Environmental factors that influence species diversity of floodplain plant communities in different flooding phases in the Okavango Delta , Botswana

Background and objectives: Species composition and distribution in seasonal floodplain plant communities are influenced by variation in flooding. However, the influence of intra-flooding variation phases on the diversity of seasonal floodplain plant communities has not been studied in the Okavango Delta. The objective of this study was to investigate environmental factors that influence species composition and distribution of seasonal floodplain communities before and after flooding. It was hypothesised that environmental factors that influence the species composition and distribution in seasonal floodplain communities will vary with intra-flooding seasons.


Introduction
Seasonal floodplain plant communities are an important component of the ecology of the wetland ecosystems. They are dynamic and heterogeneous (Benstead et al. 1997;Toogood et al. 2008) with their species composition and distribution influenced by the variation in seasonal flooding, which is a function of topography (Merritt et al. 2009;Oliveria-Filho et al. 1994;Toogood, Joyce & Waite 2008). Lowly elevated floodplains usually have higher water levels and longer flooding duration than highly elevated ones (Growing, Spoor & Mountford 1998). This results in the lowly elevated floodplains being dominated by flood tolerant species such as Hydrilla verticillata (L.f.) Royle, which grows well when fully submerged and Typha angustifolia L. (Cronk & Fennessy 2000). Highly elevated floodplains are dominated by flood intolerant species such as Chrysopogon nigritanus (Benth.) Veldkamp, Sporobolus spicatus (Vahl) Kunth and Imperata cylindrica (L.) Raeusch. (Bonyongo, Bredenkamp & Veenendaal 2000).
The influence of flooding on plants is reflected in seasonal floodplain vegetation community composition and distribution (Oliveria-Filho et al. 1994). Each plant species is morphologically and physiologically adapted to a range of flooding depth and duration conditions (Cronk & Fennessy 2001;Edwards & Kollman 2002;Merritt et al. 2009). Morphological adaptation to flooding such as growth of hytrophied lenticels, aerenchyma tissue and adventitious roots enhance oxygen transport in plants, enabling them to survive flooding conditions (Kozlowski 1984). Physiological adaptations to flooding include germination inhibition, glycolysis and ethylene production (Naiman & Decamps 1997).
In the Okavango Delta, previous studies have also attributed plant species composition and distribution in seasonal floodplain communities to flooding depth (Bonyongo et al. 2000;Murray-Hudson 2009). It has also been found that other environmental factors, including distance from the water source, ground water electrical conductivity, pH, soil chemistry (Ellery et al. 1993), timing of inundation, soil salinity, and nutrient and sediment supply (Ellery & Tacheba 2003) were responsible for controlling the species composition and distribution in seasonal floodplain plant communities. However, these conclusions were based on studies conducted during low flood. As a result they do not provide information on significant environmental factors determining seasonal floodplain vegetation community composition and distribution during a high flood.
Furthermore, the previous studies on the species composition and distribution of seasonal floodplain communities did not consider the intra-flooding variation phases of middle of rainfall season, before flooding and after flooding. To address this, the current study aimed to investigate environmental factors determining seasonal floodplain vegetation community composition and distribution before flooding and after flooding during a high flood year in the Okavango Delta. It was hypothesised that environmental factors that influence the species composition and distribution in seasonal floodplain communities in the Okavango Delta will vary with flooding phases.

Study site
This study was conducted in the Okavango Delta in the Nxaraga lagoon area (Figure 1).
The Okavango Delta is fed by local rainfall and seasonal floods from the Angolan highlands (Gumbricht, McCarthy & McCarthy 2004), which are asynchronous (McCarthy & Ellery 1998). The total flooded area in the Okavango Delta ranges between 4 000 km 2 and 13 000 km 2 (McCarthy 2006). This variation is influenced by the seasonal variation in the local rainfall and flood discharge (Gumbricht et al. 2004). The Delta receives the lowest inflow between September and November and receives the highest inflow between March and April (McCarthy & Ellery 1998). Mean maximum rainfall in the Okavango Delta ranges between 300 and 550 mm/year (Wilson & Dincer 1976). The annual mean flood discharge ranges between 6.0 × 10 9 m 3 and 16.4 × 10 9 m 3 (Gumbricht et al. 2004) of which approximately 96% is lost through evapotranspiration while 2% is lost through infiltration (Wilson & Dincer 1976). Another 2% is lost as outflow through Thamalakane River (Wilson & Dincer 1976).
There are three hydrological regions in the Delta namely: permanent swamp, seasonally flooded floodplains (primary and secondary floodplains) and occasionally flooded floodplains (tertiary floodplains) (Gumbricht et al. 2004). Each hydrological region is characterised by a particular range of hydroperiod (Wolski et al. 2006) and associated vegetation communities (Bonyongo et al. 2000) ( Table 1).
The Delta experiences mean monthly maximum and minimum summer temperatures ranging from 30.5°C to 40°C and 14.8°C to 19.2°C respectively (Ellery 1991). During winter mean monthly maximum temperature ranges from 25.3°C to 28.7°C and minimum temperature ranges from 7.0°C to 10.0°C (Ellery 1991).

