Hydrobiologia 262: 31-42, 1993. 0 1993 Kluwer Academic Publishers. Printed in Belgium.

31

Effects of periodic flooding on the water chemistry and primary production of the Mapire systems (Venezuela) Teresa Vegas-Vilarrubia” & Rafael Herrera’ ’ Centro de Ecologia y Ciencias Ambientales, Instituto Venezolano de Investigaciones Cient$cas, Aptdo, 21827, Caracas 1020 A, Venezuela; “present address: GEOHIDRA C.A.: P.O. Box 47851, Caracas 1040, Venezuela Received 25 September 1990; in revised form 22 October 1992; accepted 10 November

1992

Key words: Ecotone, tropical floodplain production, nutrient pulses

water chemistry,

lakes, lentic/lotic

alternation,

primary

Abstract The Mapire river mouth forms a complex floodplain system, where the river behaves as a river during the dry season, but changes to a transient lake which partially covers the inundation forest during the rainy season. Thus, we expected changes in water chemistry and a gradual increase of primary production during high waters. The system was sampled monthly for one year; two floodplain lakes were also studied for comparative purposes. Variations in the concentration of macro- and micronutrients occurred in a pulse-like manner and seemed to relate to mechanisms at work in the transient lake. Dissolved oxygen showed a stratification with low values at the bottom, but never reached anoxia. Net and gross primary production and respiration did not show any clear spatial pattern, reflecting a mosaic of different biochemical states within the transient lake. Heterotrophy tended to prevail in the transient lake, while autotrophy dominated floodplain lakes.

Introduction Most tropical rivers present strong periodical waterlevel fluctuations. They have large floodplains, where small lake-like water-bodies associated with the main river occur (Hasler, 1975; Junk, 1977, 1980; Welcome, 1979; Wissmar et al., 1981; Sioli, 1984; Hamilton & Lewis, 1987; Lewis, 1988; Whitton et al., 1988). Floodplain lakes and rivers together form an ecological unit that can be subdivided into subsystems in a more or less arbitrary way, linked by interfaces or ecotones (Junk, 1980). I n such transitional areas, periodic floods determine the alternating presence of

aquatic and terrestrial stages, during which the aquatic system is reduced or completely disappears until the next rainy season (Junk, 1973, 1980; Sioli, 1984; Howard-Williams, 1985). In South America, seasonal cycles in water chemistry, biotic communities and metabolism of floodplain lakes have been well documented for the Amazon River system (reviewed in Sioli, 1984; Richey, 1980; Wissmar et al., 1981) and, to a lesser extent, for the Orinoco River system (Blanco & Sanchez, 1984, 1986; Vasquez, 1984; Hamilton & Lewis, 1988). Both rivers have low nutrient levels (Lewis & Weibezahn, 1981; Sioli, 1984) and primary production (Wissmar et al.,

32

1981; Lewis, 1988) in their main channel, as compared with values commonly encountered in standing freshwaters. This fact emphasizes the role of floodplain waterbodies as sources of an inoculum for algal growth in rivers (Rai & Hill, 1984; Lewis, 1988) and of nutrients, released from the flooded area. During the high-water season, allochtonous inputs to the floodplain waterbodies in the form of leacheates, dead leaves and woody debris might be important nutrient sources, since land-water interactions are intense. However, only few studies concern the amount of nutrients released from vegetation in these aquatic environments, and particularly of their effect on water chemistry and primary production (Irmler & Furch, 1980; Wissmar et al., 1981; Hamilton & Lewis, 1987; Furch, 1988). Knowledge of the way in which they process and transform dissolved nutrients is still incipient. Junk (1980) and Howard-Williams (1985) focus on the importance of hydrology with regard to patterns of residence time, throughflow and mass flow of nutrients in such systems. Therefore, a good understanding of their hydrological regime is also needed. The Mapire River mouth forms a complex floodplain system where the river behaves as a river during the dry season, but changes to a transient lake which partially covers the inundation forest during the rainy season. More details on its hydrological behaviour are provided by VegasVilarrhbia & Herrera (1993). These particular conditions of the Mapire River mouth during high-water lead us to expect some important changes in its biogeochemistry. In the present paper, we investigate the impact of forest flooding on the macro- and micronutrient concentrations in the transient lake, and the effect of temporal impoundment of the Mapire River on its level of autothrophy.

