Research papers
Modern pollen–vegetation relationships within a small Mediterranean temporary pool (western Morocco)
Btissam Amamia,b,c,⁎, Serge D. Mullerb, Laïla Rhazia, Patrick Grillasc, Mouhssine Rhazid, Siham Bouahima,b,c
aUniversité Hassan II Aïn Chock, Faculté des Sciences, Laboratoire d'Ecologie Aquatique et Environnement, BP 5366 Maarif Casablanca, Maroc
bUniversité Montpellier-2/CNRS, Institut des Sciences de l'Evolution, Case 061, 34095 Montpellier cedex 05, France
cTour du Valat, Centre de recherche pour la conservation des zones humides méditerranéennes, Le Sambuc, 13200 Arles, France
dUniversité Moulay Ismail, Faculté des Sciences et Techniques d'Errachidia, Département de Biologie, BP 509 Boutalamine, Errachidia, Maroc
a b s t r a c t a r t i c l e i n f o
Article history:
Received 18 March 2010
Received in revised form 27 June 2010 Accepted 28 June 2010
Available online 8 July 2010 Keywords:
North Africa
hydrophytic plant community surface pollen sample quantitative reconstruction indicator taxa
Morocco is rich in temporary pools which harbour numerous rare plant species. Long-term conservation of such threatened plant communities should be based on the understanding of their past dynamics. Despite conditions unfavourable to pollen preservation, surface sediments of acidic temporary pools are shown to contain pollen assemblages likely to allow vegetation reconstruction. Knowledge of the modern relationships between pollen and vegetation is, however, necessary for interpreting fossil data in terms of past vegetation.
Surface pollen assemblages andfloristic surveys of a temporary pool in Benslimane forest, western Morocco, are compared in order to evaluate the pollen record of the local hydrophytic vegetation. Floristic surveys were carried out for 12 years (1996–2008) along two crossing permanent transects. A set of 21 surface- sediment samples, taken along the same transects in 2007, were analysed for pollen. The spatial relationships between vegetation and pollen assemblages are explored by means of multivariate analyses, statistical tests and linear regressions. The calculation of representation indices moreover allows proposing quantitative ways for pollen-based plant-abundance reconstruction.
Results reveal that the vegetation structure along the hydrological gradient is well recorded in the pollen assemblages, with: (1) a marginal zone characterised by terrestrial taxa and rare amphibious taxa (Elatine, Pilularia), (2) an intermediate zone of amphibious taxa (Alisma-type,Illecebrum/Paronychia,Isoetes velata- type), and (3) a central zone of aquatics (Myriophyllum alterniflorum,Ranunculus-type). The best correlation between the pollen record and total pool vegetation was found in the centre of the pool, which supports the reliability of the study of a single core from the centre of the pool for the reconstruction of the past dynamics of the local hydrophytic vegetation. Both the qualitative‘community’approach (representation indices and indicator pollen taxa) and the quantitative‘taxa’approach (correction factors) suggest that reconstructions of past populations can be achieved from a few taxa, namelyIsoetes velata-type,Myriophyllum alterniflorum andRanunculus-type. For these taxa, regression parameters (slope and y-intercept) have been calculated between pollen percentages and plant percentages in present vegetation, and between pollen influxes and plant abundances, respectively. These parameters can be extended to interpret fossil data from other temporary pools within the same region to reconstruct their relative and absolute past plant abundances.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Temporary wetlands, defined as endoreic depressions character- ised by seasonalfloods, are found worldwide, under climates with distinct wet and dry seasons such as the Mediterranean climate (Deil, 2005). Mediterranean temporary pools harbour many rare species (Quézel, 1998; Grillas et al., 2004) adapted to their unique ecological functioning. They are strongly threatened by human activities,
including urbanisation, artificial infilling, drainage and agriculture, in particular in North Africa (Rhazi et al., 2001, 2006). Increasing anthropogenic disturbances are often responsible for severe degra- dation of their plant communities (Rhazi et al., 2006). The ongoing conservation policies are usually based on parameters such as rarity, species richness and endemism (Myers et al., 2000) and are only designed on the short-term. The development of long-term conser- vation policies requires a historical perspective, based on the knowl- edge of initiation modes and past dynamics (Froyd and Willis, 2008;
Muller et al., 2009).
Palaeoecological data are being used for reconstructing past vege- tation dynamics at the local scale (e.g.,Muller et al., 2003, 2008;
Bottollier-Curtet and Muller, 2009). However, because of conditions
⁎Corresponding author. Université Hassan II Aïn Chock, Faculté des Sciences, Laboratoire d'Ecologie Aquatique et Environnement, BP 5366 Maarif Casablanca, Maroc.
Tel.: + 212 522230680/82.
E-mail address:ibtissam_amami@yahoo.fr(B. Amami).
