Hydrobiologia (2008) 596:31–46 DOI 10.1007/s10750-007-9055-8
PRIMARY RESEARCH PAPER
Hydrochemical fluctuations and crustacean community composition in an ephemeral saline lake (Sua Pan, Makgadikgadi Botswana)
Graham Paul McCulloch Æ Kenneth Irvine Æ Frank D. Eckardt Æ Rob Bryant
Received: 21 January 2007 / Revised: 3 June 2007 / Accepted: 16 June 2007 / Published online: 4 August 2007 Ó Springer Science+Business Media B.V. 2007
Abstract Fluctuating hydrochemistry, as a result of the lake dried out and salinities increased. pH extreme hydrological regimes, imposes major phys- estimates generally ranged between 8.6 and 10, with iological constraints on the biota of ephemeral saline maximum values recorded during initial flooding. lakes. While the inverse relationship between salinity Crustaceans comprised mainly Branchinella spinosa, and zooplankton species richness is well-known Moina belli, Lovenula africana and Limnocythere across salinity gradients, few studies have docu- tudoranceai, all of which occurred across a wide mented closely the response of zooplankton to range of salinities, while halotolerant freshwater seasonal changes in salinity. Weekly sampling during species (Metadiaptomus transvaalensis, Leptestheria two flood seasons at Sua Pan, an intermittent saline striatochonca and the ostracods Plesiocypridopsis lake in central Botswana demonstrated the impor- aldabrae, Cypridopsis newtoni and a newly identified tance of spatial and temporal salinity gradients for Potamocypris species) disappeared above conductiv- crustacean community composition, associated with a ities of 1,500 lScm 1. A unique crustacean compo- decline in species richness, from 11 to three species. sition in southern Africa was attributed to Sua Pans’ Conductivity ranged between 320 and 125,800 lS rare chemical composition among southern African cm 1 during seasonal flooding; changing from dom- saline lakes; flood waters on Sua Pan contained a 2+ 2+ + + inance by HCO3 and CO3 ,Ca and Mg , at the higher proportion of Na and HCO3 , and less K , 2+ 2þ beginning of the floods, to NaCl dominated waters as Mg and SO4 than over 80% of records from salt pans elsewhere in southern African. The freshwater species of crustaceans in Sua Pan were similar to those found in other southern Africa lakes, and these Handling editor: J. M. Melack similarities decreased in lakes with higher pH and G. P. McCulloch (&) K. Irvine proportions of Na, and less SO4 and Mg in their Department of Zoology, University of Dublin, Trinity chemical composition. The predominant saline toler- College, Dublin 2, Ireland ant species on Sua Pan, however, showed a greater e-mail: [email protected] similarity to those in saline lakes in southern and East F. D. Eckardt Africa with higher proportions of HCO3 and, partic- 2+ Department of Environmental Sciences, University of ularly, Mg in their chemical composition. Cape Town, Cape Town, South Africa Keywords Saline lake Ionic composition R. Bryant Department of Geography, University of Sheffield, Zooplankton Makgadikgadi Species richness Sheffield, UK Salinity 123 32 Hydrobiologia (2008) 596:31–46
Introduction crustacean species richness and a simultaneous change in species composition in Lake Chad were Unpredictable and seasonal flood regimes (rainfall attributed to changes in temperature, oxygen and occurrence and intensity) among shallow and ephem- predation, while salinity was not suggested as a major eral saline lakes contribute high variability to their influencing factor (Saint Jean, 1983). physical and chemical character (Langbein, 1961; Although widely distributed in southern Africa, Hammer, 1986; Comin & Williams, 1994; Meintjes few detailed studies have been done on the hydro- et al., 1994; Williams, 1998). Salinity can vary chemistry and zooplankton communities of the enormously among and during individual flood and region’s shallow and ephemeral saline lakes (Allan- drying periods; through varying