Description

of the study area

A hydrochemical, vegetational and geomorphological analysis of the study area is found in Vegas-Vilarrtibia & Herrera (1993), Rosales (1989), and in Carb6n (1991). But some details

on the two floodplain lakes situated on the margins of the Mapire River are necessary here. We called these lakes CHEN and MEAN for research purposes, since no local names exist (Fig. 1, in Vegas-Vilarrtibia & Herrera, 1992). CHEN is located in the proximity of the junction of the Mapire and Orinoco Rivers, is circular with an area of 452 m2, and its depth varies between 0.65 and 3.0 m (January to May). It probably originated from an erosive depression that became filled with water. A dense vegetation ring of Coccoloba obtusifolia surrounds it. The lake is situated in front of a sampling station called PSR. Its morphology suggests that it was once a meander. It is 96 m long, 24 m wide and has an area of 1607 m2. From January to May its depth varies from 0.5 to 2.0 m. While CHEN is connected to the main river, natural channels linking the floodplain lake with the Mapire River are present at MEAN. During highwater, both lakes are covered by flood waters and they become part of the transient lake.

Material

and methods

Sampling stations, frequency and depth were as described in Vegas-Vilarrtibia & Herrera (1992). Values obtained at DES during low-waters were used as a standard against which values obtained at the transient lake and at both floodplain lakes were compared, in order to show variations in water chemistry due to the flood. CHEN and MEAN were sampled separately during the dry season. There, only surface samples were collected. Transparency was measured with a Secchi Disk (S.D.) and pH with a field pH-meter. Dissolved oxygen (D.O.) was determined by the Winkler method (Carpenter, 1965, 1966). Water samples for nutrient analysis were filtered through washed fiber GFjC Whatman filters, fixed with HgCl, and stored at 4 “C. Nitrates were reduced to nitrites on cadmium reduction columns and determined calorimetrically (Strickland & Parsons, 1965). Dissolved silicates (Si) and orthophosphates (PO: - ) were determined colorimetrically by the methods of Mullin & Riley (1963)

33

c,

-

OnI 6.1

ORI 6.0

ORI 5.6 4.0

ORI 5.9 ORI 5.9

I 2m MAP

TER

MMI

I PSR

I DES

MAP

1. Spatial and temporal distribution of pH values at the transient lake. Horizontal vertical axis: depth in meters. Fig.

and Murphy & Riley (1955), respectively. Subsamples for cation analysis were filtered through glass fiber filters (Whatman GF/C) and acidified to pH ~2; Na and K were measured by atomic absorption spectrophotometry. Primary production and respiration were measured in the transient lake and in the floodplain lakes. Oxygen production and consumption by organisms were measured by incubating clear and dark bottles by the Winkler method (Golterman et al., 1969; Vollenweider, 1969). Samples were incubated at bimonthly intervals. Clean dark and clear bottles were incubated between the surface and S.D. depth in the transient lake, but only at the surface in CHEN and MEAN. Samples were incubated for four hours, beginning at 9:00 a.m.

TER

I PSR

I DES

axis: sampling stations (arbitrary distances),

The arithmetic differences between bottles were taken to be significant only if they were higher than 0.15 O2 mg l- ’ (Vollenweider, 1969). Results of net and gross primary production and respiration are given in mg C m- 3 h- ‘, after converting oxygen production into carbon incorporation (Margalef, 1978). Figures 1, 2, 3, 5 and 6 show the same depth scale in meters (Fig. 1).

