning of this work. Bas van Geel, Joost Dui-venvoorden, and an anonymous referee providedconstructive comments on an earlier version ofthis paper.

References

Agwu, C.O.C., Beug, H.J., 1982. Palynological studies of ma-rine sediments o¡ the west African coast. Meteor For-schungsergeb. 36, 1^30.

Alcaraz, F., D|¤az, T.E., Rivas-Mart|¤nez, S., Sa¤nchez-Go¤mez,P., 1989. Datos sobre la vegetacio¤n de Espan‹a: provinciabiogeogra¤¢ca Murciano-Almeriense. Itiner. Geobot. 2, 1^133.

Ballouche, A., 1986. Pale¤oenvironments de l’Homme FossileHoloce'ne au Maroc. Apport de la Palynologie. Ph.D. The-sis, University of Bordeaux, Bordeaux.

Braun-Blanquet, J., 1964. P£anzensoziologie. Grundzu«ge derVegetationskunde, 2nd edn. Springer, Vienna.

Burjachs, F., Giralt, S., Roca, J.R., Seret, G., Julia', R., 1997.Palinolog|¤a holoce¤nica y desertizacio¤n en el Mediterra¤neooccidental. In: Iba¤n‹ez, J.J., Valero, B.L., Machado, C.(Eds.), El Paisaje Mediterra¤neo a Trave¤s del Espacio y delTiempo. Implicaciones en la Deserti¢cacio¤n. Geoforma,Logron‹o, pp. 379^394.

Carrio¤n, J.S., Dupre¤, M., Fumanal, M.P., Montes, R., 1995. Apalaeoenvironmental study in semi-arid southeastern Spain:the palynological and sedimentological sequence at PernerasCave (Lorca, Murcia). J. Arch. Sci. 22, 355^367.

Carrio¤n, J.S., Munuera, M., Navarro, C., Burjachs, F., Dupre¤,M., Walker, M.J., 1999. The palaeoecological potential of

pollen records in caves: the case of Mediterranean Spain.Quat. Sci. Rev. 18, 1061^1073.

Carrio¤n, J.S., Munuera, M., Navarro, C., Sa¤ez, F., 2000. Pa-leoclimas e historia de la vegetacio¤n en Espan‹a a trave¤s delana¤lisis pol|¤nico. Viejas falacias y nuevos paradigmas. Com-plutum 11, 1^28.

Carrio¤n, J.S., Scott, L., Hu¡man, T., Dreyer, C., 2001. Pollenanalysis of Iron Age cow dung in southern Africa. Veg. Hist.Archaeobot. 9, 239^249.

Davis, O.K., 1990. Caves as sources of biotic remains in aridwestern North America. Palaeogeogr. Palaeoclimatol. Palae-oecol. 76, 331^348.

Davis, O.K., Anderson, R.S., 1987. Pollen in packrat (Neo-toma) middens: pollen transport and the relationship of pol-len to vegetation. Palynology 11, 185^198.

Davis, O.K., Mariscal, B., 1994. A comparison of archaeolog-ical palynology of Almer|¤a, Spain, and coastal southern Cal-ifornia, USA. In: Davis, O.K. (Ed.), Aspects of Archaeo-logical Palynology: Methodoloy and Applications. AASPContrib. Ser. 29, pp. 75^82.

El Ghazali, G.E.B., Moore, P.D., 1998. Modern lowland pol-len spectra and contemporary vegetation in the eastern SahelVegetation Zone, Sudan. Rev. Palaeobot. Palynol. 99, 235^246.

Hall, S.A., 1985. Quaternary pollen analysis and vegetationalhistory of the Southwest. In: Bryant, V.M., Holloway, R.G.(Eds.), Pollen Records of Late-Quaternary North AmericanSediments. American Association of Stratigraphic Palynolo-gists Foundation, pp. 95^123.

Horowitz, A., 1979. The Quaternary of Israel. Academic Press,New York.

Horowitz, A., 1992. Palynology of Arid Lands. Elsevier, Am-sterdam.

Horowitz, A., Weinstein, M., Ganor, E., 1975. Palynologicaldetermination of dust storms provenances in Israel. PollenSpores 17, 223^231.

Mariscal, B., 1992. Variacio¤n de la vegetacio¤n durante el Sub-boreal. Ana¤lisis pol|¤nico en Cabezo de Brujas, Almizaraque(Almer|¤a). Rev. Esp. Micropaleontol. 23, 141^149.

Moe, D., 1983. Palynology of sheep’s faeces: relationship be-tween pollen content, diet and local pollen rain. Grana 22,105^113.

Mota, J.F., Cabello, J., Cueto, M., Go¤mez-Mercado, F., Gi-me¤nez, E., Pen‹as, J., 1997. Datos sobre la vegetacio¤n delsureste de Almer|¤a. Universidad de Almer|¤a, Almer|¤a.

