Pollen and spores from surface samples in the campos region of Uruguay and their paleoecological implications

In this study we describe and illustrate pollen and spores that have been identifi ed as signifi cant in modern and fossil samples from the campos region of Uruguay. We provide new information about modern pollen assemblages and their relationship to the vegetation types of this region. We discuss the taxonomic limitations of pollen and spores and their representation in both modern and fossil samples. We highlight the importance of identifying both silent and key indicator taxa for making accurate paleoecological interpretations. We also emphasize the importance of knowing the pollination strategies of parental plants because many of them are zoophilous, and so small changes in the proportions of their pollen grains in fossil assemblages could refl ect important changes in vegetation. Th is study presents a practical approach to paleoecological research, which not only produces robust results in studies of the campos region, but can be applied to other grassland ecosystems, including those in temperate regions.


Introduction
Pollen analysis, the study of fossil pollen and spores, is the principal technique for long-term vegetation reconstructions (Seppä 2007), and it has provided fundamental advances in the understanding of the processes and mechanisms of ecological change (e.g.Jackson 2007;Rull 2010;Willis et al. 2010).However, pollen records do not directly refl ect plant abundance due to several factors such as diff erences in pollen production, dispersal and preservation of taxa (Prentice 1988), which cause that taxa may be either over-represented or under-represented in pollen samples.Species with anemophilous pollination, which produce high quantities of pollen grains, are frequently overrepresented in pollen assemblages while zoophilous, with low pollen productivity, are under-represented (Faegri & Iversen 1989).Palynologist must work with an incomplete representation of taxa because many fl oristically important families contribute little or no pollen to pollen rain (Bush et al. 2001).
One of the principles on which pollen analysis is based is the possibility of identifying diff erent pollen and spores according to their morphology (Faegri & Iversen 1989).Th ere are some diffi culties in studying modern pollen-vegetation relationships in Río de la Plata grasslands (RPG) (Southern South America) related to the limited pollen taxonomic resolution of their most abundant parental plants, such as Poaceae, Cyperaceae, Asteraceae subf.Asteroideae and some genera of the Apiaceae family (Tonello & Prieto 2008;Mourelle & Prieto 2012).However, there are a great number of minor taxa that allow the qualitative and quantitative analysis of pollen-vegetation relationships.
In particular, palynological analyses in the campos region of Uruguay revealed that pollen and spores in surface samples (Mourelle & Prieto 2012) and Holocene records (Iriarte 2006;García-Rodríguez et al. 2010;Mourelle et al. 2015a;b) are very diverse.However, even though several pollen and spores atlas (e.g.Bauermann et al. 2013;Radaeski et al. 2014) and catalogues (e.g.Neves & Bauermann 2004;Macedo et al. 2009;Evaldt et al. 2013;Masetto & Lorscheitter 2014) in the campos region of Rio Grande do Sul have been published during the last decades, there are no morphological descriptions and illustrations for the most frequent and abundant pollen and spores that appeared in surface and fossil samples from the campos region of Uruguay.
For this reason we propose the following aims in this study: (1) to describe and illustrate the main pollen and spores from surface samples from the campos region of Uruguay; (2) to provide new information about modern pollen assemblages and their relationship to the vegetation types of this region; and (3) to discuss the taxonomic limitations of pollen and spores and their representation in both modern and fossil samples, and provide a practical approach to obtaining robust results for paleoecological research in the campos region.