0034-6667/$–see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.revpalbo.2010.06.012
Contents lists available atScienceDirect
Review of Palaeobotany and Palynology
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / rev p a l b o
being generally unfavourable for pollen preservation (e.g. microbial attack, pollen oxidation and dry conditions), only few temporary pools have been investigated in the Mediterranean region (Dutil et al., 1959; Le Dantec et al., 1998; Muller et al., 2008, 2009). These works, from different geographical contexts, indicate that while pollen preservation is imperfect pollen analysis is possible. In small temporary wetlands, the reliability of vegetation reconstructions however requires previous knowledge of the relationships between pollen and vegetation. Previous studies have highlighted the sensitivity of pollen–vegetation relationships to pollen production, dispersal and preservation (Sugita, 1994; Calcote, 1995), as well as to site characteristics (Jacobson and Bradshaw, 1981; Jackson, 1990).
These works concern almost exclusively the terrestrial vegetation and rarely discuss the pollen record of local hydrophytic vegetations (Janssen and Braber, 1987; Muller et al., 2008).
The present study aims at investigating the relationships between pollen and vegetation within temporary wetlands. It is based on comparison between modern pollen spectra and vegetation surveys carried out for 12 years in a temporary pool located in western Morocco. Two approaches have been defined in order to reconstruct vegetation from pollen records: the so-called qualitative‘community’ approach and quantitative‘taxa’approach (Janssen, 1967, 1970). On thefirst hand, the‘community’approach, based on the characteriza- tion of plant communities by their total pollen record, provides information about past vegetation successions and local ecological changes within regional landscapes. Communities may also be qualitatively reconstructed using‘indicator taxa,’which indicate the possible presence of other ecologically related taxa. On the second hand, the‘taxa’approach, considering each taxon individually, aims at reconstructing quantitatively their past abundances. This approach assumes that simple linear regressions adequately reflect the rela- tionships of the respective abundance between pollen record and
vegetation (Andersen, 1970; Webb et al., 1981). The main questions addressed in this article are:
(1) How accurately do pollen records reflect the composition and structure of local vegetation?
(2) Which biases are likely to affect the pollen representation in temporary pools?
(3) What are the hydrophytic taxa likely to be used as indicators of each community?
(4) Is it possible to quantify, for selected taxa, the relationship between pollen and vegetation?
2. Materials and methods
2.1. Geographical and biological setting
The study area is located in the province of Benslimane, on the Moroccan Atlantic seaboard between the cities of Rabat and Casablanca (Fig. 1). The regional vegetation is dominated by degraded forests of cork oak (Quercus suber) and arartree (Tetraclinis articulata), in mosaic with cultivatedfields and reforestation patches of pines (Pinusspp.) and eucalyptus (Eucalyptusspp.). This region has a semi- arid Mediterranean climate, with a mean annual rainfall (1961–2006) of 450 mm and large inter-annualfluctuations (range: 142–803 mm).
The minimum and the maximum mean annual temperature are 7.5 and 29.5 °C respectively (Zidane, 1990).
The study site is a temporary pool (N 33°38′50″; W 07°05′24″; 258 m a.s.l.) located on sandstones and quartzites (Destombes and Jeannette, 1966), within the cork-oak forest of Benslimane. The site, covering a surface of 0.3 ha, is developed on hydromorphic sediments, composed of silts, clays and sands, the size of particles decreasing from the margins towards the centre (Rhazi et al., 2001). Extensive
Fig. 1.A. Map of Morocco showing the location of Benslimane region; B. Location of the pool D4(black dot) in the Benslimane province, western Morocco (drawing: E. Saber, 2009),
grazing (cattle, sheep and goats) and the presence of wild boars result in disturbances of the sediment surface.
2.2. Vegetation survey
The vegetation was monitored for 12 years (1996–2008) using 21 quadrats distributed every 8 meters along two perpendicular permanent transects crossing the deepest part of the pool (Fig. 2).
The two transects T1and T2are 80 m and 72 m long, respectively. The quadrats (0.3 × 0.3 m) are divided into 9 squares (0.1 × 0.1 m). In each quadrat, the vegetation abundance was measured as the number of occurrences, i.e. the number of squares occupied by each species (value between 0 and 9). In order to buffer the large inter-annual variations of the vegetation, the abundance of each species per quadrat is calculated as the average of the maximal abundances per year (unit: number of occurrences m−2year−1). Plant percentages are based on a vegetation sum, defined as the sum of the occurrences of all species common to both pollen and vegetation records.
Life types (Table 1) are classified as‘terrestrial’and‘hydrophytes’ related to hydrological requirements (Rhazi et al., 2001). The‘hydro- phytic’group lumps both aquatic and amphibious taxa (Table 1). In
order to standardise vegetation and pollen data, the taxonomic level of the study is defined by the precision of pollen identification.