Results The Mapive River: Lentic versus lotic conditions

Measured pH values (Fig. 1) were acidic throughout the year, ranging from 6.3 to 3.2. In August

34 and September, pH increased about one unit in the upper layers of DES and PSR, when compared with the other sampling stations. Nutrient concentrations were lower than expected and highly variable in space and time. Nitrate concentrations (NO,-N) are shown in Table 1. During most of the dry season NO,-N remained undetectable, except in January (127 pg l- ‘). Nitrite (NO,-N) was detectable only in August (Fig. 2). PO:--P was higher than 8 pg l- ’ only in November (Fig. 2) varying from 21.8 to 78.3 pg PO:- -P l- ‘. These higher concentrations occurred mainly between 0 and 5 meters. Actually, the presence of N03-N and PO: _ -P in detectable concentrations happened at the end of the rainy season. Dissolved Si values (Fig. 3) showed not noteworthy variations, except in August when a less uniform distribution pattern developed. During the dry season, concentrations increased slightly, with values up to 7 pg l- ‘. D.O. and S.D. values are seen in Fig. 4. Secchi Disk values ranged from 0.34 to 1.32 m in the transient lake, indicating low transparency. D.O. concentrations tended to decrease from MAP towards the deepest zone (DES and PSR) in July and, later, from surface to bottom, giving rise to a vertical stratification. Dissolved oxygen concentrations at the bottom were low, reaching a minimum of 1.5 mg l- ’ (19.6 y0 saturation) in August. In October, D.O. measured at TER and MAP increased and tended to homogeneity across the water column. A weak current, observed at the water surface, probably caused this. Still, some vertical stratification of D.O. persisted downstreams. In November, a stronger water current reinforced the tendency to disrupt stratification. No anoxy was observed during the rainy

al N-NO5

AUG

I 2m

b) P-PO;

I

NOV

I

I

I

PSR

MM1

MAP TER

I

DES

2. Spatial and temporal distribution of NO,-N (pg I- ‘) and PO:- (pg 1-l) at the transient lake. Horizontal axis: sampling stations (arbitrary distances), vertical axis: depth in meters. Fig.

season. Perhaps anoxia appeared locally within the flooded forest or at the sediment-water interphase. During the dry season D.O. concentrations were uniform (9-10.8 mg O,l-) always reaching saturation values. Concentrations of Ca and Mg (Vegas-Vilarrubia & Herrera, 1992) were rather low. K and Na concentrations (Figs 6,7) were more or less uniform from June to August. K concentrations increased slightly in the upper layers (O-5 m) of DES, PSR in September. In October this tendency continued, leading to the formation of

Table 1. Physical and chemical variables of DES during dry season. Results obtained for the transient lake (rainy season) and for floodplain lakes. CHEN and MEAN are compared against values presented here.

Month

02 m 1-l

PH

PO:--P !a 1-l

N,-N El-’

NO,-K Pgl-’

K mg 1-l

Na mg 1-l

Mg mgl-’

Ca mg 1-l

Jan Feb April May

11.3 10.6 10.8 9.0

5.6 6.8 6.7 6.8

n.d. n.d. n.d. n.d

137.0 n.d. n.d. n.d

n.d. n.d. n.d. n.d.

1.02 0.26 0.47 0.33

0.70 1.02 0.65 0.86

0.04 0.06 0.04 0.04

0.10 0.28 0.15 0.14

Si

Fig. 3. Spatial and temporal distribution tances), vertical axis: depth in meters.

of Si (mg I- ‘) at the transient lake. Horizontal

axis: sampling stations (arbitrary

MAP TER MAP TER MAP

TER

PSR

PSR

PSR

dis-

DES

DES

DES

OXYGEN

Fig. 4. Spatial and temporal distribution of dissolved 0, (mg 1- ‘) at the transient lake. Horizontal bitrary distances), vertical axis: depth in meters.

axis: sampling stations (ar

36 K’

Fig. 5. Spatial and temporal distribution distances), vertical axis: depth in meters.