degrees of cyclical son et al., 1990; Seaman et al., 1991; Day, 1993). dissolution of surface evaporates during flooding and Ephemeral saline lakes in Namibia and South Africa evaporative concentration during drying phases (Har- range in salinities from 3 to 276 g l 1 and are die et al., 1978; Williams, 1998). Evaporative con- generally dominated by Na+ and Cl ions, with 2 centration and sequential mineral precipitation on the HCO3 +CO3 prevalent in lower salinity waters surface and sub-surface layers alter the ionic propor- (Hutchinson et al., 1932; Seaman et al., 1991; Day, tions of vanishing flood waters (Stumm & Morgan, 1993). The aquatic fauna of these lakes are thought 1970; Eugster & Hardie, 1978). Such changes affect depauperate, with no well-defined fauna confined to lake ecology profoundly (Hammer, 1986; Williams, saline pans; appearing rather to comprise mostly 1998). tolerant freshwater forms (Seaman et al., 1991). Day Differences in zooplankton communities have (1990) found that crustacean species richness among been observed among lakes with different ionic small ephemeral pans in Namibia was largely influ- composition (e.g. Hecky & Kilham, 1973; Williams enced by isolation and did not correlate with salinity. et al., 1990; Bos et al., 1996; Derry et al., 2003) and However, in nearly all cases, where the hydrochem- the inverse relationship between zooplankton species istry and associated zooplankton of these saline lakes richness and salinity in saline lakes is well estab- and ephemeral pans have been sampled, data related lished (e.g. LaBarbara & Kilham, 1974; Green, 1986, to single samples and seasonal analysis is scarce 1993; Hammer, 1986, 1993; Williams et al., 1990). (Hutchinson et al., 1932; Allanson et al., 1990; Day, Other, or a combination of other abiotic and biotic 1990, 1993; Seaman et al., 1991). factors, such as predation and competition are also This study reports the hydrological, chemical and likely important for species richness and community biological characteristics of Sua Pan; part of the composition (Williams et al., 1990; Williams, 1998; Makgadikgadi salt pan complex, Botswana, one of Herbst, 2001). However, little is known about the the largest ephemeral lake systems in Africa. Until sequential response of zooplankton communities to our study, the chemical composition of Makgadikg- changing salinity and chemical composition in indi- adi was poorly understood and not included in vidual lakes. Melack (1988) studied changes in the reviews of saline water chemistry in southern Africa biota of Lake Elmenteita during a period of lake (Mepham, 1987; Allanson et al., 1990; Seaman et al., drying and salinity increase and suggested that 1991; Day, 1993). pH values from Sua Pan and small increased salinity directly affected a key macrofaunal pans adjacent to it, taken by the Rhodesian Schools element (Paradiaptomus africanus) and an important Exploration Society in 1957, were 9 and 10 (Eccles, primary producer (Anabaenopsis arnoldii), while D. H., Undated). Brendonck & Riddoch (1997) other biota were affected through altered biological recorded pH of 8.9–9.5 and conductivities of 5.2– interactions. Vareschi & Vareschi (1984), over sev- 21.3 mS cm 1 in the north basin of Sua Pan and eral years, not only associated disappearances of some small adjacent pans. Details on the aquatic biota Lovenula africana in Lake Nakuru with increases of of Makgadikgadi is also scant (Seaman et al., 1991). conductivities above 25,000 lS, but also identified Identification of invertebrate taxa collected by the that the availability of Spirulina platensis at the base Rhodesian Schools Exploration Society suggested the of the food chain was important for the zooplankters crustacean species composition in the Makgadikgadi and rotifer population dynamics. During a 2-year was unusual for southern Africa. Subsequent collec- period of drought, a considerable reduction in tions of ostracods (Martens, 1988), copepods (Rayner 123 Hydrobiologia (2008) 596:31–46 33