2.3. Pollen analysis
Twenty-one surface-sediment samples were collected in May 2007 within the vegetation quadrats (Fig. 2) and stored in refrigerated plastic bags until their treatment. Pollen extraction was performed on sediment volumes of 1 cm3, followingFægri and Iversen (1989). Pollen iden- tification was based on pollen atlases (Reille, 1992–1998) and on the reference collection of theInstitut des Sciences de l'Evolution(Université Montpellier-2). The notation of taxa followsBirks and Birks (1980): the suffix‘-type’groups several taxa indistinguishable by their morphology and the prefix‘cf.’indicates that the most probable taxon name is applied. A minimum of 300 pollen grains was counted for each sample.
As the study concerns the local hydrophytic vegetation, this pollen sum includes aquatic taxa, but excludes Pteridophyte spores (Berglund and Ralska-Jasiewiczowa, 1986). Pollen concentrations (grains cm−3) are based on volumetric estimations (Davis, 1965). The calculation of pollen influx (grains mm−2year−1) is based on an estimated apparent sedi- mentation rate of 0.032 cm year−1, based on two radiocarbon dates (Table 2). This apparent rate does not integrate possible sedimentation
Fig. 2.A. Temporary pool D4(photo S.D.Muller, 2009); B. Vegetation belts, quadrats and surface-sediment samples taken along transects T1and T2within the pool D4. Squares represent quadrats measuring 0.3 × 0.3 m, as represented on the right of thefigure.
Table 1
Taxa found as pollen and/or in vegetation within the pool D4. Vegetation and pollen frequencies are calculated as the number of occurrences of taxa/total number of quadrats (= 21) and are expressed in percents. Pollen source: L, local; E, extra-local; R, regional. Life type: Aq, aquatic; Am, amphibious; Te, terrestrial. Pollination mode (followingwww.erick.
dronnet.free.frandwww.telabotanica.org): A, anemophilous (wind-pollinated); E, entomophilous (insect-pollinated); H, hydrogamous (water-pollinated); S, self-pollinated.
Taxon Code Pollen source Life type Pollination mode Vegetation frequency Pollen frequency
Alisma-type Alis L Am A 48 43
Apiaceae Apia E Te E – 43
Artemisia Arte R Te A – 5
Asphodelus Asph E Te E 14 24
Asteroideae Aste L Te E 95 100
Boraginaceae Heli L Te E 62 5
Brassicaceae Bras L Te E/S – 62
Callitriche Call L Aq H/S 81 10
Cannabis/Humulus Ca/Hu R Te A – 5
Carduus-type Card L Te E/S 24 86
Caryophyllaceae Cary L Te A 33 5
Cerealia-type Cere R Te A – 86
Characeae Char L Aq H 10 –
Chenopodiaceae Chen R Te A – 29
Cicendia/Exaculum Ci/Ex L Am S/E 19 10
Cichorioideae Cich L Te E 100 100
Cistus Cist E Te E 29 41
Corrigiola Corr L Te E 52 19
Crassula Cras L Te E 14 24
Cyperaceae Cype L Am A 95 81
Elatine Elat L Am S 43 19
Ephedra major-type Ephe R Te A – 19
Erica-type Eric R Te E – 10
Eucalyptus Euca R Te A – 33
Euphorbia Euph L Te E 33 29
Fraxinus Frax R Te A – 5
Galium Gali L Te E 10 10
Genista-type Geni R Te E – 19
Geraniaceae Gera L Te E/S 14 –
Hedysarum-type Hedy R Te E – 5
Helianthemum-type Heli L Te E – 5
Hypericum Hype L Te E 33 5
Illecebrum/Paronychia Il/Pa L Am E 57 86
Isoetes histrix Ishi L Am H 14 86
Isoetes velata-type Isve L Am H 81 100
Juncus Junc L Am A 48 62
Lamium-type Lami L Te S 5 –
Linaria-type Lina L Te E 52 19
Lotus-type Lotu L Te E 14 67
Lythrum borysthenicum-type Lyth L Am E 43 43
Malvaceae Malv L Te E/S 5 –
Mentha-type Ment L Te E 38 19
Myriophyllum alterniflorum Myal L Aq A 62 71
Myriophyllum pectinatum Mype R Aq A – 10
Myrtus Myrt E Te E 5 29
Narcissus-type Narc L Te E 62 43
Olea Olea R Te A – 14
Ophioglossum-type Ophi R Am H – 5
Osmunda Osmu R Am H – 5
Phillyrea Phil R Te E – 86
Pilularia Pilu L Am H 24 48
Pinus Pinu R Te A – 62
Pistacia Pist E Te A 5 14
Plantago coronopus-type Plco L Te A 29 42
Plantago major/media Plma L Te A 10 5
Poaceae Poac L Te A 100 100
Polygonum aviculare-type Poly L Te E 14 –
Primulaceae Prim L Te E 24 –
Quercus ilex/coccifera Quer R Te A – 24
Quercus suber Quer E Te A 10 100
Ranunculus-type Ranu L Aq E 100 86
Rosaceae Rosa L Te A 5 –
Rumex-type Rume L Te A 10 24
Spergula-type Sper L Te E/S 29 38
Trifolium-type Trif L Te E 33 43
Urtica Urti R Te A – 48
Viburnum Vibu R Te A – 15
Xanthium-type Xant R Te A/S – 91
Total richness 45 61
changes or hiatuses, due to climate variability or erosion processes.