Hydrology and vegetation sampling
Flooding depth (in centimetres) was measured using a calibrated 2 m PVC pipe in mid-May 2010 (flood propagation) and end of September 2010 (flood recession). It was measured in 25 m 2 permanent plots in each of the hydrological regions (Table 1) per flooding season (before flooding and flood recession) where vegetation and soil were also sampled. A total of 40 plots were sampled. The soils were collected from the centre of each permanent plot using a soil auger at a maximum depth of 30 m. Flooding duration was recorded as the number of weeks in which the permanent plots remained inundated.
Vegetation was sampled from randomly selected plots during the same period when flooding parameters were sampled. The plots were placed in different floodplains of primary (10 plots), secondary (15 plots) and tertiary (15 plots) (Bonyongo et al. 2000;Gumbricht et al. 2004; Table 1). The dimension for the vegetation plots were 5 × 5 m. This is the minimal sampling plot area that was determined by Bonyongo et al. (2000) for sampling the seasonal floodplain vegetation in this study area. In each plot, plant percentage cover was estimated following the modified method by Braun-Blanquet (Mueller-Dombois & Ellenberg 1974). The Braun-Blanquet method was used to estimate plant species percentage cover using an ordinal scale (Table 2). Percentage cover was estimated for the emergent plant species only.

Measured environmental variables
The soil was sampled before and after floods and analysed for pH, extractable P, K, Mg, Ca and Na at the University of Botswana Okavango Research Institute Laboratory. They were collected at a depth of 30 cm from the same 25 m 2 plots where plant species were sampled. A detailed analysis of soil samples is given in Tsheboeng et al. (2014).

Statistical data analysis
The relationship between environmental variables and seasonal floodplain vegetation community composition and distribution was sought using Non-metric Multidimensional Scaling (NMS) (Kruskal 1964;Mather 1976) in PC-ORD version 6. NMS was used to relate soil nutrients, flooding depth and duration to vegetation community composition and distribution.

Results
Flooding depth was significantly (p < 0.05) higher in all hydrological zones after flooding than before flooding ( Table 3). The content of K, Ca and flooding duration were significantly higher (p < 0.05) before than after floods (Table 3).

Factors that influence floodplain plant community composition and distribution
Before floods factors NMS ordination showed that before flooding, important factors that influence the distribution of seasonal floodplain communities were Na, K, flooding depth and flooding duration (Figure 2).
Before flooding Na, Mg, K, Ca and pH were negatively correlated with species along Axis 1, but positively correlated to those along Axis 2. Flooding duration and P were negatively correlated with species oriented along Axis 2 while flooding depth was positively correlated to both axes (Table 4 and Figure 2).

After flooding
After flooding the significant factors that influenced the species composition in seasonal floodplain communities were K, Na, Mg, pH, flooding depth and flooding duration ( Figure 3).
After flooding Na, Mg, K and Ca were negatively correlated with species distributed along axes 1 and 2. Flooding duration and pH were negatively correlated to axes 1 and 2 respectively. Flooding depth was positively correlated to both axes, while P was negatively correlated to species along axis 2 (Table 5 and Figure 3).