Study area
RPG extend between 28° and 38°S latitude, covering approximately 700,000 km 2 of eastern Argentina, Uruguay and southern Rio Grande do Sul (Brazil) (Fig. 1) and is characterized by the dominance of grasses and other herbs.Based on geomorphological, hydrological and edaphic features and their link with natural vegetation and land use, the RPG were divided by León (1991) into the pampas in the eastern Argentina, and the campos in Uruguay and southern Rio Grande do Sul (Brazil).The presence of woody species and grasses with C4 metabolism gradually diminishes from north to south, one of the facts that allowed the division of this sub-region in southern and northern campos (León 1991) (Fig. 1).Most of these shrubby and arboreal species in Uruguay reach their southern distribution range and so they are absent in the pampas.These RPG are climatically determined (Bilenca & Miñarro 2004).The climate is temperate, with less extreme conditions than those in other regions located at similar latitudes due to the Atlantic Ocean moderating effect (Soriano 1991).
Natural grasslands are the predominant vegetation type of the campos region (Fig. 1) and are extended all over Uruguay (ca.77% of the territory), in flat and gentlyrolling relief, hills, gentle hills and low mountains (Puerto 1987; Lezama et al. 2011) (Tab. 1).Woody vegetation covers approximately 4.3% of the territory (Petraglia & Dell'Acqua 2006) and is composed of about 300 species of trees and shrubs, characteristic elements of forests and shrublands, respectively (Brussa & Grela 2007;Haretche et al. 2012).Forests, shrublands and other herbs-dominated vegetation (not grasslands) develop locally in environments with particular edaphic conditions that define each different vegetation type (Tab.1).
After the European settlement, the campos region of Uruguay has progressively become one of the most important areas of livestock and grain production in the world (Bilenca & Miñarro 2004).Furthermore, the expansion of cropland areas, the introduction of exotic grass species and their associated weeds, and more recently the substitution of grasslands for exotic forest plantations, deeply modified the original landscape.

Selection of pollen and spores taxa
Described and illustrated taxa were selected from 62 surface samples from the campos region of Uruguay based upon two criteria: (1) their abundance and frequency in surface samples and (2) their occurrence in fossil records previously published (Mourelle et al. 2015a; b) (Fig. 1).A total of 46 surface samples were previously studied (Mourelle & Prieto 2012) and 16 further samples were collected and incorporated in this study (additional data are provided in Mourelle 2015).All samples were collected following Adam & Mehringer (1975) sampling strategy and they correspond to seven different vegetation types: grasslands, riparian forests, low-mountain shrublands-forests; psammophilous forests, "Candela" coastal shrublands, bunchgrasses and marshes (Fig. 1).The sample sites are principally located in the eastern part of Uruguay because: (1) it concentrates a larger number of different vegetation types than the western part; (2) it has minor human disturbance; and (3) it is where the fossil pollen sequences studied are located (Fig. 1).

Laboratory work
Surface samples were dried at 60°C for 24 h.Subsamples of 5 to 15g dry sediment were used for pollen analysis.Sediment samples were sieved trough a 110 mm screen to remove coarse particles and then processed according to standard palynological techniques including warm KOH, HCl, heavy-liquid separation with ZnCl 2 , HF and acetolysis (Faegri & Iversen 1989).