& Heeg, 1994) and anostracans (Brtek, 1967; Bren- lowest part of the Makgadikgadi complex, covers donck & Riddoch, 1997) have confirmed the presence approximately 3,400 km2, with an extensive catch- of some of the species identified in Eccles’ report. ment of about 27,000 km2, comprising Kalahari Similarities to crustacean communities in East Afri- sediment based calcisols, luvisols, vertisols and can lakes have been attributed to Makgadikgadis’ gleysols underlain by Carboniferous-Jurassic sand- relative proximity to East Africa and a similar stone and basalts (Thomas & Shaw, 1991). Most of chemical composition to East African saline lakes, the soils in the area, particularly those close to the with high HCO3 and CO3 concentrations (Seaman pan, have calcrete layers at shallow or moderate et al., 1991). This article provides the first detailed depths. The pan can be divided into three hydrolog- account of the Makgadikgadis’ crustacean commu- ically separate basins, the north, middle and south nity and its association with variable salinity and basins, separated by two sand spits that protrude from water chemistry. The work involved a detailed and the eastern shore (Fig. 1). During the wet season, five systematic survey over 2 years, and the results were rivers (Nata, Semowane, Mosetse, Lepashe and compared with other saline lakes of southern and east Mosope) flow intermittently onto their respective Africa. This provides an overview of the primary basins. The north basin sump, at 890 m above sea chemical characteristics contributing to the species level, is the deepest part of the pan and floods more composition of these lakes. frequently and for longer periods than other parts of the complex (Cooke, 1979). The pan surface com- prises brine-saturated sand and clay layers with
Materials and methods surface efflorescence of trona (Na2CO3 Na2HCO3– 2H2O) and halite (NaCl) and large amounts of Site description carbonates and sulphates are thought to exist on and in the top layers of the pan surface (Thomas & Shaw, The Makgadikgadi salt-pan complex is a relict of an 1991). ancient lake that once covered most of contemporary northern Botswana (Cooke, 1979; Thomas & Shaw, Sampling and analysis 1991). Remnants of the lake exist in the form of two large endorheic salt pans, Ntwetwe Pan and Sua Pan, During two wet seasons (1999–2000 and 2000–2001), subject to intermittent flooding (Fig. 1). Mean annual daily rainfall and evaporation data were collected rainfall is estimated at between 400 and 500 mm, from the edge of Sua Pan using, respectively, a falling during the wet season, from November to standard rain gauge and Class A evaporation pan. The April, but large inter-annual fluctuations are charac- date of initial flooding and length of flood period teristic of the climate (Schulze, 1972). Sua Pan, the (duration of flood) were recorded for each basin.
Fig. 1 Map of the northern Botswana region, showing the Makgadikgadi salt pan complex, which comprises predominantly Ntwetwe and Sua Pan. Sua Pans’ inflowing rivers and catchment areas are encircled and concentric circles indicate the pattern of flood duration on the pan
123 34 Hydrobiologia (2008) 596:31–46
Conductivity and pH measurements of water variation in pH among the three basins during the samples, taken from random sites in each of the 1999–2000 and 2000–2001 wet seasons. Skewed data north, middle and south basins, were carried out bi- were normalised by square root transformation. weekly from December 1999 to June 2000, and Linear regression was used to test the relationship weekly from November 2000 to July 2001, using a between conductivity and corresponding estimates Mettler Toledo1 MC 266 meter corrected to 25°C. for total dissolved solids (TDS). Pearson Product– From December 1999 to April 2000, monthly sam- Moment correlations were used to test the association ples for total