Phytoplankton periodicity and sequences of dominance in an Amazonian flood-plain lake (Lago Batata, Pará, Brasil): responses to gradual environmental change

169

Hydrobiologia 346: 169–181, 1997. c 1997 Kluwer Academic Publishers. Printed in Belgium.

Phytoplankton periodicity and sequences of dominance in an Amazonian flood-plain lake (Lago Batata, Para´ , Brazil): responses to gradual environmental change Vera L´ucia de M. Huszar1 & Colin S. Reynolds2 1

Laboratory of Phycology, Dept. Botany, Museu Nacional, Universidade Federal do Rio de Janeiro, Quinta da Boa Vista, S˜ao Crist´ov˜ao, Rio de Janeiro, 20 940-040, Brazil 2 Freshwater Biological Association, NERC Institute of Freshwater Ecology, The Ferry House, Ambleside, Cumbria LA22 OLP, UK Received 11 September 1996; in revised form 14 January 1997; accepted 13 February 1997

Key words: phytoplankton, succession, flood-plain lake, Amazon

Abstract The composition of the phytoplankton of Lago Batata, a flood-plain lake connected to Rio Trombetas, undergoes a conspicuous annual cycle which is related to the hydrology (depth of water, rate of fluvial flushing) and the hydrography (stability, frequency of mixing of the water) of the lake. From a sparse nanoplankton at high-water and high flushing, the lake passes to desmid-diatom dominance and finally to filamentous cyanobacteria when the lake is barely 2 m deep. As it refills, the lake again becomes desmid-dominated; then, when the turbidity is least and the stratification most stable, Botryococcus becomes a major component. Eventually flushing becomes too rapid for any but the relatively fastest-growing species. These changes are gradual and, at the scale of algal generation times, cannot be explained as sharp or sudden disturbances. Neither do they have the properties of ecological successions but rather represent compositional responses to a progressive environmental modification analogous to the floristic phenomenon of gradual climate change. Introduction The concept of succession is a cornerstone of community ecology and energetics yet ecologists continue to debate its properties, its drives, its steady-state outcome and its relationship to species diversity. In this context, the ecosystem development of planktonic communities of lakes and seas should be attractive to theorists, because the short life spans of successive generations make them amenable to study, to useful experimentation and to the testing of hypotheses. To do this purposefully and convincingly, however, it is important that plankton ecologists discern the causes and courses of compositional change in planktonic assemblages and not, as has so often been the case, simply refer uncritically every sequence of changing community dominance to species succession. Taking the ‘rules’ of Odum (1969) on ecosystem development and his diagnosis of ecological succes-

*136037

sion as a model, Reynolds (1988) distinguished those changes in the structure and the composition of the phytoplankton which are the direct, autogenic consequences of population growth (nutrient depletion, increased self shading, more grazing) from those which are driven by density-independent, allogenic forcing (such as wind-mixing events, storms or floods). The former lead to structural complexity, with the selection of species with increasingly specialised adaptations, and declining productivity. The latter simplify structure, rejuvenate resources and stimulate productivity. Such externally driven events generally cut across the self-organising process of ecosystems, as it were, in mid-succession, each time resetting the structure and organisation of the community to some less well-developed state. Indeed, the frequency of these interventions, or disturbances (Connell, 1978), impacts upon the relative maturity of ecosystems, the strength of species selection, the importance of inter-

Article: hydr 3729 Pips nr 136037 BIO2KAP hydr3729.tex; 24/06/1997; 9:45; v.7; p.1

170 specific competition and the diversity of species representation: periodic disturbances resist the maturation of planktonic successions, so competitive interactions are weak and diversity is high (Reynolds, 1988, 1993; Padis´ak, 1994). If substantially correct, these deductions about planktonic successions have considerable relevance to the interpretation of structural and functional changes in terrestrial ecosystems, provided due allowance is made for the divergent time-scales. In this way, it would be possible to analogise between the process of re-establishment of canopy-forming trees in an area of woodland damaged by fire or wind to the development of a plankton bloom, after the onset of thermal stratification. In his criticism of the application of intermediate-disturbance theory to the phenomena associated with phytoplankton community dynamics, Wilson (1994) ventured that, were the scaling differences between forests and plankton consistent, it would it be difficult to determine whether the sequence of community responses represented a succession or whether it is not simply the replacement of one species by another better adapted to the gradual, seasonallyforced changes in the environment that had taken place in the few weeks separating different generations. Wilson (1994) analogised such changes to the supplanting of one dominant type of vegetation by another under the influence of gradual climate change. We acknowledge that this is well-reasoned logic and that it would be wrong to view all change in phytoplankton communities as being the result either of strict ecological succession or of non-successional disturbance responses to sudden, externally-forced environmental changes. To reinforce the analogy with changes in terrestrial vegetation, Reynolds (1993) distinguished the changing dominance of temperate forests since the last glaciation from either the role of that of storms and catastrophes, in disturbing contemporaneous progress towards continuous forests, or that of contemporaneous successions (open habitat – scrub – forest) in re-establishing the contemporaneous climax state. The changes in the potential climatic climax vegetation at the higher northern-hemisphere latitudes since the last glaciation (Picea – Betula/Pinus – Corylus/Alnus/Ulmus – Quercus/Fagus) owe to the fact that, under the influence of gradual but progressive changes in the global climate, different species have been favoured as climactic dominants, as a function of their various specialisms and predilections. As a result, the character of the vegetation has altered through the progressive supplantation of one succes-

sional sequence and outcome by another. Present-day Quercus forest is not the product of a single, uninterrupted ecological succession stretching from the time that the first tundra or periglacial meadow developed in the wake of the retreating ice fronts. We deduce that, while the mechanisms of succession and change remain broadly unchanged, the outcome, in terms of the species selected, is always liable to change. By the same argument, successional sequences are probabilistic rather than deterministic. The ‘successional template’ relates to the process and not to any single species progression, representation or outcome; indeed, Reynolds (1993) cites several distinct types of plankton succession. However, it is germane to demonstrate that gradual climate change also has a planktonic analogue and to do so is the primary purpose of this paper. We attempt to demonstrate how one successional pathway can be supplanted by another in response to a gradual change in the hydrographic environment and to show that the process complies neither with any well-characterised planktonic succession nor to a sequence of biotic reactions to external disturbances. Our evidence is drawn from the phytoplankton periodicity of an Amazonian flood-plain lake, where striking cyclicity in the organisation, structure and species composition is related to the annual hydrological cycle of the Amazon Basin.

Description of the study site Lago Batata (hereinafter, Batata Lake: 1 280 S, 56 140 W) is a clear-water lake situated on the right bank of the Trombetas River, Par´a, Brazil (Figure 1). This region has a humid, tropical climate, with an annual median temperature of 27  C and a mean annual precipitation of 2100 mm (Brasil, 1976). Although separated from the river along most of its eastern edge by a lateral levee, Batata Lake is in permanent connection with the Trombetas River (Panosso, 1993). The level of the river fluctuates on a conspicuous annual cycle determined mainly by seasonal variations in the intensity of precipitation in the catchment and by its outfall to the Amazon River. At its highest winter levels, the Trombetas overflows the levee. As the depth of the water in the river fluctuates with the discharge, however, so does the depth and extent of the lake, varying between a maximum of 12.0 m and 31 km2 at high water level (between March and July) and 2 m and 18 km2 at low water level (October to December). These dates are approximate but they serve to identify two further

hydr3729.tex; 24/06/1997; 9:45; v.7; p.2

171 definable periods in the hydrological cycle of the lake: one of falling water level (August–September) and one of rising water level (January–February). It is clear that the changes in the physical, chemical and biotic environment of Batata Lake are driven principally by the hydrological variability. A further feature of the study site is the bauxite tailings which cover some 30% of the total lake area at high-water level (Roland & Esteves, 1993). These arose from ore processing by Minerac¸a˜ o Rio do Norte S.A. and were discharged into the west part of Batata Lake between 1979 and 1989. The impact of the tailings on the phytoplankton of the lake was investigated by Huszar (1994). They have not influenced the results presented and discussed here.

