Coastal forest and Miombo woodland history of the Vilankulo region,Mozambique

Abstract

The present day distribution of Miombo savanna-woodland in Mozambique has been attributed to an expansion due to the clearing of original coastal forests through agriculture and use of fire. Here, we test this hypothesis using palaeoecological data from Lake Nhauhache, situated in the Vilankulo region. Our analysis shows that Brachystegia, one of the main constituents of the Miombo, has varied over time, and its variability seems to be driven by hydrological changes related to climatic variability rather than by land-use changes. The analyses show that Brachystegia was most common during ad 200–700 when a marshy forest/shrub community was dominant. After ad 700, this community changes to a dominance of Syzygium and Fagara linked to gradually rising water levels. Brachystegia remains in low abundance and fluctuating over time. From ad 1000, a general decline in trees/shrubs in favour of grasses concurs with an increase in grass pollen (possibly cereal) and charcoal, most probably as a result of farming activities. The decline in tree taxa was probably exacerbated by periodic droughts after c. ad 1200 as indicated by the diatom assemblage. In the period ad 1700 to late 1800, arboreal pollen is well represented, and this is concurrent with the diatom record suggesting high lake levels.

Keywords

land-use history late Holocene Miombo Mozambique palaeoecology savanna ecology vegetation history

Introduction

The coastal lowlands of southern Mozambique and the Vilankulo area are commonly reported as a forest or forest transition region (White, 1983) or as a Miombo savanna-woodland with a coastal ticket or forest along the sea stretch (Wild and Grandvaux Barbosa, 1967). The forest or forest transition region, the Indian Ocean coastal belt, is described as a mosaic of semi-deciduous forest, thickets, woodland, savannas and edaphic grasslands (Moll and White, 1978; Werger, 1978). Despite representation in vegetation maps, islands of semi-deciduous forests, are, however, only present in the Inhambane region and by the Save River (Wild and Fernandes, 1967). Miombo is a colloquial term of savanna-woodland dominated by the genera Brachystegia, Julbernardia and/or Isoberlinia (Frost, 1996). In the phytogeographies of Werger (1978), the Miombo savanna-woodland was suggested to be invasive on the coast with its original centre of endemism in the south-central Africa (hence the name Zambesian phytochorion). The expansion of the Miombo savanna-woodland has been attributed to the use of fire and farming in the coastal regions. In previous discussions on landscape dynamics (e.g. Hall, 1981; Morais, 1988), it has been assumed that the coastal regions were forested prior to the arrival of farmers in southern Africa c. ad 300. Presently, the seaward-facing coastal slopes in the Vilankulo region carry coastal thickets, but the landscape is dominated by Julbernardia globiflora and Strychnos spinosa savannas, but Brachystegia spiciformis is actually very rare (Massinga in Ekblom, 2004; Telford and De Castro, 2001, see section ‘Discussion’). As in the case of the absence of coastal forests, the absence of B. spiciformis in the Vilankulo region has also been attributed to intensive farming and use of fire and swidden agriculture (Telford and De Castro, 2001); however, none of these hypotheses have been tested using historic data or palaeoecology. This paper sets out to test this hypotheses using historical sources or palaeoecology. The aim of this paper is therefore to explore the dynamics between forests and Miombo savanna-woodlands in the Vilankulo region.

Earlier palaeoenvironmental investigations in the Vilankulo region have shown that the coastal area in ad 400 consisted of a mosaic of forests, B. spiciformis savanna-woodlands, challenging the idea of the Miombo as an invasive type of vegetation (Ekblom, 2008; Ekblom et al., 2013). B. spiciformis is underrepresented in pollen diagrams, as it is pollinated by insects. Despite this, it is relatively well represented in the pollen diagrams from southern Africa (Scott, 1982). Meanwhile, Julbernardia is rarely reported in pollen diagrams. B. spiciformis will thus be used here as a sole indicator of Miombo, even though it should include also other species.

Here, we present a new vegetation record from the Vilankulo region to explore the Miombo/forest dynamics. The analysis is based on stratigraphic occurrences of pollen, spores and microfossil charcoal from Lake Nhauhache and combined with the diatom assemblage presented previously (Holmgren et al., 2012). The results are discussed in relation to changes in land use, and in regional vegetation and hydrological changes, with a specific focus on the long-term ecology of the Miombo.

There are several processes that can be responsible for the hypothesised transition from forests to a Miombo savanna-woodland. One process is land-use change and swidden agriculture, whereby farmers converted coastal forests into agricultural lands and fallow periods were too short to allow forests to return. Another process that would have promoted Miombo constituents on the cost of forest species is a possible shift in fire regimes (e.g. seasonality, intensity and frequency). As the Miombo constituents are well adjusted to the influence of fire, they sprout from the root, and Miombo woodlands in the African interior have also been reported to burn approximately in four cycles (Chidumayo, 1997). These two processes are likely to be interlinked as forest fires usually tend to spread mainly on the forests floor, for example, not affecting trees once they are matured. A third possibility that will be tested here is that the presence of Miombo constituents can be explained as a long-term constituent of the coastal landscape and as unrelated or weakly related to land use. Indirect evidence of farming communities in southern Mozambique is recorded from ad 300 (see below), but the earliest material evidence of farmers in the Vilankulo region only dates from ad 600. The record presented here dates from ad 200 and will be used to reconstruct the landscape prior to the onset of farming. We also set out to discuss the long-term dynamics of the Miombo/forest ecology and possible causes when it comes to ecological transformations.