Nevertheless, such changes are expected to have diminished for millennia, along with the progressive stabilisation of the pool system, revealed by the progressive increase infine-grained sediments. Pollen diagrams were constructed using GPalWin (Goeury, 1997). The pollen source area (Janssen, 1973; Jacobson and Bradshaw, 1981) is classified as local (inside the pool), extra-local (10 m-wide belt just outside the pool) and regional (N10 m from the pool edge). In the following text, the term‘pollen’will be used in an extended sense encompassing pollen s.s.
and spores.
In order to specify the possible effect of disturbances on the pollen distribution, the depth of the disturbances caused by domestic herbivores (cattle and sheep) in the temporary pool was systemat- ically measured along both transects (T1and T2) just after the pool had completely dried (May 2008).
2.4. Data analysis 2.4.1. Vegetation structure
The vegetation structure was studied using correspondence analysis (CA). For each taxon, the mean abundance per quadrat for 12 years (expressed in number of occurrences m−2year−1) was used.
The CA was carried out on 40 taxa, excluding the trees and shrubs (Fig. 3) characterising the surrounding cork-oak forest (Cistus,Myrtus, Pistacia,Quercus suber). The similarity between the 21 vegetation quadrats was estimated using a hierarchical classification based on Ward's distance (Ward, 1963). The quadrats were grouped according
to the principle of variance minimization, based on the Euclidean distance (Fig. 4).
2.4.2. Pollen assemblages
The similarity of pollen concentration between the samples was estimated independently for transects T1and T2, using a hierarchical classification analysis (HCA). The influence of pollination modes (Table 1) and life types on pollen concentrations was tested using variance analysis (ANOVA), because data have a normal (Gauss uni- modal) distribution. The variation in total pollen concentration along the topographic gradient was tested separately for both transects T1
and T2 using the non-parametric Kruskal–Wallis test, because the distribution of data is not normal. The variation in pollen concentra- tion along the topographic gradient was tested for each hydrophytic
Fig. 3.Ordination of modern plant taxa on the twofirst axes of the correspondence analysis (CA) performed on transects T1and T2of the pool D4.Squares: aquatic plants (Aq);
triangles: amphibious plants (Am); circles: terrestrial plants (Te). For taxa code, seeTable 1.
Fig. 4.Cluster analysis of quadrats of transects T1and T2of the pool D4. The quadrats are grouped into three belts: marginal (M), intermediate (I), and central (C).
Table 2
AMS (accelerator mass spectrometry) radiocarbon ages of unidentified organic matter contained in sediments from the central belt of the pool D4. The calibrated ages (cal BP) were computed with the CALIB 5.0 program (Stuiver and Reimer, 1993), using the calibration dataset INTCAL04.14c (Reimer et al., 2004).
Depth (cm) Laboratory code 14C age (BP) Calibrated age 2δ(cal BP)
45–47.5 Poz-28627 1530 ± 30 1520–1350
105–107.5 Poz-28634 3690 ± 60 4230–3860
taxon using ANOVA and Kruskal–Wallis test, respectively. Finally, the variation in the depth of disturbance caused by herbivores along the topographic gradient was analysed using Kruskal–Wallis test.
2.4.3. Modern pollen–vegetation relationships
The differences in frequency between hydrophytic and terrestrial taxa were examined using Kruskal–Wallis test and between local/extra- local and regional pollen sources using ANOVA. The modern relation- ships between pollen and vegetation were tested using linear regres- sions (Andersen, 1967). The correlation was studiedfirst at the scale of the pool (transects combined), and secondly per quadrat. In both cases, the pollen influx of each taxon (grains mm−2year−1) was compared to
its total vegetation abundance (number of occurrences m−2year−1). At the pool scale, the analyses were carried out on mean pollen influx and vegetation abundance.