The rate of community compositional change ( ) was calculated according to Lewis’ (1978) summeddifference method: if bi (t) is the abundance of the ith species and B (t) is the sum of the individuals making up the sampled community, the rate of change in composition between two given dates t1 and t2 is solved from:

Materials and methods

Observations

The data considered in the present study are derived from samples of the phytoplankton collected at weekly intervals between September 1988 and October 1989, from just beneath the water surface at a fixed point in the main basin (Figure 1). The sampling programme was carried out as a background to more intensive studies of the limnology of Batata Lake, which were undertaken during September (when the level of the lake was falling) and December of 1988 (low water) and during March (rising) and July (high water) of 1989. The results of these studies are reported elsewhere (Panosso, 1993; Roland & Esteves, 1993; Bozelli, 1994; Huszar, 1994; Melo, 1996). The focus here is on the changes in the dominant phytoplankton species through the year but appropriate supporting information is used from the detailed studies wherever it is relevant. The samples were fixed with Lugol’s Iodine and the phytoplankton were identified and enumerated some time after collection, following the sedimentation and inverted-microscope method of Uterm¨ohl (1958). Specific biomass was estimated from the product of the population and the mean unit volume (Edler, 1979). Species diversity was approximated by application of the Shannon-Wiener Index (Shannon & Weaver, 1963):

Physical aspects

H 00 =

n =N log2 (n =N ); i

i

where ni is the measure of the ith species (as specific wet biomass calculated as the product of a cell-volume approximation and the number of cells counted; unfortunately, no direct measurements are available) and N is  ni .

 = if[bi (t1 )=B (t1 )]

[b (t2 )=B (t2 )]g=(t2 i

t1 ):

Note that the quantity of  is important. The sign is disregarded.

In Figure 2, fluctuations in the depth of Batata Lake are compared to the annual variation in the level of the Trombetas River. The pattern of hydrographic variability is driven almost exclusively by seasonal differences in the distribution of precipitation in the headwaters of the Trombetas River (Huszar, 1994). In contrast, daylength, ambient air- and water temperatures, even the clarity, gas- and solute- content of the lake water, fluctuate little through the year (Table 1). The dominant role of the river hydrology in determining the seasonality of Batata Lake is clearly recognisable.The principal effect of level fluctuation in Batata Lake is to impose an annual cycle of alternation between being stratified to being effectively unstratified, with important transitional conditions in either direction. Illustrations of the depth-distributions of isotherms during high- and low-water periods are compared with episodes of falling and rising water in Figure 3. It is also relevant to the development of phytoplankton that a close correspondence between transparency (as determined by Secchi-disc extinction) and water level has been found in later work on the lake (Figure 4). The poor transparency at low water is attributed partly to algal biomass and partly to the frequent resuspension of fine sedimentary material. The fact that the lake is most turbid when it is shallow emphasises the between-season variation and the extent of the change in the underwater light climate that is progressively undergone with each cycle.

hydr3729.tex; 24/06/1997; 9:45; v.7; p.3

172

Figure 1. Maps to show (above) the location of Batata Lake and (below) its relation to the Trombetas River.

Biological aspects One hundred and seventy four species of phytoplankton were noted during this study (Huszar, 1994 and

in press). Here, we are concerned principally with the fluctuating abundance of the phytoplankton and of the dominating species (listed, with authorities, in Appendix 1). Fluctuations in abundance, expressed as fresh

hydr3729.tex; 24/06/1997; 9:45; v.7; p.4

173 Table 1. Climatological, morphometrical, physical and chemical data, considering trimonthly sampling in all lake Batata (means). (from Huszar, 1994) Feature

Falling

Low-water

Area (km2 ) Level (m) Depth-Central Body (m) Accumulated Precipitation (mm) Air Temp. ( C) Water Temp. ( C) Secchi Disk (m) Conductivity (S.cm 1 ) Dissolved Oxygen - surface (%) Dissolved Oxygen - bottom (%) pH PO34 (g.l 1 ) TP (g.l 1 ) NO33 (g.l 1 ) NH14+ (g.l 1 ) TN (mg.l 1 ) SiO2 (mg.l 1 ) Si/PT TN/TP