Pollen and spores descriptions
Standard morphological descriptions of palynomorphs (pollen and spores) follow Punt et al. (2007) and Hesse et al. (2009).Pollen shape classes are described according to ratios of their polar and equatorial axes (P:E ratios) in agreement with Erdtman (1971).Descriptions were checked using the pollen and spores reference collection of the Laboratory of Palaeoecology and Palynology, Universidad Nacional de Mar del Plata, Argentina.Identifications were supplemented by published atlas and photographs (Markgraf & D'Antoni 1978;Prieto & Quattrocchio 1993;Pire et al. 1998;2001;2006;Bauermann et al. 2013).
Under the light microscope, at least 20 pollen grains or spores per each taxonomic entity were measured to elaborate each morphological description, using a 100x oil immersion objective.Measurements were made using an ocular micrometer in the microscope.All images were taken by a Nikon Coolpix P3 digital camera.Pollen slide preparations have been deposited in the palynological collection at the Laboratory of Paleoecology and Palynology, Universidad Nacional de Mar del Plata, Argentina.
For pollen and spores nomenclature we followed some conventions recommended by Birks & Birks (1980): (1) if pollen or spores types are known to be produced by only one species they are named after this species (e.g.Phyllanthus sellowianus); (2) if a complete family produces pollen or spores that cannot be morphologically distinguished, the type is named after the family (e.g.Poaceae); (3) if the pollen or spores morphological category is known to be produced by two taxa, then it is named after these two taxa (e.g.Lithraea/Schinus); (4) if three or more plant taxonomical entities are known to produce the same morphological pollen or spore type, the suffix "type" is added after the taxonomical name (e.g.Alternanthera-type).Since there are significant differences between the concepts of "plant taxa" and "palynomorphological types" and in order to provide the necessary clarity on the nomenclature, names of pollen and spores were put in small capitals and refrain from using italics (Klerk & Joosten 2007).
Pollen were classified according to the classification system APG III (2009) and spores in accordance with the Missouri Botanical Garden (MOBOT 2016).The palynological terminology follows Punt et al. (2007) and Hesse et al. (2009).
On the other hand, pollen and spores from surface samples were counted using a microscope with 1000x magnification.A minimum of 300 pollen grains were counted, including pollen from herbs and aquatic herbs (herbaceous pollen), shrubs (shrubby pollen), trees (arboreal pollen) and climbers, and they represent the total pollen sum.Due to the over-representation of Poaceae in grassland samples, 300 pollen grains excluding Poaceae were counted in order to obtain a good representation of the subordinated taxa.Pteridophytes and bryophytes were counted and expressed as percentages of the total palynomorphs sum.Palynomorph percentages were calculated and diagrams were drawn using TGView 2.0.4 program (Grimm 2004).(Chebataroff 1960;Puerto 1969;1987;León 1991;Alonso-Paz & Bassagoda 1999;2002;Fagúndez & Lezama 2005;Brussa & Grela 2007;Lezama et al. 2011) and the characteristic pollen and spores from modern surface samples in the campos region of Uruguay.

Vegetation type
Habitat Vegetation frequent taxa Surface samples: characteristic taxa

Numerical analyses
Numerical analyses were performed in order to determine whether the vegetation types are associated with characteristic pollen spectra from which the parent vegetation can be identified.Pollen types were selected if the mean value of each type in percentage was higher than 2% in at least one sample (spores were excluded due to their local over-representation).The percentage data were square root transformed prior to numerical analyses in order to stabilize their variances.Pollen samples were classified into groups using Cluster Analysis unconstrained (Birks & Gordon 1985) with Euclidean distance (Overpeck et al. 1985), performed with TGView 2.0.4 program (Grimm 2004).
In order to evaluate the woody pollen dispersal from riparian forests to grasslands, seventeen grassland surface samples were analyzed considering (1) their distance to the riparian forests and (2) their woody pollen content.Grassland samples were selected only if a riparian forest represented the closest vegetation type.Woody pollen dispersal was assessed by plotting woody pollen content in grassland surface samples (as percentages) versus the distance to the nearest riparian forest (in meters).The power trend line has been added to the scatter plot.

Results
From the 62 surface samples, the total palynomorph sum reached a total of 86 pollen and spores related to the natural vegetation of the campos region of Uruguay.Only 35 pollen grains and 8 spores were the most representative in terms of abundance and frequency (Figs. 1, 2; Tab. 1).They are described in Tables 2 and 3 and illustrated in figures  3, 4 and 5. Unconstrained Cluster Analysis shows distinctive palynological assemblages which characterized the main vegetation types of the campos region of Uruguay (Fig. 6).Seven vegetation types in tree groups of pollen assemblages were identified.Group I represents grasslands (Ia), bunchgrasses (Ib) and marshes (Ic).Group II comprises samples from "Candela" (Dodonaea viscosa) coastal shrublands (IIa) and from low-mountain shrublands-forests and psammophilous forests (IIb).Group III represents riparian forests.
The power trend line clearly demonstrates that the woody pollen from riparian forests shows very low dispersion capacity to grasslands.R-squared value of 0.86 indicates a very good fitness of the curve to the data set (Fig. 7).
The 43 palynomorph from surface samples described in this study were also the most abundant and frequent in the Arroyo Solís Grande (ASG) and Paso Barranca-India Muerta (PB-IM) fossil sequences (Mourelle et al. 2015a; b) (Fig. 8).They represent 52% and 58% of the pollen types that compose the total palynomorphs identified in ASG and PB-IM sequences, respectively (additional data are provided in Mourelle 2015).

Discussion
The importance of these pollen and spores, in both modern and fossil samples, is discussed below and represents the basis of this study, as well as it provides a practical approach to obtaining reliable results.

Modern pollen-vegetation relationship
Pollen analyses from modern surface samples revealed the good relation between modern pollen assemblages and the principal vegetation types of the campos region of Uruguay, mainly related to either the dominance of some taxa (e.g.Poaceae in grasslands; Dodonaea viscosa in "Candela" coastal shrublands) or to the presence of key indicator taxa (e.g.hydrophilous woody taxa in riparian forests) (Figs. 2, 6).However, the low taxonomic resolution of some pollen taxa (e.g.Asteraceae subf.Asteroideae, Fig. 5U-V; Myrtaceae, Fig. 5A; Rhamnaceae, Fig. 4V) and the absence of pollen from some species that develop exclusively in some vegetation types (e.g.Sideroxylon obtusifolium, Rollinia maritima and Varronia curassavica in psammophilous forests; Juncus acutus in brackish marshes) make it difficult to differentiate some vegetation types only by the palynological assemblages.As a consequence, a loss of ecological information is expected.
Grassland pollen assemblages are dominated by herbs (ca.80%; Group Ia, Fig. 6) while riparian forests are dominated by woody pollen taxa (ca.70%; Group III, Fig. 6).Because riparian forests are dense, pollen dispersed from grasslands cannot easily enter into these forests.Consequently, grasslands pollen signal, regionally dominant, is under-represented in the pollen spectra from samples taken inside the riparian forests, although they consist mostly of trees and shrubs with zoophilous pollination (Tab.3).Since other forests and scrublands of the campos region are less dense, the grassland regional pollen signal is better represented in such pollen assemblages, which are co-dominated by herbs, trees and shrubs (Group II, Fig. 6).
In grassland surface samples shrubby and arboreal pollen is present in traces, mainly represented by mesophilous taxa (Myrtaceae; Myrsine, Fig. 5-R and Celtis, Fig. 4U).Blepharocalyx salicifolius (Myrtaceae) and Myrsine laetevirens trees reach up to 10-15 m height and constitute the forest canopy, over which Celtis iguanaea climbs.As a result, pollen of these mesophilous taxa, besides falling gravitationally into the forest, can be more easily dispersed into grasslands than pollen from hydrophilous taxa, which only reach up to 4 m height and grow restricted to water body margins (e.g.Phyllanthus sellowianus and Cephalanthus glabratus).In addition, the frequent presence of Celtis and Myrsine in grassland samples could be related to wind dispersal (Otegui & Cocucci 1999;Gassmann & Pérez 2006;Torretta & Basilio 2009).
Uniform morphology of Poaceae pollen presents problems in the reconstruction of past grassland ecosystems and, therefore, many efforts have been made in order to elaborate a method to differentiate grassland ecosystems based on measurable Poaceae pollen parameters.Jan et al. (2015) observed that neither the scores of C3/C4 Poaceae species nor those diploid/polyploid species are clearly separable, despite there are clear trends showing that pollen sizes are larger in C4 as compared with C3 species, buy also polyploids are larger than diploid species.Grasslands dominated by C3 and C4 Poaceae species from different regions and habitats could then be separated by studying the pattern of trends in their pollen size (Jan et al. 2015).In this regard, Schüler & Behling (2011) used Poaceae pollen size as a tool for distinguishing past grasslands in South America, demonstrating that a distinction between grassland ecosystems is possible based on measurable Poaceae pollen grain parameters.They observed that the RPG Poaceae pollen was considerably smaller compared to the one from other grassland regions, but they did not find significant differences between Poaceae pollen from the campos and the pampas regions.In particular, for many taxa from different genus or tribes present in the campos region, only the extreme forms are readily separable, but there is a continuum of forms blurring the distinction between the extremes (Radaeski 2014).A false separation of forms is statistically more misleading than a lumping of types, although this involves losing ecological information for paleoecological reconstructions.In consequence, we still suggest considering Poaceae pollen grains at the family level in combination with minor taxa of grasslands.This also applies to Asteraceae subf.Asteroideae which, despite differences between their pollen grains are clearer (e.g.Punt & Hoen 2009;Cancelli et al. 2010), the difficulty classifying all Asteraceae subf.Asteroideae pollen present in modern and fossil assemblages might generate misleading paleoecological interpretations.Therefore, we propose considering them at the subfamily level.Therefore, the limited taxonomic resolution of the most abundant pollen in grassland samples (Poaceae, Cyperaceae and Asteraceae subf.Asteroideae) makes it impossible to differentiate samples from northern and southern campos at a regional scale since this division is based on different Poaceae genera (Mourelle & Prieto 2012).However, some divisions at a local scale are possible (see below).
The presence of shrubby and arboreal pollen in the campos differentiates their pollen assemblages from those of the pampas where these taxa are absent, taken into account that forest and shrubland hardly develop in the pampas region.Samples belonging to grasslands where palm individuals grow could be differentiated from the others by the presence of Butia pollen (Fig. 4D).However, not all surface samples from grasslands where palm individuals develop exhibited Butia pollen.This absence prevents us to interpret this as a lack of palms in the vegetation (Mourelle & Prieto 2012).