Discussion
This study showed that there was variation in environmental factors that influence species composition and distribution of seasonal floodplain plant communities in the Okavango Delta. Factors that significantly influenced plant species composition before flooding are Na, K, water depth and flooding duration. However, after flooding there were additional factors of K, Mg and pH. This suggests that during flooding these cations are deposited on floodplains due to lateral water flow resulting in their increased concentration in these habitats. Another explanation of this could be evapo-concentration of K and Mg as the flood recedes, which results in increased concentrations of these elements. As a result their influence on the composition and distribution of seasonal floodplains becomes significant.
Species whose distribution was influenced by Na, Mg, K and pH include Sporobolus spicatus, Sporobolus acinifolius Stapf, Cynodon dactylon and Imperata cylindrica. These species are generally adapted to saline conditions that are associated with these chemicals. Sporobolus species are adapted to salinity through increased area of the root; stem, leaf blade and leaf sheath (Hameed, Ashraf & Naz 2011). Enlargement of these organs enhance the excretion of saline ions such as Na + and Clin large quantities (Flowers & Colmer 2008). To further enhance the excretion of saline ions, Sporobolus species have increased vesicular hair density and developed aerenchyma tissue (Hameed, Ashraf & Naz 2011). Cynodon dactylon survives saline conditions through exclusion of toxic ions through leaves enhanced by increased density of vesicular hairs on both adaxial and abaxial leaf surfaces (Marcum 1999). Imperata cylindrica has also been found to survive saline conditions in wetland ecosystems (McDonald 2004). This plant copes with salinity through anatomical adaptations such as increased succulence of the midrib and cortical parenchyma, which may help in the sequestration of ions (Hameed, Ashraf & Naz 2009). In addition to this, Hameed, Ashraf & Naz (2009) found that Imperata cylindrica develops enlarged bulliform cells, which help in folding the leaves to minimise water loss during salt stress. Other anatomical adaptation strategies in Imperata cylindrica include reduced root area, which helps it to absorb Na + and Clin lower quantities and development of aerenchyma tissue to enhance ion excretion (Hameed, Ashraf & Naz 2011).
Flooding duration and depth influenced species composition and distribution before flooding and after flooding. This may have implications for their survival during drought conditions caused by climate change. It is predicted that hydrological changes resulting from climate change will affect species composition and distribution such that only species that are tolerant of drought conditions survive during low water levels (Middleton 2009). The influence of flooding duration and depth is important from a management point of view in the Okavango Delta. Flooding can be manipulated from upstream impoundments of the Okavango River Basin, which will reduce the inflow into the distal regions such as the Okavango Delta in Botswana, which in turn may result in changes in floodplain species community composition and distribution. This suggests that any water abstraction from the Okavango River Basin should take into consideration the fact that flooding duration and depth are important in sustaining the species composition and distribution of seasonal floodplain plant communities such that those developments do not disturb this. However, experimental studies are still needed to give accurate predictions of seasonal floodplain plant communities to flooding duration and depth. In this study, species that were influenced by water depth include Oryza longistaminata A.Chev. & Roehr, Nymphaea lotus L., Leersia hexandra Sw., Eleocharis dulcis (Burm.f.) Trin. ex Hensch and Nymphaea nouchali Burm.f. These species are tolerant of prolonged flooding duration and high depth. They survive flooding conditions through development of fleshy, hollow stems and adventitious roots that grow from the submerged nodes (Ellenbroek 2012). These adaptations help in the absorption of oxygen and carbon dioxide for the processes of respiration and photosynthesis respectively. For the absorption of light they have developed large leaves to increase the surface area (Ellenbroek 2012).
The findings of this study also agree with results from studies conducted elsewhere (e.g. Gregory et al. 1991;   Junk 1997;Rees 1978;Zeilhofer & Schessl 1999). However, it should be noted that these studies did not investigate the intra-annual variation of environmental factors that influence the species composition and distribution in seasonal floodplain communities. In a study in the Pantanal seasonal floodplains, Zeilhofer and Schessl (1999) found that seasonal floodplain vegetation community composition and distribution were influenced by flooding depth and duration gradient. A short grassland vegetation community dominated by flood tolerant Vochysia divergens Pohl was found in longer-duration, deeply flooded sites, whereas a medium tall grassland vegetation community occurred in areas experiencing short flooding duration and shallow flooding depth. The influence of flooding depth and duration on seasonal floodplain plant communities was also observed in the Amazon seasonal floodplains (Gregory et al. 1991;Junk 1997;De Simone et al. 2003). The findings of Rees (1978) from a study conducted in the Kafue seasonal floodplains also agree with the observation made in this study. In that study, floodplain plant communities were distributed according to their tolerance to flooding duration and depth gradient. Regions that were frequently flooded with high flooding duration and depth were dominated by Vossia cuspidata (Roxb.) Griff and Echinochloa stagnina (Retz.) P.Beauv. In the Okavango Delta past studies also found that flooding duration and depth influence vegetation community composition and distribution (e.g. Biggs 1976;Bonyongo et al. 2000;Ellery & Tacheba 2003;Ellery et al. 1993;Murray-Hudson 2009;Smith 1976;Tsheboeng et al. 2014).

Conclusion and implications for management of floodplain plant communities in the Okavango Delta
This study has shown that there are seasonal variations in the environmental factors that influence the species composition and distribution of seasonal floodplain plant communities in the Okavango Delta. Factors that influenced plant species composition and distribution before flooding were Na, K, water depth and flooding duration. After flooding there were additional factors of K, Mg, pH. From a management perspective, this study suggests that the influence of these environmental factors should be considered before any major developments such as impoundments are implemented so that the species composition and distribution are not disturbed. To make accurate quantitative projections on the changes of species composition and distribution that may result from changes in hydrological regime, future experimental studies are needed. Those studies should quantify the quantity of water needed to sustain the seasonal floodplain plant communities in the Okavango Delta. Future studies should also investigate the changes in soil nutrient content and toxic substances associated with extended flooding duration. Such studies should also investigate the influence of flooding on primary production of seasonal plant communities.