23 44.1 5.0 394 27.8 29.9 1.3 11 82 41 6.0 < 5.0 19.4 85.0 158.2 0.53 2.0 111 60

18 41.1 2.0 826 26.3 30.7 0.9 9 95 91 5.6 < 5.0 26.7 25.9 108.8 0.46 1.2 50 38

Rising ? 45.7 7.0 1284 26.0 28.5 1.5 9 108 75 5.9 < 5.0 10.7 9.6 118.3 0.59 2.7 279 124

High-water 31 48.4 9.5 239 26.3 29.6 1.0 11 97 91 5.4 7.5 23.7 21.4 72.3 0.64 2.3 110 60

 Panosso (1993-);  Roland & Esteves (1993);  Atomic ratio

Figure 2. Fluctuations of the level of the Trombetas River (solid line, right hand scale in m elevation) and the depth of the lake (broken line). The symbols LW and HW refer to the periods of low and high water, R to the filling (‘rising’) phase and F to the phase of falling water level. The extent of the water column calculated to be vulnerable to diel mixing is shown in the light stipple; that requiring abnormal mixing strength to penetrate the full depth is shown by starker stippling. In the period of its duration, full-mixing events would have occurred irregularly and a thermocline at beyond 5 m would have been a relatively persistent feature.

hydr3729.tex; 24/06/1997; 9:45; v.7; p.5

Figure 3. Examples of the diel distribution of isotherms in Batata Lake, from selected episodes of falling water level, low water, filling and high water.

174

hydr3729.tex; 24/06/1997; 9:45; v.7; p.6

175

Figure 4. Fluctuations in Secchi-disk transparency of Batata Lake in relation to fluctuations in the level of the Trombetas River.

biovolume, are shown in Figure 5. The plot is subdivided to show major phylogenetic affinities. The size distribution of the biomass is shown in Figure 6. The calculated rates of change in the species composition, which distinguish periods of rapid modification (high values) from phases of more stable community structure or only gradual change, are shown in Figure 7. Fluctuations in the instantaneous species diversity of the phytoplankton assemblage are represented in Figure 8. At the beginning of the study, a modest plankton biomass was dominated by Cryptomonas cf. marssonii, Cyclotella stelligera, Peridinium umbonatum and Merismopedia tenuissima. With falling waterlevel, relatively light rain, initially good light penetration and increasingly frequent mixing of the full water column, the biomass began to increase quite rapidly (Figure 5). The expanding biomass involved Cryptomonas, Cyclotella and, over the next five weeks, an increasing dominance of microplanktonic desmids – especially Staurodesmus triangularis, Pleurotaenium tenuissimum and Staurastrum pseudotetracerum – and the diatom Aulacoseira granulata. These changes are reflected in the shift in the spectrum of sizes represented (small to large: Figure 6), in a rise and then a stabilisation in the rate of compositional change (Figure 7), and a parallel small fluctuation in species diversity

(Figure 8). The highest biomass (16.3 mm3 l 1 ) was attained in October, during the low-water period. During November, however, the populations of diatoms, then of desmids, declined somewhat, while those of Jaaginema geminatum (formerly Oscillatoria geminata) and Oscillatoria cf. transvaalensis increased to dominance. This phase was relatively short-lived. The slow ingression of river water during December, the gradual rise in water level and the onset of incomplete diel mixing coincided with the development of a more diverse plankton, involving small chlorophytes (Chlorella homosphaera, Scenedesmus ellipticus), diatoms (Cyclotella stelligera, Urosolenia eriensis), small (Cosmarium arctoum) and larger desmids (Staurodesmus triangularis, Pleurotaenium tenuissimum). A desmid-dominated maximum, equivalent to 7.0 mm3 l 1 , was attained in February. By March, the lake was 8 m in depth, more or less continuously stratified and, relatively clear (an estimated 60% of the lake volume lay within the euphotic zone in March: Huszar, 1994). Colonies of the chlorophyte Botryococcus fernandoi and of the cyanobacterium Merismopedia tenuissima became more prominent to share dominance with desmids and Jaaginema geminatum. As time progressed, however, the ingress of river water became significant, with progressively