Marshes and bunchgrasses
Pollen assemblages of brackish and freshwater marshes are co-dominated by Poaceae and Cyperaceae.Most Poaceae pollen comes from the regional grasslands, while Cyperaceae comes from plants from soil depressions where water accumulates, promoting the local growth of species such as Schoenoplectus californicus and Scirpus giganteus (Rodríguez-Gallego et al. 2012) (Fig. 2; Tab. 1).The absence of Juncus acutus in brackish marshes surface samples does not allow the differentiation between them and the freshwater marshes, and highlights the importance of identifying the silent taxa in order not to misinterpret the fossil pollen spectra.Furthermore, on the Uruguayan coast there are not salt marshes dominated by Chenopodioideae, as observed in other South American areas (Isacch et al. 2006) where Chenopodioideae pollen often reaches a strong local representation (up to 80%) in Bahia Samborombón (Vilanova & Prieto 2012) and Laguna Mar Chiquita (Stutz & Prieto 2003).This fact must be taken into account when interpreting coastal fossil sequences, as undertaken in ASG, where the development of salt marshes dominated by Chenopodioideae during the mid Holocene was inferred (Mourelle et al. 2015a).Bunchgrasses are dominated by Poaceae species that develop together with some Cyperaceae in soils with high substrate water availability, and hence Poaceae and Cyperaceae also codominate their pollen assemblages (Fig. 2; Tab. 1).Relatively dense Eryngium pandanifolium clumps, locally known as "caraguatales", are associated to bunchgrasses and are also represented in their pollen spectra (Eryngium).In addition to grasslands, both marshes and bunchgrasses are observed in open environments, which favored the deposition of woody pollen dispersed from the canopy of the neighboring riparian forests (Fig. 2).
The fact that all these taxa are shared by different forest types developed in the campos region not only makes these pollen grains and spores characterize and define them in the palynological spectra but it also means that they are not useful to differentiate between forests.
For palynologists interpreting pollen diagrams is important to understand which groups of taxa are likely to be silent in the pollen record.For instance, pollen from the characteristic species of the psammophilous forest vegetation (Sideroxylon obtusifolium, Rollinia maritima and Varronia curassavica) were absent in the pollen assemblages (Fig. 2; Tab. 1).In consequence, such silent taxa prevent the psammophilous forest to be clearly identified from other forests in the pollen assemblages, mainly from low-mountain shrublands-forests (Fig. 6).However, it is important to note that pollen spectra from psammophilous forest have higher values of Ephedra triandra (Fig. 4A) and Rhamnaceae, and minor of Asteraceae subf.Asteroideae, related to the abundance of their parental plants on each vegetation type (Figs. 2,6;Tab. 1).
In contrast, some different vegetation types can be identified and statistically separated from others by a few key indicator taxa (e.g.Bush 1991;Rodgers III & Horn 1996;Ortuño et al. 2011).Pollen samples from riparian forests are dominated by woody taxa, represented by the mesophilous taxa described above, but also by hydrophilous ones that develop exclusively in these forests (Tab.1), represented in pollen spectra by Sebastiania (Fig. 4N-O), Salix chilensis (Fig. 4Q-R), Phyllanthus sellowianus (Fig. 4P) and Cephalanthus glabratus (Fig. 5S).These pollen types are riparian forest key indicator taxa.Aquatic herbs such as Myriophyllum (Fig. 4H), Polygonum (5P-Q) and Nymphoides indica (Fig. 5W) reflect plants that develop in the margin of the water bodies and in the floodplains.In addition, riparian forests understorey is more humid than in the other forest types, and that favored the development of Isoetes (Isoetes) and many pteridophytes of genera such as Campyloneurum, Pleopeltis and Microgramma (Polypodiaceae).
Ocotea pollen grains are very fragile (Behling & Negrelle 2006) and Erythrina crista-galli is pollinated by birds and so their flowers have stamens hidden deep in the corolla tube (Hidalgo 2008), two facts that make that these grains are not usually found in pollen assemblages.However, this does not generate problems concerning the correct identification of riparian forest pollen assemblages, both qualitatively (Fig. 2) and quantitatively (Fig. 6), principally due to the presence of the key indicator pollen types from hydrophilous taxa that grow exclusively in these forests.
Therefore, when we analyze the pollen spectra of a fossil sequence we must take into account that the presence of key pollen types from woody hydrophilous taxa indicate the development of riparian forests, whereas the presence of pollen types from mesophilous taxa do not allow us to distinguish what type of forest these pollen grains might be coming from.

Shrublands
There exists a considerable difference between shrubland and forest soils.Shrublands develop in soils with coarse sand, low humidity and water availability for plants, as well as low organic matter and nitrogen levels (Bartesaghi 2007).These conditions hinder the development of many taxa, mainly trees, bryophytes and pteridophytes.Dodonaea viscosa (Fig. 5G-H) dominates the pollen spectra and vegetation from "Candela" coastal shrublands (Fig. 6; Tab. 1).Asteraceae subf.Asteroideae pollen is also well represented, mainly due to shrubs such as Baccharis spp.and Eupatorium buniifolium, and herbs such as Senecio spp.

Woody pollen dispersal
Woody taxa from riparian forests have mostly zoophilous pollination (Tab. 1) and so they produce low quantities of pollen grains poorly dispersed.As a consequence, most of the woody pollen grains produced by riparian forests drop directly into the ground and have strong local effect (Figs. 2, 7).Inside the riparian forests, woody pollen represents ca.70%, while at distances >250 m woody pollen does not reach up to 2% of the pollen assemblage, represented by mesophilous taxa (Myrtaceae, Myrsine and Celtis).Blepharocalyx salicifolius (Myrtaceae), Myrsine laetevirens and Celtis iguanaea reach riparian forest canopy and so some of their pollen grains can be dispersed away to grasslands.This could also be related to Celtis and Myrsine wind dispersal (Otegui & Cocucci 1999;Gassmann & Pérez 2006;Torretta & Basilio 2009).
Pollination mechanisms will predictably bias the fossil record: pollen grains arriving to lake sediments are most likely to be from anemophilous taxa or from zoophilous ones exhibiting "messy" pollination syndromes, in the sense that excess pollen is spilled into the air, for example, due to vibrations from wings of a visiting bee (Bush & Rivera 2001).Our results reflect that just very small amounts (or even traces) of key pollen types from hydrophilous taxa found in the fossil spectra from sequences taken in open environments (such as lagoons or marshes) indicate the development of riparian forests in the surroundings.However, further detailed investigations are required to obtain more robust results.