2.4.4. Representation indices
The pollen record of vegetation percentages was tested for the different zones. The zones established by cluster analyses from vegetation data and pollen data, respectively, are not exactly identical (6 quadrats are attributed to different zones). We have conserved each zonation for linear correlations at the community scale. Indices of association (A), under-representation (U) and over-representation (O) were calculated, for all taxa present in plant and pollen datasets,
Fig. 5.Surface pollen diagram of transect T1, pool D4. Grey curves represent pollen concentrations (PC; scales in 1000 grains cm−3, not constant) and empty ones pollen percentages (%; scales constant). Dots represent percentages less than 1% and stars indicate the presence of the concerned plants at the sampling points.
using presence/absence data from the vegetation survey (Davis, 1984;
Hjelle, 1997; Bunting, 2003):
A=B0×ðP0+P1+B0Þ−1 U=P1×ðP1+B0Þ−1 O=P0×ðP0+B0Þ−1
B0is the number of samples in which both the pollen type and its associated plant are present; P0is the number of samples in which the pollen type is present in the surface sample but not the plant taxa; and P1is the number of samples in which the plant but not the pollen type is present in the surface sample. The Davis indices are used for determining the representation of each taxon:
- Strongly-associated taxa: AN65%;
- Associated taxa: 50%bAb65%;
- Weakly associated taxa: low A values and relatively high values for both U and O;
- Unassociated taxa: A = 0, and high O and U values;
- Over-represented taxa: low A values, high O values and U = 0.
2.4.5. Correction factors (R-values)
Two numerical methods based on linear regressions were applied for calculating pollen–vegetation conversion factors. The analysis was limited to the strongly-associated and associated taxa, as defined above. The first method (absolute data) studies the relationship between pollen influx (Pai, in grains mm−2year−1) and vegetation abundance (Vai, in number of occurrences m−2year−1). The equation used for estimating the correction factor (Rai) and the y-intercept (Dai) for absolute data and for the taxoniis:
Pai=Rai*Vai+Dai ð1Þ
Fig. 6.Surface pollen diagram of transect T2,pool D4. SeeFig. 5for details.
The second method (relative data) studies the relationship between pollen percentages (Pri) and vegetation percentages (Vri), both calculated on the sums of the taxa present in the respective records within the pool. The equation used for estimating the correction factor (Rri) and the y-intercept (Dri) for relative data and for the taxoniis:
Pri=Rri*Vri+Dri ð2Þ
In these linear equations, the correction factor (slope of the regression line) is considered to represent an estimate of the pollen production of the concerned taxon, while the y-intercept would refer to‘background pollen rain,’originating from beyond the defined area.
3. Results
3.1. Vegetation structure
The 102 species found in the studied temporary pool have been grouped in 48 pollen taxa (Table 1). Thefirst two axes of the CA on modern vegetation explainca. 49% of the total variance and the biplot (Fig. 3) shows a strong Guttmann effect, indicating interdependence between the two axes. The vegetation is obviously organised on a hydrological gradient, from the wettest zones characterised by
CallitricheandMyriophyllumto the driest ones, harbouring Primula- ceae andTrifolium-type. The HCA (Fig. 4) confirms this gradient and groups the 21 vegetation quadrats in the three classical vegetation belts of temporary pools (Grillas et al., 2004; Deil, 2005): marginal (M), intermediate (I) and central (C). The marginal and intermediate belts overlap partially in the studied pool (Fig. 4).
3.2. Pollen assemblages
Pollen diagrams representing surface pollen assemblages along transect T1and T2are presentedFigs. 5 and 6. Terrestrial taxa (Cistus, Phillyrea,Quercus suber) and some amphibious ones (Crassula-type, Elatine,Isoetes histrix,Juncus,Pilularia) are primarily recorded at the margin of the pool. Other amphibious taxa (Alisma-type, Cyperaceae, Illecebrum/Paronychia, Isoetes velata-type) and Asteroideae present their highest pollen percentages within the intermediate zone. Finally, pollen of aquatic taxa (Myriophyllum alterniflorum,Ranunculus-type) is more abundant at the centre of the pool. Regional terrestrial taxa, including wind-pollinated taxa such as Cerealia-type and Poaceae, show a homogeneous distribution within the pool.
The hierarchical classifications carried out on pollen taxa concen- trations found in samples of transects T1(11 quadrats) and T2(10 quadrats) distinguish the quadrats of the central and intermediate zones from the quadrats of the marginal zone (Fig. 7). The total pollen Fig. 7.Cluster analysis of pollen concentrations, arranging the quadrats of transects T1and T2in three zones: marginal (M), intermediate (I), and central (C).
Fig. 8.Variation of total pollen concentration (TPC, in grains cm−3) along transects T1and T2, pool D4. The median, the minimum and the maximum are shown for each zone.
Different letters indicate a significant difference between the concerned pair of belts (pb0.05), and identical letters indicate non-significant differences (pN0.05).
concentration varies significantly between the three zones for each transect (T1: Chi2= 6.5; dF= 2; p= 0.03; T2: Chi2= 8.01; dF= 2;
p= 0.01): it appears to decrease from the centre to the edge of the pool (Fig. 8). This is consistent with the fact that pollen concentration is significantly higher for hydrophytic taxa compared to terrestrial ones (F= 5.33;dF= 1;p= 0.02). On the other hand, pollination mode does not appear to significantly affect the pollen distribution at the local scale (pN0.05).