Vilankulo region and research background

The Vilankulo region is marked by the presence of several lake systems, most of which are closed lakes with a characteristic circular or semi-circular shape. The Pleistocene dune system of the region forms an aquifer independent from the interior that is fed mainly by rainfall (Coetsee and Hartley, 2001). Vilankulo receives most of its rainfall (832 mm/yr) between November and March, when temperatures are the highest. The region experiences a c. 20-year rainfall cyclicity (Tyson and Preston-Whyte, 2000), in addition to high inter-annual rainfall variability. The dune system of the coastal plain is characterised by weakly developed soils with a low organic content and moderate weathering (Fränzle, 1984; Wood, 2001). These soils support a savanna and woodland savanna where J. globiflora and S. spinosa are well represented, interspersed with grasses. B. spiciformis is rarely found in the Vilankulo region (Massinga in Ekblom, 2004; Telford and De Castro, 2001). In some parts, the coastal dunes support a dense shrub characterised by Commiphora zanzibarica, Phyllanthus reticulatus and Turrea nilotica. These occur with Grewia monticola, Deinbollia oblongifolia, Clerodendrum glabrum and Acalypha glabrata and also climbers such as Cocculus hirta (Massinga in Ekblom, 2004). Coastal forests, constituted by species such as Chlorophora excelsa, Ficus spp., Morus mesozygia, Celtis africana, Afzelia quanzensis, Dialium schlechteri and Brachyleana discolor, are absent in the Vilankulo region today (Ekblom, 2004; Wild and Fernandes, 1968). However, patches of closed canopies sometimes occur and are characterised by species, such as Millettia stuhlmanni, Alchornea laxiflora, Euclea natalensis and A. quanzensis. Along streams, species such as Ficus trichopoda and Myrica cf. pilulifera are recorded together with J. globiflora and, occasionally, B. spiciformis (Telford and De Castro, 2001).

The earliest date of farming communities in Mozambique comes from south of Maputo, where early farming community–style pottery (Matola) was found in layers dated to ad 300. Hunter and gatherer communities were active in the area at this time and probably used fire to renew the grass and provide good grazing for game, although there is no direct evidence of this. Hunter and gatherer groups producing late Stone Age lithics are attested to the Bazaruto Archipelago, northeast of Vilankulo town. The Vilankulo area is not well surveyed archaeologically, with the exception of Chibuene, situated 7 km south of Vilankulo. From ad 600–900, this was an important regional hub of trade (Sinclair et al., 2012). An earlier period of settlement before ad 600 is suggested by the presence of Matola-type pottery, but the presence of ceramic producing groups and, possibly, farmers in the early first millennium is yet to be confirmed. The finds of early domesticates, for example, bones and cereals, remain sparse (Badenhorst et al., 2011; Ekblom et al., 2013; Sinclair, 1987).

Palaeoecological investigations have been carried out in Lake Nhaucati and Lake Xiroche in the Chibuene area (Ekblom, 2008). These records span from ad 400 and 600 to the present, respectively. They are both small lakes fed by the same aquifer as Lake Nhauhache. The pollen records of Chibuene indicated a mosaic of forests, Miombo savanna-woodlands and grasslands at ad 400, but with extensive forests constituted by species such as Moraceae, Celtis and Trema. A progressive decrease of forests occurred after ad 1400, which was tentatively linked to droughts associated with the cold and dry periods sometimes referred to as the ‘Little Ice Age’ (ad 1400/1600–1800, Ekblom, 2008; Holmgren et al., 2012). The original intention of Lake Nhauhache investigation presented here was to investigate whether the same pattern as in Chibuene could also be found here; however, the research question was reformulated in course of this study to focus on the Miombo (see section ‘Discussion’). The pollen records in Chibuene also suggested phases of cultivation decline and expansion. An expansion of cultivation and cattle herding took place after ad 1000 followed by a decline after ad 1400 (e.g. contemporary with the decline of forests in the area). An additional expansion of farming might have occurred in the 20th century as indicated by a decline in Brachystegia (Ekblom et al., 2013). There are few other studies of present-to-historic vegetation dynamics available from the coastal lowlands that can provide a comparison to this Miombo/forest controversy. The closest analogue to this study is the work carried out in the coastal KwaZulu-Natal region in Lake Eteza (Neumann et al., 2010; Scott and Steenkamp, 1996) and in Lake Sibaya (Neumann et al., 2008). The Lake Sibaya record does not indicate any marked human influence on the vegetation before the last 200 years (Neumann et al., 2008), while in Lake Eteza marked changes are indicated from ad 700, but whether this is due to human influence, climate variability or the combination is inconclusive (Neumann et al., 2010).

Lake Nhauhache

Lake Nhauhache is situated at 21°58′50″S, 35°17′39″E (at 6 m a.s.l.), about 3 km northwest of Vilankulo town (Figure 1). Nhauhache is uppermost in a system of smaller coastal lakes linked to the Inhamalene River, which feeds into the Indian Ocean. The lake varies in size depending on regional rainfall. In July 2008, at the time of coring, it measured 150 m in diameter and was 2.2 m deep. The lake is characteristic of other lakes in this region, circular with steep surrounding slopes. In the southeast, the flatter topography allows for drainage of the lake during high lake levels.

Figure 1.

Location of the Vilankulo region; (inset) Vilankulo town, Lake Nhauhache and other investigated lakes in the vicinity.

The surroundings of the lake are today densely populated and cultivated, and the biological production within the lake is high. Hydrophilous plants, such as Nymphaea lotus, Hyparrhenia sp. and Phragmites sp., are found in the limnic and telmatic zones. Local residents reported that the lake was dry from ad 1990 to 2000 but retained water during the prolonged droughts in the 1980s when other lakes in the area were reported to be almost, or completely, dry (Holmgren et al., 2012). The diatom assemblages of the Lake Nhauhache record, previously presented (Holmgren et al., 2012) and also used here, were inferred to indicate wetter conditions from c. 300 bc to ad 800, fluctuating conditions at ad 800–1150 and dry conditions at ad 1150–1700, followed by more humid conditions until the present.

Methodology

Lake Nhauhache was sampled in July 2008 using a dinghy and a Russian coring device. The bathymetry was examined using a handheld echo-sounder, and a sediment core was subsequently collected in the deepest part of the lake, at c. 2.2 m water depth. Lacustrine sediment with a thickness of 2.3 m was cored before it reached a layer of pure dune sand. All sediment depths in this paper are referred to in terms of centimetre from the bottom of the lake. The lithology of the lake sediment is homogeneous and composed of reddish to grey sandy gyttja. In the lower portion of the core, nodules of iron are observed, and a 1-cm-thick layer of rootlets appears at 355 cm depth. There are no clear signs of erosion horizons or palaeosoil surfaces. As the lake is small, we expect it to reflect local conditions of vegetation change (Jacobsen and Bradshaw, 1981; Prentice, 1985).