Application to interpret Holocene sequences from the campos region
Paleoecological investigations based on pollen and spores content in fossil sediments from campos region (ASG and PB-IM, Fig. 8) brought information about environmental and climatic changes impact in vegetation during the last ca.8000 cal.yr BP. (1) Palynomorph analyses carried on ASG sequence from the marsh in the lower reach of Arroyo Solís Grande (34°45'35"S -55°25'55"W; Figs. 1, 8) allowed the reconstruction of vegetation history in the Río de la Plata northeastern coast (Uruguay) during the mid and late Holocene in relation to the relative sea-level fluctuations and the geomorphologic evolution (Mourelle et al. 2015a;Prieto et al. 2016).( 2) Palynomorph analyses carried on PB-IM sequence from the wetland Paso Barranca-India Muerta at Laguna Merín basin (33°40'36"S -53°49'15"W; Figs. 1, 8) allowed the reconstruction of vegetation history during the late Holocene in relation to edaphic and climatic conditions (Mourelle et al. 2015b).Despite fossil pollen assemblages from campos region showed considerable pollen and spores diversity, for which it is indispensable to have a good palynomorph morphology knowledge, many taxa could only be identified at genus or family level, causing a loss of ecological information that had to be taken into account while reconstructing paleovegetation.However, the good relationship between modern pollen and spores assemblages and the principal vegetation types of the campos region of Uruguay provided useful data for the interpretation of Holocene pollen sequences from this area.
The dominance of Poaceae in the fossil pollen spectra makes it imperative to analyze the other variables in great detail, paying special attention to the pollen types representing small herbs and woody taxa.The knowledge of the pollination strategies of parental plants allows the evaluation of the significance of their pollen grains in the fossil spectra.Then, small changes in their proportions in fossil assemblages have been really relevant for making accurate paleoecological interpretations.
The pollen types that dominate the pollen assemblages of the two Holocene sequences studied (ASG and PB-IM, Fig. 8) represent anemophilous plants (Poaceae accompanied by Cyperaceae) associated to zoophilous ones (Asteraceae subf.Asteroideae, Monocotyledoneae, Apiaceae and pollen from small herbs) (Fig. 2).These pollen assemblages represent the regional dominance of grasslands in this region during mid and late Holocene, with characteristics similar to the current ones.
Despite Poaceae high proportions in ASG sequence are related to the dominance of grasslands all over the region, their grains may also have a local origin.Its co-dominance with Chenopodioideae and the geomorphological characteristics of the coastal area revealed the development of salt marshes around an estuary during the sea-level rise and highstand (between ca.8000 and 2900 cal.yr BP).In consequence, Poaceae pollen probably also corresponds to Spartina sp., a dominant taxa in these environments.This scenario is comparable with the salt marsh vegetation that currently occurs in some coastal areas of the pampas (Mourelle et al. 2015a).
Amaranthoideae (Gomphrena-t ype and Alternanthera-type) and Ranunculaceae represent taxa that are very common in grasslands.However, they also represent plants that develop in humid environments and near water bodies, such as Alternanthera philoxeroides and Ranunculus spp.Other herbs that also develop there are Polygonum punctatum, Nymphoides indica, Myriophyllum aquaticum and M. quitense, and so their pollen grains (Polygonum, Nymphoides indica and Myriophyllum) were all found in the fossil assemblages.These associations represent freshwater areas that locally developed during the Holocene, but also suggest the occurrence of runoff that transported pollen from the hinterland source area to the estuary (Fig. 8).Relatively high Eryngium proportions in the fossil sequences reflect the Eryngium pandanifolium dense clumps that frequently occur in wetlands (PB-IM) and coastal areas (ASG) (Fig. 8).
The occurrence of certain pollen types in the fossil spectra, even in low proportions, turns out to be good indicators of paleoenvironmental changes.For instance, the appearance of the low salinity-tolerant Myriophyllum in ASG sequence at around ca. 5100 cal.yr BP reflect an increase in freshwater input and a decrease in salinity derived from the transition from brackish to freshwater conditions during sea-level fall (Fig. 8).At the same time, Dodonaea viscosa grains also appeared, indicating the development of psammophilous coastal vegetation in coastal sand bars (Fig. 8).On the contrary, Butia pollen was found in low proportions and only in some PB-IM samples, disappearing towards the top of the core (Fig. 8).This initially suggests the development of palms interspersed among grasslands, close to the sampled site, at least since ca.1300 cal.yr BP.However, as Butia pollen has not been found in all surface samples where palm individuals develop, its absence in the fossil assemblages prevents us to interpret this as an absence of palms in the past vegetation.Nevertheless, Butia disappearance in PB-IM assemblages corresponds to the last decades and probably reflects the palm vegetation decline due to human activities, such as pasture burning.This practice also reduces Eryngium pandanifolium clumps and Eupatorium spp.and Baccharis spp.shrubs, and this was also observed in the fossil spectra (Fig. 8).
Regarding the woody taxa, pollen from hydrophilous and mesophilous taxa were registered in the fossil sequences (Fig. 8).Key pollen types from hydrophilous taxa suggest the development of riparian forests.However, it is not possible to distinguish what type of forest or shrubland the pollen of mesophilous taxa comes from, since these plants are shared by all vegetation types.The knowledge of the geomorphology of the region (e.g. the proximity of the study area to hills) and its evolution (e.g.coastal progradation) is indispensable to determine if any of these forests or shrublands could have developed nearby in the past.
High spores proportions, mainly Phaeoceros, suggest the occurrence of runoff transported from the hinterland source area to the estuary in ASG, as these spores were not recorded in surface samples from coastal vegetation (Fig. 8).Very low proportions of deteriorated pollen grains and high pollen sum in the fossil spectra allowed us to rule out an over-representation of bryophyte spores due to poor preservation, which in addition amplifies the imperative value of the spores account.Isoetes relatively high proportions in PB-IM pollen assemblages probably reflect the edaphic characteristic of this wetland area that is usually covered by water.