of K (mg l- ‘) at the transient lake. Horizontal

patches of higher concentrations of K, then also present at MMI. Within the upper five meters of these stations, the concentrations of K were about 3.7 times higher than in the flowing Mapire water (dry season, Table 1). In November these patches disappeared: K concentrations decreased and tended to become uniform through the whole lake. Na concentrations followed a similar pattern during the same period, also showing patches of higher concentrations and reaching values up to three times those of the Mapire (Table 1). In November, differences in Na concentrations tended to disappear. Ca and Mg concentrations increased notably in August and September over the preceeding months (figs 7 and 8 in VegasVilarrubia & Herrera, 1993) and over flowing Mapire water. Results of Ca and Mg have been discussed by these authors, who attribute them to a temporary inflow of Orinoco water into the Ma-

axis: sampling stations (arbitrary

pire basin during the peak flood. This happens only at DES and PSR, and coincides with higher conductivity and alkalinity and with the presence of suspended solids from the Orinoco. Most chemical differences in Ca, Mg and conductivity disappeared in October, although localized patches of high Ca and Mg existed between 1 and 3 m from the surface in November, once the influence of the Orinoco was no longer observed. From January to May, Ca and Mg concentrations ranged from 0.1 to 0.28 mg I- i and 0.04 to 0.06 mg I- ‘, respectively. Table 3 shows net (NPP), and gross (GPP) primary production, and respiration (RESP) in the transient lake. They were variable among sampling sites and through time. Values for NPP were within the range given by Kaul (1985) for tropical latitudes. In most cases, the final oxygen concentrations of the inoculated clear bottles were higher than those of

Na’

IMAP ITER

IHMl

Fig. 6. Spatial and temporal distribution tances), vertical axis: depth in meters

IPSR IDES IHAP JTER of Na (mg l- ‘) at the transient lake. Horizontal

the dark bottles, but lower than the initial oxygen concentrations. Thus, photosynthetic carbon incorporation may be equaled or surpassed by carbon production due to respiration. RESP values were mostly higher in deeper layers, but NPP and GPP were not. In some cases, GPP values were zero (Table 3) because the 0, content of the dark bottles was higher than that of the corresponding clear bottles. Only in one case did 0, of the dark bottle equal the initial O2 concentrations, the respiration by computation being equal to zero. Errors in the procedure are unlikely, because this method has been standardized in our laboratory for a long time. Occasionally, records have been made of such anomalous and significantly higher 0, contents measured in dark bottles after exposure, compared to clear bottles (Dugdale & Wallace, 1960; Vollenweider, 1960; Hutchinson, 1975) but no conclusive explanation of this anom-

lMM,

IPSRIDES

axis: sampling stations (arbitrary

dis-

aly has been given. Ratios between GPP RESP (P/R), were calculated. Table 3 shows 73% of them indicate dominance of RESP GPP, this situation being stronger at DES PSR.

and that over and

Floodplain lakes

Both floodplain lakes also showed acidic pH values (Table 4). The highest value, 6.3, measured at MEAN in April, coincided with an increase in primary production (Table 5). In nearly all samples NO,-N and NO, -N were below detection limits, except at CHEN in April, were NO,- N was 64.8 pg 1-l and NO,- N 6.9 pg 1-l. PO:- -P was also below 8 pg l- ‘, except at CHEN in May, where higher values of primary production were observed. Si concentrations of

38 DRY

SEASON

’ I

!

1

1

I

RAINY

SEASON

TRANSIENT

Mg concentrations were only slightly variable during the dry season. Table 4 shows that D.O. concentrations in both floodplain lakes tended to decrease from February to May, but never reached values below 50 y. saturation. NPP, GPP and RESP values can be seen in Table 5. At MEAN, NPP was zero in February, increasing later continuously till May. RESP increased from February to April only and decreased later. Consequently P/B < 1 was in February, strongly increasing later to P/B > 1. At CHEN, NPP and RESP augmented from January to May, remaining always at a P/B > 1. In both lakes GPP tended to prevail over RESP. No anomalies in relative 0, contents of the exposed bottles occurred. NPP values were also within the range given by Kaul (1985) for tropical latitudes.

Discussion and conclusions Fig. 7. Hypothetical model of the Mapire transient system. Subsystems = by boxes. Arrows = fluxes of energy and matter. 1. Yield of terrestrial material to the transient lake. 2. Internal cycling of organic matter within the transient lake. 3. Partial return ofmineralized nutrients to the terrestrial system, aquatic phase. 4. