Fig 1.
Simplified schematic of the pelagic food web of Lake Tanganyika.
Zooplankton are depicted as primary or secondary consumers and the three fish species are shown as juveniles or adults. Arrows indicate major predator-prey relationships. These depictions are coarse, and the precise location of the arrow tip does not necessarily indicate a preference for zooplankton higher or lower in the food web. The relative sizes of the food web members are not to scale.
Fig 2.
The limnological cycle of Lake Tanganyika with its four major phases according to Plisnier et al. [18] and Verburg et al. [19].
(a) Stagnant, highly stratified waters during the warm rainy season (November-March) only support low nutrient availability. (b) The onset of the cool dry winds in March-May initiates the upwelling in the south leading to high nutrient fluxes in this region. (c) The lake circulation reverses during the dry season (May-September). Water column stratification is low and the nutrient availability high across the lake, with a maximum in the convective mixing area in the south. (d) The trade winds cease in October slowing down the lake circulation, while the water column re-stratifies. A weaker secondary upwelling leads to a nutrient pulse at the northern end of the lake. During the dry season, wind-driven upwelling and mixing are the dominant driving force behind nutrient injections into the euphotic zone, whereas internal waves are particularly important in the rainy season. The color gradient indicates the level of thermal stratification. Note that this latitudinal cross-section is not to scale and that the outlined mechanism primarily affects the upper water column (<200 m). Our two sampling campaigns were timed at the seasonal transitions in September/October and April/May to compare the effects of the preceding dry and rainy seasons. (e) The map shows the nine stations for water column and plankton sampling. Fish samples representing the pelagic catch were collected from the respective coastal villages/towns.
Fig 3.
Physical and chemical properties of Lake Tanganyika along our north-south transects (from station 1–9) at the end of the dry season (left) and the end of the rainy season (right).
(a,b) Schmidt stability (Sc) of the 50–100 m depth interval and buoyancy frequency of the primary thermocline (N2). Distribution of (c,d) temperature (T), (e,f) dissolved oxygen, (g,h) nitrate, and (i,j) phosphate. The solid white line depicts the thermocline, whereas the dashed white line represents less pronounced secondary thermoclines. No clear thermocline had formed at station 9 at the end of the dry season. Samples are indicated by vertical lines (continuous profiles) or points (discrete samples).
Fig 4.
Phyto- (left) and zooplankton (right) parameters sampled in the north and south basins of Lake Tanganyika at the end of the dry season (Sep/Oct 2017) and the end of the rainy season (Apr/May 2018).
(a) Depth-integrated chlorophyll-a concentration, (b) phytoplankton (>10 μm) abundance, (c) depth-integrated δ13C, and (d) δ15N values of POM. (e,f) Abundances, (g,h) δ13C, and (I,j) δ15N values of the 95 μm and 250 μm zooplankton fractions, respectively. Values represent averages with standard deviations. Sample sizes are given in S1 Table (n ≤ 3). * p-value ≤ 0.1 between north and south.
Fig 5.
Carbon (normalized for C:N mass ratio according Post et al. [56]) and nitrogen stable isotope signatures of the major pelagic food web members.
(a,b) POM (c,d) zooplankton, (e,f) the bivalve Pleiodon spekii as well as the fish (g,h) Stolothrissa tanganicae, (i,j) Limnothrissa miodon, and (k,l) Lates stappersii at the end of the dry season (left) and the end of the rainy season (right). Orange dots represent the northern basin and blue dots represent the southern basin. Numbers indicate the mean δ13C (black) and δ15N (grey) values from each basin and Δ denotes the δ13C difference between the southern and northern mean values. Ellipses encompass approximately 67% of the data for plankton samples (a-d) and 95% of the data from each basin for tissue samples (e-l). Note different axis limits. * p-value < 0.05 between north and south.
Fig 6.
Mass C:N ratios of primary consumers (left) and fish tissue (right) for the different sampling campaigns and basins of Lake Tanganyika.
(a) Pleiodon spekii, (c, e) zooplankton, (b) Stolothrissa tanganicae, (d) Limnothrissa miodon, and (f) Lates stappersii. Numbers depict the % change in estimated lipid content according to Post et al. [56]. Note varying y-axis scaling. * p-value < 0.05 between Sep/Oct and Apr/May.
Fig 7.
Spearman-rank correlation matrix of physical, plankton, and δ13C variables for (a) the end of the dry season and (b) the end of the rainy season.
For each season, we selected the five stations across the north-south transect with the highest overlap among all variables (Sep/Oct: stations 1, 2, 6, 7, 9; Apr/May: stations 1, 2, 4, 7, 8; S2 Table). Insignificant correlations (p > 0.05) are marked by grey crosses. Depth thermo: depth of the primary thermocline; N2 thermo: buoyancy frequency of the primary thermocline; Sc50-100 m: Schmidt stability of the 50–100 m depth interval; Phyto10 μm: phytoplankton abundance of the >10 μm size fraction; Zoo25/95/250 μm: zooplankton parameters of the >25, >95, or >250 μm size fractions.
Fig 8.
Principal component analyses of the C and N isotopic compositions as well as mass C:N ratios of the food web.
Analyses were done (a) on all food web members and (b) on bivalve and fish tissue samples. Variables in panel a are log transformed.
Fig 9.
Schematic synthesizing the main conclusions and hypotheses of the study.
(a) Biological productivity of phyto- and zooplankton based on abundance and δ13C data, which were used to (b) infer the distribution of regional fish stocks from their δ13C signatures. (c) The regional isolation of the fish stocks at seasonal timescales does not translate into suppressed gene flow at generational timescales, as indicated by a lack of regional genetic structure in these species. (d) The regional fish stocks as well as different genetic clusters did not exhibit systematic differences in δ15N. (e) The clupeid Stolothrissa exhibited strong seasonal changes in C:N, i.e. lipid content, indicating lipid storage after the productive dry season. *results from Junker et al. [35] and Rick et al. [36].