The spatial distribution of pollen concentration varies between the 13 hydrophytic taxa (Figs. 5 and 6; Table 3). It varies significantly (pb0.05) along the topographical gradient for Cyperaceae,Illecebrum/
Paronychia,Isoetes histrix,Isoetes velata-type,Lythrum borysthenicum- type, Myriophyllum alterniflorum, and Ranunuculus-type. There are finally no significant differences between zones in the distribution of pollen concentration for the other taxa (Alisma-type, Callitriche, Cicendia/Exaculum,Elatine,Juncus,Pilularia;Table 3).
3.3. Modern pollen–vegetation relationships
Within the studied temporary pool, 44 taxa were found in the vegetation and 61 in surface pollen samples, resulting in a total of 68 taxa (Table 1). Vegetation includes four aquatic taxa (Charophytes excluded), nine amphibious and 31 terrestrial. Pollen assemblages containfive aquatic, 11 amphibious, and 45 terrestrial taxa. 48 taxa have a local or extra-local origin, and 20 represent the regional input.
The listed taxa are predominantly insect-pollinated (48%) and wind- pollinated (38%).
The most frequent taxa common to both vegetation and pollen assemblages are Asteroideae, Cichorioideae, Cyperaceae,Isoetes velata- type,Myriophyllum alterniflorum,Narcissus-type, Poaceae andRanun- culus-type (Table 1). The relative frequencies of local and extra-local taxa in vegetation and pollen are weakly but significantly correlated (r2= 0.25;pb0.05;n= 21). While hydrophytic taxa are of course more frequent than terrestrial ones in vegetation (Chi2= 8.12; dF= 1;
pb0.01), no significant difference exists between the two groups in pollen (pN0.05). The best fits between pollen concentrations and vegetation abundance are found for Isoetes velata-type (r2= 0.47;
pb0.05;n= 21),Myriophyllum alterniflorum(r2= 0.56;pb0.05;n= 21) and Ranunculus-type (r2= 0.40; pb0.05; n= 21), and a marginal correlation (r2= 0.16; p= 0.06;n= 21) is found for Pilularia. Other
hydrophytic taxa do not present significant correlation between pollen and vegetation (pN0.05). Finally, the correlation between pollen concentration and vegetation abundance is significant in the central zone (r2= 0.25; pb0.001; N= 50), weak in the intermediate zone (r2= 0.12;pb0.01;N= 56), and insignificant at the margin of the pool (pN0.05).
3.4. Representation indices and indicator pollen taxa
Because correlations are non-significant for absolute data, the Table 4only present results concerning percentage data. These results show that the vegetation of the pool is significantly correlated to pollen records in the central and intermediate zones. Nevertheless, the pollen assemblages of the central zone also record the vegetation of the two external zones, while the ones of the intermediate zone mainly record thein-situvegetation and only weakly the vegetation of the marginal zone.
Table 5presents the Davis indices (Davis, 1984) that are calculated for 13 hydrophytic taxa. Four groups of taxa are identified based on these scores: (1) strongly-associated taxa (Isoetes velata-type,Myriophyllum Table 4
Linear regression model for the community approach, presenting the coefficient of determination (r2), thep-value, the slope of the regression line (R) and the y-intercept (D). The regression was performed on mean relative frequencies (pollen percentage;
vegetation percentage), respectively for each taxon (N = 38 taxa; n = 21 quadrats).
Because of their underground spore production that induces very high spore aggregative distribution, the pollen influxes ofIsoetes histrixandIsoetes velata-type have been divided by 100. ns = not significant (pN0.05); **=pb0.01. For vegetation data set, the correlation between pollen and vegetation was studiedfirstly at the scale of the studied pool (transects combined), and secondly at the three zones (centre, intermediate and margin).
Pollen Vegetation r2 p R D
Centre Pool 0.30 ** 0.32 1.76
Centre 0.36 ** 0.53 1.23
Intermediate zone 0.17 ** 0.24 1.99
Margin 0.03 ns – –
Intermediate Pool 0.18 ** 0.33 1.75
Centre 0.08 ns – –
Intermediate zone 0.25 ** 0.38 1.61
Margin 0.19 ** 0.28 1.88
Margin Pool 0.03 ns – –
Centre b0.01 ns – –
Intermediate zone 0.05 ns – –
Margin 0.19 ** 0.22 2.03
Table 3
Comparison of concentrations of hydrophytic taxa (= aquatic + amphibious taxa) along the topographical gradients of transects T1and T2(ANOVA and Kruskal–Wallis test). n = 21 quadrats ;dF= 2; C: centre; I: intermediate zone; M: margin. The numbers indicate significant differences between zones (pb0.05). ns = not significant;
*=pb0.05; **=pb0.01; ***=pb0.001. In the column‘Comparison’, different numbers in superscript indicate a significant difference between the concerned pair of belts, although identical numbers in superscript indicate non-significant differences.