Pollen

Pollen and microscopic charcoal were extracted through digestion of cellulose with NaOH and concentration of pollen through acetolysis (Faegri and Iversen, 1992; Moore et al., 1991). A known amount of exotic Lycopodium marker spores to the sediment to calculate total concentration of pollen grains per square centimetre of sediment. Every 8 cm (or more) was sampled in the upper part of the core. As the original focus of this investigation was on the period ad 1200–1800, the lower part was sampled at larger intervals (4–45 cm spaced, depending on observed changes in the record). Glycerol was used as an embedding medium, and samples were analysed in 250× to 1000× magnification. Counts were made until the sum of trees, shrubs, herbs and grass amounted between 450 and 500 grains. Sedges (Cyperaceae) were not included in the pollen sum, although they may include both dry grassland species and limnic communities. Aquatics and eroded grains were also excluded from the pollen sum. Pollen identification was carried out using the African Pollen Database (http://apd.sedoo.fr/apd/accueil.htm) together with photographs from the reference collections available at the Palynology Laboratory (Department of Plant Sciences, University of the Free State, Bloemfontein) and from Oxford Long-term Ecology Laboratory (Oxford University) together with published references (Bonnefille and Riollet, 1980; El Ghazali, 1993; Scott, 1982; Van Zinderen-Bakker, 1953, 1956; Van Zinderen-Bakker and Coetzee, 1959). Ecological interpretations are based on Coates Palgrave (2002). Pollen diagrams were plotted using Tilia and Tilia Graph (Grimm, 1991).

The pollen of indigenous African cereal grasses (Pennisetum americanum, Sorghum bicolor and Eleusine coracana) overlap in size and are morphologically similar to wild grass pollen (Tomlinson, 1973). It has been noted in pollen analyses from the Mozambique interior (Ekblom and Gillson, 2010) that larger grass pollen (e.g. <40 µm) tend to occur in greater numbers together with maize pollen; thus, it is likely that the grass pollen >40 µm represent indigenous cereal grasses. We have therefore used a tentative size separation of grasses of <40 µm here. Glycerol tends to swell pollen grains to some extent (Cushing, 1961). There may therefore be an overrepresentation of grains in the <40 µm size class due to the effect of glycerol. We have only included pollen grains that are decisively >40 µm in size, and we did not note any significant changes in sizes in other taxa.

Charcoal

The area of charcoal, that is, black opaque angular particles, was estimated in relation to sediment volume (cm²/cm³) and added Lycopodium spores, using the point count method, with counting minimum of in total of 50 items (Clark, 1982, 1988). This gives a rough estimate of the amount of charcoal in the sample, although a larger count, of 200 Lycopodium/charcoal, would be ideal for higher statistical reliability (Finsinger and Tinner, 2005). We also separated two main classes of charcoal: 10–100 µm interpreted as linked to local–regional fires (here referred to as ‘regional charcoal’) and above 100 µm (here referred to as ‘local charcoal’) interpreted as linked to local fires (see review in Pitkänen et al., 1999).

Spores

All spores were counted in relation to the added Lycopodium spores and presented as concentrations. Most types were identified or grouped together into main type groups (Graf and Chmura, 2006; Jarzen and Elsik, 1986; Prager et al., 2006; Van Geel and Aptroot, 2006; Van Geel et al., 1983, 1986, 2003). Here, we will focus on the coprophilous fungi, mainly Sordariaceae, but also Coniochaeta lignaria (Marinova and Atanassova, 2006; Van Geel and Aptroot, 2006; Van Geel et al., 2003). Algal spores from Spirogyra spores and Glomus (only found in very low numbers and not shown here) were also distinguished. Botryococcus, most probably Botryococcus braunii, occurred in large numbers throughout the core but was not quantified as they occur as mass material.

Diatoms

The diatom assemblage has been presented in detail elsewhere (Holmgren et al., 2012). Only the ecological groups are presented here, that is, planktonic, benthic, halophilous, aerophilous and ungrouped taxa. Every 3–4 cm intervals were sampled for diatoms, and the samples were oxidised with 17% H₂O₂. Clay particles were removed by decanting from 100 mL beakers in 2-h intervals (Battarbee, 1986). The remaining samples were mounted on glass slides using Naphrax and studied under a light microscope using oil immersion. Approximately 500 diatom frustules were counted per level using standard references (Cleve-Euler, 1953, 1955; Gasse, 1986; Krammer and Lange-Bertalot, 1986, 1988, 1991a, 1991b; Tynni, 1979).

Dating

Four samples were accelerator mass spectroscopy (AMS) ¹⁴C dated at the Radiocarbon Laboratory in Poznan, Poland, to complement the 10 samples presented by Holmgren et al. (2012; Table 1). Due to the lack of macrofossils in the sediments, bulk sediments (1 cm thickness each) were used for dating. OxCal 4.0 (Bronk Ramsey, 1995) was used for calibration to calendar scale with the calibration curve for the southern hemisphere (McCormac et al., 2004; Figure 2). The new ¹⁴C dates strongly support the previously proposed age model (Holmgren et al., 2012). The record spans the period from c. 2300 bc to ad 1800, but here, we have focused on the time period from ad 200 to 1800, as our original focus was to investigate vegetation change in this period (see section ‘Discussion’). Sediment accumulation is relatively slow in the lower part of the sequence (c. 0.05 cm/yr) but increases between ad 500 and 1000 (c. 0.2 cm/yr). From ad 1000 sediment, accumulation again slows down (c. 0.1 cm/yr).

Table 1.

Calibration dates for Lake Nhauhache. Probability at 1σ and 2σ. Calibration curve for the southern hemisphere used (McCormac et al., 2004). New dates are marked in bold.

Depth from lake surface/sample name	Depth from bottom (cm)	Laboratory number	Age ¹⁴C yr BP	Cal. 68.2% (ranges ad/bc)	Cal. 95.4% (ranges ad/bc)	Material
N1°230	10–11	Poz-27386	180 ± 30	ad 1674 (19.5%) 1710	ad 1668 (44.7%) 1785	Sandy gyttja
				ad 1720 (10.3%) 1740	ad 1794 (9.6%) 1818
				ad 1798 (7.4%) 1812	ad 1827 (25.3%) 1894
				ad 1837 (5.6%) 1849	ad 1909 (15.9%) 1954
				ad 1854 (12.2%) 1880
				ad 1924 (13.2%) 1952
N1°240–241	20–21	Poz-50380	135 ± 30	ad 1700 (12.1%)1722	ad 1687 (19.0%) 1729	Sandy gyttja
				ad 1810 (16.2%) 1838	ad 1804 (76.4%) 1952
				ad 1847 (11.6%) 1867
				ad 1879 (28.3%) 1928
N1°261	41–42	Poz-27387	415 ± 30	ad 1456 (48.3%) 1504	ad 1448 (56.7%) 1515	Sandy gyttja
				ad 1591 (19.9%) 1615	ad 1540 (38.7%) 1625
N1°264–265	44–45	Poz-50381	460 ± 30	ad 1441 (68.2%) 1483	ad 1429 (88.0%) 1503	Sandy gyttja
					ad 1593 (7.4%) 1614
N1°283	63	Poz-27389	750 ± 30	ad 1274 (56.1%) 1302	ad 1231 (2.9%) 1247	Sandy gyttja
				ad 1365 (12.1%) 1365	ad 1263 (68.8%) 1320
					ad 1351 (23.7%) 1385
N1°283 shell	63	Poz-27447	845 ± 35	ad 1216 (68.2%) 1269	ad 1180 (95.4%) 1280	Fossil shell
N1°303–304	83–84	Poz-50383	970 ± 30	ad 1045 (33.2%) 1087	ad 1032 (95.4%) 1181	Sandy gyttja
				ad 1106 (35.0%) 1157
N1°313	93	Poz-27408	1020 ± 30	ad 1025 (22.6%) 1047	ad 1016 (95.4%) 1152	Sandy gyttja
				ad 1085 (45.6%) 1133
N1°356	136	Poz-27409	1370 ± 30	ad 681 (50.7%) 708	ad 654 (95.4%) 772	Sandy gyttja
				ad 748 (17.5%) 766
N1°380	160	Poz-27390	1800 ± 30	ad 245 (68.2%) 338	ad 215 (95.4%) 402	Sandy gyttja
N1°403	183	Poz-27391	1565 ± 30	ad 535 (68.2%) 607	ad 434 (17.1%) 492	Sandy gyttja
					ad 508 (2.1%) 519
					ad 527 (76.3%) 635
N1°411–412	191–192	Poz-50384	1755 ± 30	ad 261 (12.6%) 280	ad 255 (95.%) 415	Sandy gyttja
				ad 325 (55.6%) 399
N1°425	205	Poz-27392	1775 ± 30	ad 258 (28.6%) 299	ad 243 (95.4%) 404	Sandy gyttja
				ad 319 (27.7%) 359
				ad 364 (12 %) 382
N1°447	227	Poz-27393	2285 ± 30	375 (13.9%) 351 bc	387 (95.4%) 203 bc	Sandy gyttja
				301 (54.3%) 201 bc

Figure 2.

Age–depth diagram with the range of dates published in Holmgren et al. (2012, grey) and additional dates (black). The probability distribution (2σ) of each date is shown in the respective graph. The pollen and diatom zones are shown to the right.

Results

The pollen assemblage has been divided into three zones through cluster analysis. The analysis is based on unweighted paired group average and Euclidean similarity index using the software PAST (Hammer et al., 2001). This method is unconstrained by depth and more suitable for cores with possible temporal gaps (Figure 3). The most commonly represented species is presented here (Figures 4 and 5). The tree/shrub group has been divided into two main ecological groupings to assist interpretation, a riverine forest/forest group and a savanna/generalist group (Table 2).

Figure 3.

Cluster analyses of analysed levels, the analysis that is unconstrained by depth is based on unweighted paired group average and Euclidean similarity index using the software PAST (Hammer et al., 2001).

Figure 4.

Pollen diagram, with terrestrial pollen percentages grouped into arboreal types and savanna and generalist types. The lithology consists of sandy gyttja. Herbs and aquatics are also shown. The pollen types with a low representation have been exaggerated (hatched silhouette) with a factor of 5.

Figure 5.

Summary pollen diagrams with the main ecological groups shown in percentages. The lithology consists of sandy gyttja. Distribution of spores (number/cm²), regional and local charcoal (number/cm²) and total pollen and spore concentrations (left). The pollen diagram has been divided into three zones based on the pollen assemblage. A summary diatom diagram is shown to the right with the four zones based on the diatom assemblage indicated for comparison. The sparse representation of charcoal in the lower part of the diagram is due to the low sampling resolution here.

Table 2.

Represented pollen taxa and ecological groupings in the Lake Nhauhache sediment core.

Pollen included in the terrestrial pollen sum
Trees and shrubs associated with forest/riverine forests
Alchornea	Eugenia	Myrica
Acalypha	Meliaceae	Olea
Celtis	Millettia-type	Syzygium
Dialium-type	Mimusops	Trema
Diospyros	Moraceae	Myrsine
		Stereospermum-type
Generalist trees and bush
Acacia	Combretaceae	Julbernardia
Brachystegia	Cordia	Rhamnaceae
Canthium-type	Dichrostachys	Rhus-type
Capparaceae	Dodonaea	Sclerocarya
Cassia-type	Euclea	Sterculia
Cassine-type	Euphorbia-type	Spirostachys
Commiphora-type	Fagara-type	Uapaca
Celastraceae	Grewia	Vitaceae
Herbs
Acanthaceae	Liliaceae	Tribulus
Asteraceae	Heteromorpha-type	Poaceae
Boraginaceae	Lamiaceae	Polygonum-type
Chenopodiaceae (Chen/Am)	Malvaceae	Ranunculaceae
Apiaceae-type	Oxygonum	Solanaceae
Commelina
Ericaceae
Ungrouped types (included in the terrestrial pollen sum)
cf. Vitex	cf. Caesaria spp.	Borassus/Hyphaene
cf. Meliaceae	cf. Peltophorum	Podocarpus
cf. Scrophulariaceae	Borreria-type	Unknown
Pollen excluded from the terrestrial pollen sum
Limnic plants
Cyperaceae	Borassus (note this is included in pollen sum)
Herbaceous vegetation associated with damp and marshy vegetation
Apiaceae-type	Heliotropium-type	Onagraceae
Commelina	Heteromorpha-type	Polygonum-type
Dissotis-type	Ipomoea-type	Ranunculaceae
Ericaceae	Liliaceae	Solanaceae
Gentianaceae	Menthe-type
Submerged or partially submerged aquatic plants
Equisetaceae	Nymphaea	Typha
Hydrocharitaceae	Potamogeton	Utricularia
Laurembergia

Nhauhache 1: 210–135 cm depth (ad 200–700)

Total pollen concentrations are generally low in this zone. Trees and shrubs are well represented, particularly Stereospermum, Combretaceae, Myrsine Africana, Alchornea (well represented throughout the core), Moraceae and Trema, are here combined into one curve as their representation is low, and they have similar ecological requirements. The identification of Stereospermum remains provisional (see Table 2). Brachystegia shows its highest values in this zone compared with other parts of the core. There are high values of Nymphaea throughout the zone together with Laurembergia and Hydrocharitaceae. Poaceae and Cyperaceae show moderate levels. The herbs (Chenopodiaceae, Asteraceae and Acanthaceae) are relatively well represented. Charcoal particles and Poaceae >40 µm are few in this zone. Spores are represented in low numbers and represented mainly by fern spores.

Nhauhache 2: 135–42.5 cm depth (ad 700–1500)

This zone could possibly be divided into two subzones 135–90 cm and 90–42.5 cm (ad 700–1100 and ad 1100–1500) on the basis of a decline of trees/shrubs in the upper part of the zone. As there is only one pollen level in the lower part, they are not statistically separated. Total pollen concentrations increase in the whole zone as does the number of degraded grains. Trees and shrubs are still well represented but decrease in relation to previous zone and particularly in the upper part of the zone from 90 cm depth (ad 1100). M. africana declines markedly together with Moraceae/Trema, Stereospermum and Combretaceae. Brachystegia is represented in low numbers throughout the zone. Myrtaceae, mainly represented by Syzygium with some contribution of Eugenia, is well represented. Phyllanthus is more common in this zone than before. Alchornea shows a small increase in the upper part. Poaceae increases while the Cyperaceae curve remains stable. Nymphaea decreases progressively to low numbers in the upper part of the zone. The herbs decrease slightly, but Asteraceae is better represented than in previous zone. Poaceae >40 µm remains low in numbers until 90 cm depth (c. ad 1100), when they increase significantly together with both local and regional charcoal particles. Spore numbers continue to be low, but there is a slight increase in spores associated with coprophilous fungi (C. lignaria) from 90 cm depth. Fern spores decline from the beginning of the zone but increases together with trilete fern/moss spores.

Nhauhache 3: 42.5–10 cm depth (ad 1500–1800)

The uppermost zone is of short duration and is distinguished by an increase in trees/shrubs and in Cyperaceae and the aquatics Nymphaea and most markedly of Typha. Among the trees/shrubs, Alchornea remains well represented together with Myrtaceae and Phyllanthus. Stereospermum and Brachystegia occur only in low numbers, and the latter is absent in the upper part of the zone. Dodonaea is common (as also in the upper part of the previous zone). The herbs show a significant increase, particularly the Lamiaceae group. Fern and fern/moss spores occur in moderate numbers, and there is an overall increase in spores, particularly spores associated with coprophilous fungi C. lignaria and Sordariaceae that are very common in the uppermost part of the core. Local and regionally transported charcoal remains well represented together with grasses >40 µm. Pollen concentrations in this zone are higher than in previous zone, and the amount of degraded grains remains high.

Discussion

Summarising the changes in vegetation communities

The results of the Lake Nhauhache pollen analysis show that Brachystegia was most common in the beginning of the sequence. At ad 200, a vegetation community with Brachystegia, Stereospermum, M. africana and Combretaceae was present near Lake Nhauhache. These occurred with a few representatives of typical forest vegetation. The forest taxa (Moraceae, Trema, Celtis) that represented 20–40% of the pollen sum nearby Chibuene (Ekblom, 2008; Ekblom et al., 2013) are not well represented in Lake Nhauhache (forest taxa here represent only c. 2% of the pollen sum). Thus, the pollen diagram from Lake Nhauhache does not indicate the presence of a large forest community at ad 200 or later in the pollen diagram, which is also why we rephrased our research question in course of this study. The Stereospermum, M. africana and Combretaceae community was dominant until around c. ad 700 (135–136 cm depth), probably as a marshy forest/shrub community with Brachystegia common in the well-drained areas. The palm tree Borassus also occurs sporadically during this phase, and this is normally associated with temporarily flooded areas. The diatom assemblage (Holmgren et al., 2012) indicates a marshy lake environment with variably high values of aerophilous taxa, suggesting, perhaps, that the area was only temporarily waterlogged. High values of Nymphaea parallels the results from Chibuene, which also displays a peak in Nymphaea at the beginning of the lake phase when areas high in organic matter were flooded by rising water tables. Spirogyra is common in this period, and it usually grows as macroalgal mats in shallow standing waters (Joska and Bolton, 1996). The abundance of Botryococcus suggests oligotrophic conditions at the base of the core as well as in the upper part.

In the period ad 700–1000 (135–136 cm depth), there is a change in species composition with Syzygium and Fagara being more dominant. This is most likely linked to gradually rising water levels and exacerbated conditions for a marshy forest community. The diatom assemblage continues to display high values of aerophilous taxa until 120 cm depth (c. ad 840) when there is a peak in planktonic taxa at 125 cm depth (Figure 5), which has been interpreted as a flooding event (Holmgren et al., 2012). Higher lake levels in this period are also supported by high pollen concentrations and the fast accumulation of sediment. Thus, the change in species representation is probably best explained by a change in hydrology and/or an increase in rainfall. The high lake levels are likely to have drenched the swamp forest community and other species were favoured on the steeper slopes around the lake. Brachystegia is still represented with 2–4% after ad 700, but other taxa, as M. africana and Stereospermum, only occur sporadically after ad 700. The transition is somewhat obscure, however, due to a low sample resolution in this part, something which can hopefully be amended in the future.

From ad 1000 (90 cm depth), there is a general decline of trees/shrubs in favour of grasses. This is contemporary with evidence of an agricultural expansion (see below). Pollen concentrations in this period are moderate. The decline of trees/shrubs in this phase may initially be linked with the agricultural expansion, later exacerbated by periodic droughts. The diatom assemblage indicates overall dry conditions in the whole period from ad 1180 to 1700. Within this period, there were periods of desiccation and re-flooding such as for instance at c. ad 1250. Despite these reoccurring dry periods, after ad 1500, the trees/shrubs increase, particularly Myrtaceae and Phyllanthus, suggesting an expansion of a riparian forest. This is somewhat surprising since the 18th century was reportedly marked by the most severe periods of droughts (Holmgren et al., 2003; Lee-Thorp et al., 2001; Scott and Lee-Thorp, 2005), but a lack of vegetation change has also been recorded in other regions in southern Africa in this time period (see section ‘Discussion’ in Ekblom et al., 2012).

The period from ad 1700 to late 1800 is suggested by the diatom record to have had high lake levels throughout (Holmgren et al., 2012). Arboreal pollen is well represented in this period, but the species composition is slightly different from other periods (Alchornea, Myrtaceae, Phyllanthus). Herbs and aquatics increase while grasses decline slightly. Brachystegia declines and disappears in the upper part of the zone.

Testing the hypothesis of Miombo expansion

Brachystegia is the only species of the Miombo that frequently occurs in the pollen diagram, Julbernardia type has been identified but very rarely, and this is a pattern also in other pollen analyses as discussed in section ‘Introduction’. Therefore, Brachystegia is here treated as a representative of the Miombo. Brachystegia is underrepresented in the pollen diagram, and even a low presence is therefore interpreted to indicate considerable coverage. It is unlikely that the low representation of forest taxa throughout the diagram can be explained as the result of land use and/or shift in fire regime as was posited in section ‘Introduction’. It cannot be excluded that farmers were active in the area from ad 300, or that hunter and gather communities may have had an impact on vegetation patterns before that, but this would have been on a small scale and not sufficient to reshape vegetation patterns. The material evidence of farming communities only dates from ad 600 in nearby Chibuene. This area probably supported a large population, but there is no evidence of an expansion/intensification of farming until ad 1000–1400. Thus, we argue that the Lake Nhauhache record, notwithstanding a low resolution in the bottom of the core, supports the third hypothesis posed in the introduction, namely, that Brachystegia and by extension – the Miombo – is a long-term constituent of the coastal landscape in southern Mozambique.

Our results indicate that Brachystegia has varied over time, with the highest representation at ad 200–700. The similarity between the appearance of Brachystegia and the diatom record from the same lake indicates that the presence of Miombo has been driven mainly by hydrological changes related to climatic variability.

Grass pollen <40 µm that may represent cereal grasses is represented throughout the core, and in the base of the diagram, this is most likely wild grasses. An increase in grass pollen <40 µm can be seen in the transition between zones 2a and 2b at ad 1000 (90 cm depth). At the same time, there are also high values of charcoal, which is here interpreted as linked with clearing and swidden agriculture taking place in the vicinity of Lake Nhauhache. However, there is no significant increase in coprophilous spores at this time. Charcoal and possible cereal grain pollen remain high throughout zone 2b until ad 1500. Farming and use of fire kept the landscape relatively open as arboreal pollen is represented by less than 20% in this period. The decline in riparian taxa at c. ad 1500 may be linked with this agricultural expansion that is also seen in Chibuene. It may also be linked with the repeated droughts occurring from ad 1150 onwards. Brachystegia remains represented in lower numbers and fluctuates over time with small peaks at ad 1000 and c. ad 1450. Possibly, this can also be related to land use, but it cannot be excluded that this is related to edaphic or hydrological changes. At ad 1500, there appears to be a decline in agricultural activities with a grass pollen <40 µm and charcoal. However, from ad 1700, there is an increase in grass pollen <40 µm, charcoal and most significantly in coprophilous spores. The latter are most probably linked with an increase of domestic stock in the area.

Brachystegia disappears at the top of the record, from ad 1700. As the resolution is low in the upper part and as we are missing the last 200 years of the record, it is unclear whether this is a temporary decline or whether this decline can be linked with the near absence of this species today in the Vilankulo region.

Conclusion

The Lake Nhauhache sequence shows that Brachystegia was most common at ad 200–700 years ago. At ad 200, a marshy forest/shrub community consisting of Stereospermum, M. africana and Combretaceae was dominant with Brachystegia common in the well-drained areas. After ad 700, this community changes to a dominance of Syzygium and Fagara linked with gradually rising water levels at the same time as Brachystegia decreases. From ad 1000, there was a general decline of trees/shrubs in favour of grasses. This was most probably linked with the opening of the landscape for farming as it is concurrent with increase in possible cereal grass pollen and charcoal. The further opening of the landscape was probably exacerbated by periodic droughts after c. ad 1200 as suggested from the diatom assemblage in the period of ad 1180–1700. Brachystegia remains represented in lower numbers after ad 700 and fluctuates over time over. The small peaks at ad 1000 and c. ad 1450 can possibly be related to land use, but it can also be related to edaphic or hydrological changes. The period from ad 1700 to late 1800 is suggested by the diatom record to have had high lake levels throughout, and arboreal pollen is well represented in this period, but the species composition is slightly different from other periods. It is clear from the analysis that Brachystegia has varied naturally over time, and its variability seems to be driven mainly by hydrological changes related to climatic variability. The hypothesis that the Miombo savanna-woodland appeared as a result of human-induced land-use changes is thus not supported by this study. The reasons behind the observed decline of Brachystegia in the last centuries can, however, not be explained without more detailed studies.

Footnotes

Funding

The work was, in part, carried out within the framework of Karin Holmgrens VR-funded project Holocene climate variability in southern Africa.

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

Badenhorst

Sinclair

Ekblom

. (2011) Faunal remains from Chibuene, an Iron Age coastal trading station in central Mozambique. Southern African Humanities 23: 1–15.

Battarbee

(1986) Diatom analysis. In: Berglund

(ed.) Handbook of Holocene Palaeoecology and Palaeohydrology. Chichester: John Wiley & Sons, pp. 527–570.

Bonnefille

Riollet

(1980) Pollens des Savanes d’Afrique Orientale. Paris: Centre National de la Recherche Scientifique.

Bronk Ramsey

(1995) Radiocarbon calibration and analysis of stratigraphy: The OxCal program. Radiocarbon 37: 425–430.

Chidumayo

(1997) Miombo Ecology and Management: An Introduction. London: Intermediate Technology.

Clark

(1988) Stratigraphic charcoal analysis on petrographic thins sections: Application to fire history in northwestern Minnesota. Quaternary Research 30: 81–91.

Clark

(1982) Point count estimation of charcoal in pollen preparations and thin sections of sediments. Pollen et Spores 24: 523–535.

Cleve-Euler

(1953) Die Diatomeen von Schweden und Finnland, T. 3, Monoraphideae, Biraphideae 1. Stockholm: Almqvist & Wiksell.

Cleve-Euler

(1955) Die Diatomeen von Schweden und Finnland, T. 4, Biraphideae 2. Stockholm: Almqvist & Wiksell.

10.

Coates Palgrave

(2002) Trees of Southern Africa. 3rd Edition. Cape Town: Struik.

11.

Coetsee

Hartley

(2001) Specialist study 3: Hydrology. In: Wood

(ed.) Environmental Impact Study of a Proposed Natural Gas Field at Temane & Pande in Mozambique. Available at: http://www.mysasol.com/natural_gas/EIA/temane_dev3.asp#s3 (accessed 19 April 2013).

12.

Cushing

(1961) Size increase in pollen grains mounted in thin slides. Pollen et Spores 3(2): 265–274.

13.

Ekblom

(2004) Changing Landscapes: An Environmental History of Chibuene, Southern Mozambique. Uppsala: Uppsala University.

14.

Ekblom

(2008) Forest-savanna dynamics in the coastal lowland of southern Mozambique, c. 400 ad–present. The Holocene 18: 1247–1257.

15.

Ekblom

Gillson

(2010) Fire history and fire ecology of Northern Kruger (KNP) and Limpopo National Park (PNL), southern Africa. The Holocene 20: 1063–1077.

16.

Ekblom

Eichorn

Sinclair

. (2013) Land use history and resource utilisation 400 ad to the present, at Chibuene, southern Mozambique. Vegetation History and Archaeobotany. Available at: http://www.springerlink.com/openurl.asp?genre=article&id=doi:10.1007/s00334-013-0392-4.

17.

Ekblom

Gillson

Holmgren

. (2012) Rainfall variability and vegetation dynamics of the lower Limpopo Valley, southern Africa, 500 ad to present. Palaeogeography, Palaeoclimatology, Palaeoecology 363–634: 69–78.

18.

El Ghazali

GEB

(1993) A study of the pollenflora of Sudan. Review of Palaeobotany and Palynology 76: 99–347.

19.

Faegri

Iversen

(1992) Textbook of Pollen Analysis. Chichester: Wiley.

20.

Finsinger

Tinner

(2005) Minimum count sums for charcoal-concentration estimates in pollen slides: Accuracy and potential errors. The Holocene 15: 293–297.

21.

Fränzle

(1984) Bodenkunde-Südafrika (23°10′–26°52′S, 29°50′–35°40′E). Berlin: Gebrüder Borntraeger.

22.

Frost

(1996) Chapter 2: The ecology of miombo woodlands. In: Campbell

(ed.) The Miombo in Transition: Woodlands and Welfare in Africa. Bogor: Center for International Forestry research, pp. 11–58.

23.

Gasse

(1986) East African Diatoms, Taxonomy, Ecological Distribution. Berlin: Gebrüder Borntraeger.

24.

Graf

Chmura

(2006) Development of modern analogues for natural, mowed and grazed grasslands using pollen assemblages and coprophilous fungi. Review of Palaeobotany and Palynology 141: 139–149.

25.

Grimm

(1991) TILIA and TILIAGRAPH. Springfield, IL:Illinois State Museum.

26.

Hall

(1981) Settlement Patterns in the Iron Age of Zululand, an Ecological Interpretation. Oxford: British Archaeological Reports.

27.

Hammer

Harper

DAT

Ryan

(2001) PAST: Paleontological statistics software package for education and data analysis. Palaeontologia Electronica 4(1): 9 pp.

28.

Holmgren

Lee-Thorp

Cooper

GRJ

. (2003) Persistent millennial scale climatic variability over the past 25,000 years in southern Africa. Quaternary Science Review 22: 2311–2326.

29.

Holmgren

Risberg

Freudendahl

. (2012) Variations in precipitation during the last 2300 years at Lake Nhauhache, Mozambique. Journal of Paleolimnology 48: 311–322.

30.

Jacobsen

Bradshaw

RHW

(1981) The selection of sites for paleovegetational studies. Quaternary Research 16: 80–96.

31.

Jarzen

Elsik

(1986) Fungal palynomorphs recovered from recent river deposits, Luangwa Valley, Zambia. Palynology 10: 35–60.

32.

Joska

MAP

Bolton

(1996) Filamentous freshwater algae in South Africa – A literature review and perspective on the development and control of weed problems. Hydrobiologia 340: 295–300.

33.

Krammer

Lange-Bertalot

(1986) Süßwasserflora von Mitteleuropa. Bacillariophyceae 1. Teil: Naviculaceae. Stuttgart: Gustav Fischer Verlag.

34.

Krammer

Lange-Bertalot

(1988) Süßwasserflora von Mitteleuropa. Bacillariophyceae. 2. Teil: Bacillariaceae, Epithemiaceae, Surirellaceae. Stuttgart: Gustav Fischer Verlag.

35.

Krammer

Lange-Bertalot

(1991a) Süßwasserflora von Mitteleuropa. Bacillariophyceae.3. Teil: Centrales, Fragilariaceae, Eunotiaceae. Stuttgart: Gustav Fischer Verlag.

36.

Krammer

Lange-Bertalot

(1991b) Süßwasserflora von Mitteleuropa. Bacillariophyceae. 4. Teil: Achnanthaceae, Kritische Ergänzungen zu Navicula (Lineolatae) und Gomphonema. Stuttgart: Gustav Fischer Verlag.

37.

Lee-Thorp

Holmgren

Lauritzen

. (2001) Rapid climate shifts in the southern African interior throughout the mid to late Holocene. Geophysical Research Letters 28: 4507–4510.