Navarro, C., 2000. Potential of Pollen Records in Caves:Taphonomy and Palaeoecology. Ph.D. Thesis, Universityof Murcia, Murcia.

Navarro, C., Carrio¤n, J.S., Navarro, J., Munuera, M., Prieto,A.R., 2000. An experimental approach to the palynology ofcave deposits. J. Quat. Sci. 15, 603^619.

Nogueras, P., Burjachs, F., Gallart, F., Puigdefa'bregas, J.,2000. Recent gully erosion in the El Cautivo badlands (Ta-bernas, SE Spain). Catena 40, 203^215.

Pantaleo¤n-Cano, J., 1997. Estudi Palinolo'gic de Sediments Li-torals de la Provincia d’Almer|¤a. Transformacions del Pai-

PALBO 2453 12-8-02


satge Vegetal dins un Territori Semia'rid. Ph.D. Thesis, Uni-versidad Auto¤noma, Barcelona.

Parra, I., 1994. Quanti¢cation des Pre¤cipitations a' partir desSpectres Polliniques Actuels et Fossiles: du Tardiglaciaire a'l’Holoce'ne Supe¤rieur de la Co“te Me¤diterrane¤enne Espag-nole. Ph.D. Thesis, Universite¤ de Montpellier, Montpellier.

Peinado, M., Alcaraz, F., Mart|¤nez-Parras, J.M., 1992. Vege-tation of Southeastern Spain. Cramer, Berlin.

Prieto, A.R., Carrio¤n, J.S., 1999. Tafonom|¤a pol|¤nica: sesgosabio¤ticos y bio¤ticos del registro pol|¤nico en cuevas. Asocia-cio¤n Paleontolo¤gica Argentina, Publ. Esp. 6, 59^64.

Reille, M., 1976. Analyse pollinique de se¤diments postglaci-aires dans le Moyen Atlas et le Haut Atlas marocains: pre-miers re¤sultats. Ecol. Mediterr. 2, 153^170.

Riera, S., Esteban, A., Go¤mez, A., 1995. El depo¤sito turbosode la Can‹ada Larga del Cerro del Sotillo (1890 m, Sierra deBaza-Filabres). Estudio pol|¤nico y geomorfologico. Avancepreliminar. Actas 3 Reuniao do Quaternario Ibe¤rico, Uni-versidade de Coimbra, Coimbra, pp. 491^497.

Ritchie, J.C., 1987. A Holocene pollen record from Bir Atrun,northwest Sudan. Pollen Spores 29, 391^410.

Rossignol, M., 1969. Se¤dimentation palynologique re¤centedans la Mer Morte. Pollen Spores 11, 17^38.

Sa¤nchez-Go¤mez, P., Guerra, J., Coy, E., Herna¤ndez, A., Fer-na¤ndez, S., Carrillo, A.F., 1998. Flora de Murcia. DiegoMar|¤n, Murcia.

Schulz, E., 1980. Zur Vegetation der o«stlichen zentralen Sa-hara und zu ihrer Entwicklung im Holoza«n. Wu«rzburgerGeographische Arbeiten 51, Wu«rzburg.

Scott, L., 1982. Late Quaternary fossil pollen grains from theTransvaal, South Africa. Rev. Palaeobot. Palynol. 36, 241^278.

Scott, L., Cooremans, B., 1992. Pollen in recent Procavia (hy-rax), Petromus (dassie rat) and bird dung in South Africa.J. Biogeogr. 19, 205^215.

Targarona, J., 1997. Climatic and oceanographic evolution ofthe Mediterranean Region over the last GlacialÎnterglacialtransition. A palynological approach. LPP Contribution Se-ries 7, Utrecht.

Thompson, R.S., van Devender, R.S., Martin, P.S., Foppe, T.,Long, A., 1980. Shasta ground sloth (Nothoceriops shastenseHo¡stetter) at Shelter Cave, New Mexico: environment, dietand extinction. Quat. Res. 14, 360^376.

Van Campo, M., 1975. Pollen analyses in the Sahara. In:Wendorf, F., Marks, A.E. (Eds.), Problems in Prehistory:North Africa and the Levant. Southern Methodist Univer-sity Press, Dallas, TX, pp. 45^64.

Yll, E.I., Pantaleo¤n-Cano, J., Pe¤rez-Obiol, R., 1994. Ana¤lisispol|¤nico de una secuencia holoce¤nica en Roquetas de Mar(Almer|¤a). In: Mateu, I., Dupre¤, M., Gu«emes, J., Burgaz,M.E. (Eds.), Trabajos de Palinolog|¤a Ba¤sica y Aplicada.Universidad de Valencia, Valencia, pp. 189^198.

PALBO 2453 12-8-02


Documents

Ataphonomicstudyofmodernpollenassemblagesfromdung ... Carrion_RPP.pdftum,Rhamnusoleoidessubsp.angustifolia,Pista-cialentiscus,Oleaeuropaeavar.sylvestris,Stipa tenacissima, Ephedra