Pollen taxa (T1/T2)
ANOVA test F-Ratio p Comparison
Cyperaceae 4.17 * C1I1M2
Illecebrum/Paronychia 5.3 * C1I1M2
Isoetes histrix 17.29 *** C1I1M2
Isoetes velata-type 6.83 ** C1I1M2
Myriophyllum alterniflorum 42.84 *** C1I2M3
Ranunculus-type 30 *** C1I2M3
Kruskal–Wallis test Chi2 p Comparison
Alisma-type 3.89 ns C1I1M1
Callitriche 1.57 ns C1I1M1
Cicendia/Exaculum 1.94 ns C1I1M1
Elatine 3.12 ns C1I1M1
Juncus 0.75 ns C1I1M1
Lythrum borysthenicum-type 7.54 ** C1I2M2
Pilularia 0.12 ns C1I1M1
Table 5
Representation of hydrophytic taxa in the pollen record, based on the calculation of Davis indices (Davis, 1984): associated (A), under-represented (U) and over- represented (O). The two right columns present the coefficient of determination (r2) and thep-value of the tests of linear correlations between pollen concentrations and vegetation (n = 21 quadrats). ns = not significant. **=pb0.005; ***=pb0.001.
Taxons A U O r2 p
Strongly-associated types
Cyperaceae 80 20 0 0.06 ns
Isoetes velata-type 81 0 19 0.47 ***
Myriophyllum alterniflorum 75 8 20 0.56 ***
Ranunculus-type 86 54 0 0.40 **
Associated types
Illecebrum/Paronychia 50 17 44 b0.01 ns
Weakly associated types
Alisma-type 27 60 56 b0.01 ns
Elatine 18 78 50 0.01 ns
Isoetes histrix 11 33 89 0.01 ns
Juncus 21 60 69 0.09 ns
Lythrum borysthenicum-type 20 67 67 0.08 ns
Pilularia 36 20 60 0.16 0.06
Unassociated types
Callitriche 6 94 50 b0.01 ns
Cicendia/Exaculum 0 100 100 0.02 ns
alterniflorum,Ranunculus-type), (2) associated taxa (Illecebrum/Parony- chia), (3) weakly associated taxa including the rare species (Elatine, Isoetes histrix,Pilularia), and (4) unassociated taxa (Callitriche,Cicendia/
Exaculum).
These results are consistent with the linear correlation tests (Table 5): the hydrophytic taxa with highr2values are characterised by good correspondences between pollen and vegetation. They moreover point to Cyperaceae andIllecebrum/Paronychiaas strong- ly-associated and associated taxa, respectively, despite their lowr2 values (respectively 0.06 and 0.006;pN0.05).
3.5. Correction factors (R-values)
Table 6presents correction factors (R-values) and estimates of pollen dispersion (y-intercept) for the 5 strongly-associated and associated taxa. These results are consistent with the linear regres- sions (Table 5) that show good correspondences between pollen and vegetation for three hydrophytes (Isoetes velata-type,Myriophyllum alterniflorum,Ranunculus-type). It should be noted thatIsoetes velata- type present significant correlation only for absolute data (Table 6).
Because of its high spore productivity and high percentages, the Fagerlind effect is certainly important for this taxon, and it would be necessary to use the extended R-value (ERV) model (Parsons and Prentice, 1981; Prentice and Parsons, 1983; Sugita, 1994) to analyse the relationship for percentages. Unfortunately, the necessary para- meters have not yet been measured.
4. Discussion
4.1. Composition of vegetation and pollen assemblages at the local scale Former investigations carried out on the pool D4 from 1997 to 2009 (Rhazi et al., 2001, 2007, 2009) reveal its highfloristic richness, characterised by 102 species, among which six are rare and threatened in Morocco:Elatine brochonii,Exaculum pusillum,Pilularia minuta,Lythrum thymifolia,Isoetes velata, andMyriophyllum alterni- florum(Fennane and Ibn Tattou, 1998).Isoetes velataandMyriophyl- lum alterniflorum are abundant within the pool, while Elatine brochonii, Exaculum pusillumand Pilularia minuta are restricted to small peripheral patches (Table 1). Some of these species typical of temporary pools are sporadic and require particular hydrological conditions for recruitment, development and reproduction. The studied pool also harbours terrestrial species (31 taxa including Asteroideae, Cichorioideae and Poaceae), the abundance of which varies between years depending on hydrology (Rhazi et al., 2009).
During dry seasons, such terrestrial taxa originating from the surrounding cork-oak forestfind favourable hydrological conditions at the margin (Table 1) of the pool which is irregularlyflooded. The
alternation betweenflooded and dry phases allows coexistence of different communities with specific spatial patterns of distribution and thus creates an intra-pool mosaic (Grillas et al., 2004; Rhazi et al., 2007).