dissolved salinity (TDS) and ionic between species richness and salinity for the entire composition calculation were taken from each basin salinity gradient on Sua Pan, the individual basins, and three of the inflowing rivers (the Nata, Semow- and over the saline categories sub-saline (0 3gl 1), ane and Mosetse rivers). Water samples were hyposaline (3–20 g l 1), mesosaline (20–50 g l 1) collected in 500 ml plastic bottles, previously washed and hypersaline (>50 g l 1)(sensu Hammer, 1986), with a 5% acid solution and rinsed with distilled using Data Desk1 Version 6. The BrayCurtis water, and stored at <4°C until analyses. Chemical dissimilarity coefficient index was used to compare analysis was carried out at the National Centre for US the crustacean community assemblage of Sua Pan Geological Survey (12201 Sunrise Valley Drive, with that of similar lakes and small wetlands in Reston, VA 20192) using the methods of Fishmen & southern and East Africa. Relationships among Friedman (1989). Results were plotted on a Ternary crustacean community similarity between Sua Pan plot diagram to illustrate the changes in Sua Pans’ and the other sample saline lakes (expressed as Bray chemical composition during a flood period and to Curtis index), and their various abiotic characteristics compare its chemistry with other saline lakes in were examined using linear regressions. southern and east Africa (Day, 1993). Invertebrates were sampled weekly; using a stan- dard 240-lm sweep net (25 cm wide) to sample the Results entire water column and benthic surface or, when deep, the top 25 cm of the water column. In addition, Flood regime and water chemistry a perspex tube (diameter 5.5 cm) was dropped vertically through the water column to the lake Annual rainfall for the 1999–2000 and 2000–2001 bottom to collect an integrated sample of the entire wet seasons were, respectively, 851 mm and water column (Edmonson & Winberg, 1971). Each 348 mm. Maximum air temperatures of 38–41°C sample technique was used at five sites, chosen and evaporation rates of up to 20 mm d 1 led to rapid randomly, in each of the three basins; the north, water loss. Each basin was subject to different flood middle and south basin. Samples were then sieved regimes (Fig. 2). By the end of January 2000, the through a 70-lm mesh and each sample was north basin was completely flooded, with maximum preserved in >80 % industrial alcohol. Invertebrates depths exceeding 2 m. It remained flooded for the were identified under a dissecting microscope, remainder of the study period and only dried out according to Day et al. (1999, 2001), Brendonck & completely in February 2002. Both the middle and Riddoch (1997), Martens (1984, 1988) and Martens south basins flooded for approximately 8 months et al. (1996). Verification of ostracods was provided between November 1999 and September 2000, with a by Dr. Koen Martens (Royal Belgian Institute of brief flood period in November–December. Flooding Natural Sciences, Brussels). was considerably less during the 2000–2001 wet season, lasting less than 2 months in each basin. Statistical analysis Conductivity and pH showed both a significant spatial difference among the basins (RM-ANOVA,
Significant changes in salinity (conductivity) among F1,54 = 265, P 0.0001 for conductivity; the three basins were tested using Repeated Measures F1,53 = 593, P 0.0001 for pH) and a temporal ANOVA (RM-ANOVA) on Data Desk1 Version 6. change during both wet seasons within each basin A similar RM-ANOVA test was carried out, with H+ (Fig. 2). The north basin was least saline, increasing concentration as the dependent variable, to test the from 320 to 24,400 lScm 1 during a two-year 123 Hydrobiologia (2008) 596:31–46 35
Fig. 2 pH and conductivity measurements from the north (diamonds), middle (Solid squares) and south basins (triangles) during the 1999–2000 and 2000–2001 wet seasons. Bold lines represent flood periods in the north, middle and south basin