38.

McCormac

Hogg

Blackwell

. (2004) SHCal0₄ Southern Hemisphere Calibration 0–1000 cal. BP. Radiocarbon 46: 1087–1092.

39.

Marinova

Atanassova

(2006) Anthropogenic impact on vegetation and environment during the Bronze Age in the area of Lake Durankulak, NE Bulgaria: Pollen, microscopic charcoal, non-pollen palynomorphs and plant macrofossils. Review of Palaeobotany and Palynology 141: 165–178.

40.

Moll

White

(1978) The Indian Ocean coastal belt. In: Werger

(ed.) Biogeography and Ecology of Southern Africa. The Hague: Junk, pp. 561–598.

41.

Moore

Webb

Collins

(1991) Pollen Analysis. Oxford: Blackwell Science.

42.

Morais

(1988) The Early Farming Communities of Southern Mozambique. Stockholm: Central Board of National Antiquities.

43.

Neumann

Scott

Bousman

. (2010) A Holocene sequence of vegetation change at Lake Eteza, coastal KwaZulu-Natal, South Africa. Review of Palaeobotany and Palynology 162: 39–53.

44.

Neumann

Stager

Scott

. (2008) Holocene vegetation and climate records from Lake Sibaya, KwaZulu-Natal (South Africa). Review of Palaeobotany and Palynology 152: 113–128.

45.

Pitkänen

Lehtonen

Huttunen

(1999) Comparison of sedimentary microscopic charcoal particle records in a small lake with dendrochronological data: Evidence for the local origin of microscopic charcoal produced by forest fires of low intensity in eastern Finland. The Holocene 9: 559–567.

46.

Prager

Barthelmes

Theuerkauf

. (2006) Non-pollen palynomorphs from modern Alder carrs and their potential for interpreting microfossil data from peat. Review of Palaeobotany and Palynology 141: 7–31.

47.

Prentice

(1985) Pollen representation, source area, and basin size: Toward a unified theory of pollen analysis. Quaternary Research 23: 76–85.

48.

Ramsey

(1995) Radiocarbon calibration and analysis of stratigraphy: The OxCal program. Radiocarbon 37: 425–430.

49.

Scott

(1982) Late Quaternary fossil pollen grains from the Transvaal, South Africa. Review of Palaeobotany and Palynology 36: 241–278.

50.

Scott

Lee-Thorp

(2005) Holocene climatic trends and rhythms in southern Africa. In: Battarbee

Gasse

Stickley

(eds) Past Climate Variability through Europe and Africa. Dordrecht: Springer, pp. 69–91.

51.

Scott

Steenkamp

(1996) Environmental history and recent human influence at coastal Lake Teza, Kwa Zulu Natal. South African Journal of Science 92: 348–350.

52.

Sinclair

PJJ

(1987) Space, Time and Social Formation: A Territorial Approach to the Archaeology and Anthropology of Zimbabwe and Mozambique c. 0–1700 AD. Uppsala: Societas Archaeologica Upsaliensis.

53.

Sinclair

PJJ

Ekblom

Wood

(2012) Chibuene, understanding the dynamics between evolving social complexity, long distance trade and state formation in southern Africa in the late first millennium ad. Antiquity 86: 723–737.

54.

Telford

De Castro

(2001) Specialist study 4, flora and fauna. In: Wood

(ed.) Environmental Impact Study of a Proposed Natural Gas Field at Temane & Pande in Mozambique. Available at: http://www.mysasol.com/natural_gas/EIA/temane_dev3.asp#s4 (accessed 19 April 2013).

55.

Tomlinson

(1973) The Inyanga Area, an Essay in Regional Biogeography. Salisbury: University of Rhodesia.

56.

Tynni

(1979) Über Finnlands rezente und subfossile Diatomeen, VIII. Geological Survey of Finland Bulletin 274: 1–55.

57.

Tyson

Preston-Whyte

(2000) The Weather and Climate of Southern Africa. Oxford: Oxford University Press.

58.

Van Geel

Aptroot

(2006) Fossil ascomycetes in Quaternary deposits. Nova Hedwigia 82: 313–329.

59.

Van Geel

Hallewas

Pals

(1983) A late Holocene deposit under the Westfriese Zeedijk near Enkhuizen (Prov. of N-Holland, The Netherlands): Palaeoecological and archaeological aspects. Review of Palaeobotany and Palynology 38: 269–335.

60.

Van Geel

Buurman

Brinkkemper

. (2003) Environmental reconstruction of a Roman Period settlement site in Uitgeest (The Netherlands), with special reference to coprophilous fungi. Journal of Archaeological Science 30: 873–883.

61.

Van Geel

Klink

Pals

. (1986) An Upper Eemian lake deposit from Twente, eastern Netherlands. Review of Palaeobotany and Palynology 47: 31–61.

62.

Van Zinderen-Bakker

(1953) South African Pollen Grains and Spores. Cape Town: Balkema.

63.

Van Zinderen-Bakker

(1956) South African Pollen Grains and Spores. Cape Town: Balkema.

64.

Van Zinderen-Bakker

Coetzee

(1959) South African Pollen Grains and Spores. Cape Town: Balkema.

65.

Werger

(1978) Biogeographical division of southern Africa. In: Werger

(ed.) Biogeography and Ecology of Southern Africa. The Hague: Junk, pp. 145–170.

66.

White

(1983) The Vegetation of Africa: A Descriptive Memoir to Accompany the UNESCO/AETFAT/UNSO Vegetation Map of Africa. Paris: UNESCO.

67.

Wild

Fernandes

(1967) Flora Zambesiaca. (Mozambique, Malawi, Zambia, Rhodesia, Botswana) (supplement.) Vegetation Map of the Flora Zambesiaca Area. Salisbury: Rhodesia.

68.

Wild

Grandvaux Barbosa

(1967) Vegetation Map of the Flora Zambesiaca Area. Salisbury: M.O. Collins (Pvt.) Ltd.

69.

Wood

(2001) Environmental Impact Study of a Proposed Natural Gas Field at Temane & Pande in Mozambique. Available at: http://www.mysasol.com/natural_gas/EIA/temane_dev2.asp (accessed 19 April 2013).