The representation of vegetation by pollen assemblages is likely to be affected by several factors, including site size, structure of the surrounding vegetation, capacities of pollen production and dispersal, resistance of pollen grains to degradation, disturbances by domestic and wild herbivores, and the accuracy of identification during pollen analysis (Janssen, 1970; Jacobson and Bradshaw, 1981; Muller et al., 2006). These factors are likely to limit the interpretation of the pollen dataset, in particular in terms of biodiversity (Odgaard, 1999; Brun et al., 2007). The obtained pollen data are clearly affected by pollen degradation (e.g. weak record ofJuncus), and taxonomical level of identification (e.g. Poaceae, Cyperaceae, Asteroideae, and Cichorioi- deae). Moreover, the small size of pool D4certainly explains the better record of hydrophytic (15 taxa for 26 species) than terrestrial taxa (30 taxa for 74 species). Hydrophytic taxa include rare and sporadic taxa (Cicendia/Exaculum, Elatine, Pilularia), which are infrequently recorded in pollen assemblages in the studied pool. Thefinding of their pollen in fossil sediments appears uncertain, and hinders the study of their past local dynamics.
Along both transects, the correlation between taxa frequencies in vegetation and surface pollen (r2= 0.22;Table 1) partly reflects the weak pollen dispersal, which does not appear to be affected by the disturbances generated by domestic and wild herbivores. It is noteworthy that local and extra-local pollen representca. 70% of the total pollen input within the pool, despite temporary pools represent only 2% of the whole surface area of Benslimane province (Fig. 1;
Rhazi et al., 2001). Such a feature confirms theoretical models predicting that small sites records predominantly pollen of local origin (e.g.Jacobson and Bradshaw, 1981; Davis, 2000).
4.2. Structure of vegetation and pollen assemblages at the local scale The vegetation relevés of the studied pool (Figs. 3 and 4) confirm the differential species composition of the vegetation within each transect along the topographical gradient typical of the temporary pools of Benslimane province (Rhazi et al., 2001, 2006). The zonation is mainly controlled by hydrology (depth and submersion duration), the principal factor controlling the vegetation structure of temporary pools (e.g.
Bauder, 2000; Grillas et al., 2004; Deil, 2005). Palynological data reveal that the three zones are well recorded by pollen assemblages (Figs. 5 and 6):
The marginal zone. Wind-pollinated and insect-pollinated terres- trial taxa (e.g. Asteraceae,Plantago coronopus-type,RumexandUrtica) are represented by low percentages at the margin of the pool. Most of these taxa characterise grazed areas with soils compacted by herbivores (Behre, 1981; Court-Picon et al., 2005). Their low abundance in pollen assemblages reflects their irregular occurrences, both at the margin of the pool and within surrounding forests. The low percentages ofJuncuscan be attributed to the difficulty of identifica- tion and to the sensitivity to degradation of its pollen (Fægri and Iversen, 1989). The low percentages of the rare taxa Elatine and Pilularia could be attributed, first to their scarcity and sporadic development (Rhazi et al., 2001).
The intermediate zone. Cyperaceae,Illecebrum/ParonychiaandIsoetes velata-type pollen are recorded from the intermediate zone, where they reach their maximal values, and in the centre of the pool. This pattern corresponds well to their distribution in vegetation.Alisma-type pollen occurs essentially within the intermediate zone, which reflects the occurrence there of a few individuals ofBaldellia ranunculoides. This taxon has developed several adaptations to hydrological constraints, such as preferential allocation of resources to vegetative reproduction (Sculthorpe, 1967; Vuille, 1987) likely to explain its restricted distribution. Finally, Asteroideae taxa present similar patterns of Table 6
Linear regression model for‘taxa’approach, presenting the coefficient of determination (r2), thep-value, the slope of the regression line (Ri) and the y-intercept (Di) for the best associated taxa in pollen and vegetation along transects T1and T2. The regression was performed on absolute frequencies (pollen influx, in grains mm−2year−1; vegetation abundance, in occurrences m−2year−1) and relative frequencies (pollen percentage; vegetation percentage), respectively for each quadrat (n = 21 quadrats).
ns = not significant; *=pb0.05; **=pb0.005.
Taxa Absolute frequencies Relative frequencies
r2 p Rai Dai r2 p Rri Dri
Cyperaceae 0.16 ns 0.26 13.12 0.05 ns −0.005 0.26
Illecebrum/
Paronychia
0.08 ns −2.41 88.45 b0.01 ns 0.004 0.19 Isoetes velata-type 0.20 ⁎ 592.76 5261.10 b0.01 ns 0.300 73.02 Myriophyllum
alterniflorum
0.39 ⁎⁎ 37.33 114.70 0.29 ⁎ 0.313 0.43
Ranunculus-type 0.30 ⁎ 26.05 −370.99 0.13 ⁎ 0.160 0.24