continuous flood period, from December 1999 to June evident in both the rivers and basins (Table 1 and 2 2001. Middle basin conductivity increased from 730 to Fig. 3). Anion dominance was HCO3 >Cl >SO4 91,600 lScm 1 in 1999–2000 and from 2,800 to (88% of the time) in the rivers, with Cl dominant in 1 65,600 lScm in 2000–2001. The most saline con- the first Nata River flood discharge. Cl > HCO3 > 2 ditions were recorded in the south basin where SO4 occurred 58% of the time in the basins, but conductivity increased from 22,400 to 111,000 lS HCO3 was the dominant anion at the beginning of cm 1 in 1999–2000 and from 46,800 to 125,800 lS the flood periods in the north and middle basins. 1 2 cm in 2000–2001. A strong positive correlation SO4 concentrations varied little and rarely exceeded (r = 0.998, n = 12, P 0.001; variables square-root 5% of anionic composition. The pattern of cation transformed) was found between conductivity and dominance was Ca2+ >K+ >Na+ >Mg2+ (63% of corresponding TDS estimates (Table 1). In the north the time) in the rivers, with Na+ dominating for most basin, pH was highest at the beginning of the flood of the time in the Nata River discharge. In contrast, (10.1), but generally increased from 8.6 in December cation dominance in the basins was Na+ >K+ > 1999 to 9.8 in June 2001. Highest pH values in both the Ca2+ >Mg2+ for 92% of the time. Na+ increased middle and south basins were also recorded at the with salinity, while the divalent cations Ca2+ and beginning of the flood periods and ranged between 9 Mg2+ decreased, disappearing completely in the south and 10. basin when the highest salinities were recorded. Si+ Spatial and temporal changes in the relative ions made up a substantial proportion of the cations proportions of the major cations and anions were in both river and basin waters. 123 36 Hydrobiologia (2008) 596:31–46
Table 1 Ionic composition of three inflowing rivers and the three basins of Sua Pan during the 1999–2000 wet season, with TDS in gl 1 and individual ion concentrations in mg l 1
+ + 2+ 2+ 2 Date TDS Na K Ca Mg Si Cl HCO3 &CO3 SO4
Rivers Nata Nov’99 1.71 560 (91.7) 30 (4.9) 11 (1.8) 3.9 (0.6) 5.5 (0.9) 670 (60.9) 350 (31.8) 75 (6.8) Jan’00 0.12 6.4 (15.9) 8.1 (20.2) 14 (34.9) 3.1 (7.7) 8.4 (20.9) 6.8 (8) 75 (87.9) 2.1 (2.5) Feb’00 0.24 35 (44.5) 7.6 (9.7) 22 (28) 4.3 (5.5) 9.5 (12.1) 35 (21.5) 120 (73.6) 6.2 (3.8) Semowane Nov’99 0.09 2.2 (7.7) 5.3 (18.5) 13 (45.3) 2.6 (9.1) 5.5 (19.2) 4.8 (7.7) 55 (88) 2.3 (3.7) Jan’00 0.1 3 (9.9) 6.1 (20.3) 14 (46.6) 2.7 (9) 4.11 (13.7) 4.8 (6.4) 65 (87.4) 2.1 (2.8) Mosetse Nov’99 0.09 4.1 (14.5) 5.6 (19.8) 10.5 (37) 2.6 (9.2) 5.4 (19) 4.9 (8) 50 (83) 2.6 (4.3) Jan’00 0.086 8.3 (29) 5.7 (20) 8.4 (29.5) 1.6 (5.6) 4.4 (15.5) 9.3 (16) 45 (77.4) 2.4 (4.1) Feb’00 0.012 4.6 (13) 5.6 (15.7) 16 (44.8) 3.3 (9.2) 6.1 (17) 5.8 (6.9) 75 (89.6) 2.5 (3) Basins North Dec’99 0.26 65 (70.2) 9.8 (10.6) 4.4 (4.8) 0.5 (0.5) 12.8 (13.8) 55 (33) 105 (63.3) 5.8 (3.5) Jan’00 1.1 335 (86) 25 (6.4) 5.2 (1.3) 0.37 (0.1) 24 (6.2) 360 (52) 300 (43.4) 29 (4.2) Feb’00 0.7 205 (81) 19 (7.5) 10.3 (4.1) 1.1 (0.4) 17 (6.7) 210 (46) 220 (48) 24 (5.3) Mar’00 1.12 365 (86.2) 27 (6.4) 6.5 (1.5) 0.79 (0.2) 24 (5.7) 425 (60.6) 240 (34.2) 35 (5) Middle Dec’99 0.58 175 (85.5) 8.4 (4.1) 8.8 (4.3) 0.8 (0.4) 11.6 (5.7) 180 (47.6) 185 (49) 11 (2.9) Jan’00 0.23 55 (75.4) 6.4 (8.8) 2.3 (3.2) 0.14 (0.2) 9 (12.4) 18 (11.6) 135 (87) 2.2 (1.4) Feb’00 6.75 2550 (93) 150 (5.5) 4.8 (0.2) 0.05 (0.01) 38 (1.4) 3600 (89.7) 155 (3.9) 250 (6.2) Mar’00 11.8 4300 (93.7) 230 (5) 4.25 (0.1) 0.05 (0.1) 55 (1.2) 6250 (86.6) 475 (6.6) 475 (6.6) Apr’00 12.9 4800 (91.2) 400 (7.6) 3.4 (0.1) 0.05 (0.1) 58 (1.1) 7000 (91.2) 160 (2.1) 500 (6.5) South Dec’99 17.8 6500 (92) 350 (5) 2.25 (<0.1) 0.4 (<0.1) 215 (3) 8500 (79.4) 1800 (16.8) 380 (3.6) Mar’00 31.5 11500 (93.5) 650 (5.3) 3.7 (<1) 0.009 (<1) 140 (1.1) 16400 (85.1) 1770 (9.2) 1050 (5.4) Apr’00 31.2 11500 (94) 600 (4.9) 4.9 (0) 0.035 (0) 125 (1) 16000 (84.4) 1700 (9) 1200 (6.3) Values in parenthesis are percentage equivalents of cations or anions
Fig. 3 Ternary diagram illustrating the patterns of ionic squares) basins. Apices indicate 100% equivalent of ions in dominance among the major cations and anions in the Nata gl 1. Areas showing the chemical composition of waters from (plus), Mosetse (cross) and Semowane (triangle) rivers, and saline lakes in East Africa and South Africa (encircled in the north (solid circles), middle (circles) and south (solid broken line), are adapted from Day (1993) 123 Hydrobiologia (2008) 596:31–46 37
Table 2 Taxonomic species list of crustaceans and other small invertebrates found on Sua Pan between November 1999 and June 2001 Class Order Species Basin Reference
Crustacean Branchiopoda Anostraca Branchinella spinosa N, M, S (Milne-Edwards, 1840) Branchinella ornata N, M, S (Daday, 1910) Spinicaudata Leptestheria striatoconcha N (Barnard, 1924) Anomopoda Moina belli N, M, S (Gurney, 1904) Daphnia barbata N (Weltner, 1897) Copepoda Calanoida Lovenula africana N, M (Daday, 1908) Cyclopoida Metadiaptomus transvaalensis N (Methuen, 1910) Ostracoda Podocopida Limnocythere tudoranceai N, M, S (Martens, 1990) Sclerocypris exserta makarikarensis N, M (Martens, 1988) Potamocypris new sp ? N (Martens, pers comm.) Plesiocypridopsis aldabrae N (Meisch, 1988) Others Coryxid hemiptera of Sigara and Notonecta genus Dytiscus species of beetle Beetle larvae, Cybister Dragonfly larvae, Libellulidae Nematode worms N, M and S represent the presence of species in the north, middle and south basins, respectively
Crustacean community composition and species- Table 3 Pearson product-moment correlations between salinity relations salinity and species richness in Sua Pan Salinity range (lScm 1) nr Significance Twelve species of crustaceans were recorded in Sua Pan (Table 2). Metadiaptomus transvaalensis, Sua Pan (320 125,800) 74 0.45 P 0.001 Plesiocypridopsis aldabrae and a new species of North Basin (320 24,400) 40 0.294 P 0.05 Potamocypris were, however, rare, occurring in low Middle basin (730 91,600) 21 0.207 P 0.2 numbers at the beginning of the 1999–2000 wet South basin (22,400–125,800) 12 0.216 P 0.5 1 season on the north basin only. Macro-invertebrates Sub-saline (0–3 g l )* 21 0.556 P 0.005 1 also occurred mainly in the north basin and Hyposaline (3–20 g l )* 28 0.159 P 0.2 included some predatory beetles and species of Mesosaline (20–50 g l 1)* 38 0.502 P 0.001 insect larvae. Hypersaline (>50 g l 1)* 19 0.691 P 0.001 There was a striking reduction in crustacean * represents salinity ranges converted from conductivity species richness, from 11 to 4, among the basins as estimates to g l 1 estimates according to Williams & salinity increased along a north–south gradient. Sherwood (1994), where the relationship between salinity and conductivity is described with a high degree of confidence Overall species richness was correlated, strongly 1.0878 2 1 by: s = 0.466 K (r = 0.988), s is the salinity in g l and negatively with salinity during the 2-year flood (between ca. 3 and 70 g l 1) and K is conductivity at 25°Cin period (Table 3). This was most noticeable in the mS cm 1 north basin, where the number of species was reduced from 11 to 3 during the flood period and the except hyposaline (Table 3). The relationship found correlation was significant, while the relationships between species richness and pH was not significant in the middle and south basins were not significant (r = 0.23; P > 0.025; n = 74). (Table 3). The species-salinity relationship examined Some species tolerated large variations in salinity within narrower salinity categories revealed highly and pH (Fig. 4). Metadiaptomus transvaalensis and significant negative correlations in all categories the ostracods, Plesiocypridopsis aldabrae, and a 123 38 Hydrobiologia (2008) 596:31–46
Fig. 4 Salinity ranges for the crustacean species at Sua Pan