Conclusions
This is the first catalogue that describes and illustrates pollen and spores in modern and fossil samples from the campos region of Uruguay.Distinctive palynological assemblages characterized the main vegetation types, mainly related to the dominance of some taxa (e.g.Poaceae in grasslands, Dodonaea viscosa in "Candela" coastal shrublands) or to the presence of key indicator taxa (e.g.hydrophilous woody taxa in riparian forests).However, the low taxonomic resolution of some pollen taxa (e.g.Asteraceae subf.Asteroideae, Myrtaceae, Rhamnaceae) and the absence of pollen from species that develop exclusively in some vegetation types (e.g.Sideroxylon obtusifolium, Rollinia maritima and Varronia curassavica in psammophilous forests; Juncus acutus in brackish marshes) makes it difficult to differentiate some vegetation types only by the palynological assemblages.The latter are considered silent taxa, and their identification proved to be very important in order not to misinterpret the fossil pollen spectra.
The knowledge of pollination strategies of parental plants was necessary to explain the presence of small amounts of key pollen types that result indicative of the vegetation (e.g.hydrophilous woody taxa indicate riparian forests development, aquatic herbs indicate freshwater areas, and Butia indicates palms development grasslands).
Studies like this one help and guide the palynologist to make identifications in surface or fossil samples, and thereby they are useful to focus on the taxa most likely to occur in the fossil record.It presents a practical approach to paleoecological research, which not only produces robust results in studies of the campos region, but can also be applied to other grassland ecosystems, including those in temperate regions.

Figure 6 .
Figure 6.Summary percentage pollen diagram of the 62 surface soil samples ordered according to the Cluster Analysis unconstrained results.

Figure 7 .
Figure 7. Woody pollen dispersal curve from riparian forests to grasslands.Two riparian forest samples were considered as pollen sources (samples 62 and 63).Numbers correspond to surface samples (see Fig. 1).

Table 1 .
Description of the vegetation types