hydr3729.tex; 24/06/1997; 9:45; v.7; p.7

176

Figure 5. Fluctuations in the phytoplankton biomass (as fresh biovolume) in Batata Lake, subdivided according to phylogenetic affinities, in relation to the level of the Trombetas River and the hydrological phase of the lake.

Figure 6. Fluctuations in the proportion of the phytoplankton biomass (= 100%) allocated among four size categories.

more frequent and more rapid flushing, as well as a reduction in water clarity. As the high-water condition became established in April, the abundance of the plankton became subject to rapid dilution (Figure 5) and its species composition to more rapid fluctuation (Figure 7). The community soon became characterised by a low biomass dominated by small, nanoplanktonic algae, including Cryptomonas cf. marssoni, Cyclotella stelligera and Chromulina cf.gyrans, with a slow rate

of change and diminished diversity. Larger algae were conspicuously absent during this phase (Figure 6). The high-water condition persisted until June when the flushing weakened and the water level of the lake began started to drop consistently. The biomass, still dominated by cryptomonads and small centric diatoms, increased somewhat. As the emptying phase progressed and the rate of dilution declined yet further, so the community diversified and the biomass expanded

hydr3729.tex; 24/06/1997; 9:45; v.7; p.8

177

Figure 7. The daily rate of community change, calculated by the method of Lewis (1978) and based upon weekly data, in relation to the hydrological phase of Batata Lake.

Figure 8. Changes in the Shannon-Wiener diversity index of phytoplankton in relation to the hydrological phase of Batata Lake.

further: Staurastrum longipes was a prominent desmid at this time and Aulacoseira was represented among the diatoms. The structure of the community was, thus, quite similar to that which had been carried one year previously.

Discussion The physical environment of Batata Lake It is readily apparent that the seasonal fluctuations in the biomass and species composition in Batata Lake relate principally to the hydrological and hydrographic cycle, rather than to any marked cyclicity of heat exchanges that characterise lakes of higher lati-

hydr3729.tex; 24/06/1997; 9:45; v.7; p.9

178 tudes. The importance of fluvial stage to the dynamics of flood-plain ecosystems (Sioli, 1984; Neiff, 1990; Garc´ıa de Emiliani, 1993) and the hydrological pulsing of their structure, metabolism and productivity (Junk et al., 1989; Esteves et al., 1994) is, of course, wellunderstood. The stage cycle imparts an overriding seasonality to the environmental conditions that is quite independent of the annual cycle of temperature and solar irradiance that regulates the thermal structure of lakes at high latitudes. From information collected during the intensive studies (Huszar, 1994; Melo, 1996), it is clear that this tropical lake experiences a typical circadian cycle of diurnal warming and nocturnal cooling, with appropriate changes in mixed-layer depth (for details see, especially, Imberger, 1985; Henry, 1995). The depth of mixing through convectional cooling is predictable. In both shallow and deep lakes, the depth of mixing shrinks towards the surface by day and then, during the afternoon, expands again. In the deep lake, mixing events contribute to the formation of a metalimnetic layer corresponding to the base of the convectional mixing over periods of several consecutive days. In the shallow lake, the mixing events reach to the bottom and turn over the entire water volume, on most days and sometimes continuously. The switch between the lake remaining stratified at night and being convectionally overturned occurs when the maximum depth of the water is some 4–5 m. The precise point on any given occasion has been resolved (Huszar, 1994) by reference to calculations involving Wedderburn Number (Imberger & Hamblin, 1982). When the density gradient between 4 and 5 m beneath the surface remains between 0.1 and 0.4 kg m 3 m 1 (corresponding to a temperature gradient of 1–2  C m 1 within the range 27–30  C), the structure is resistant to windspeeds of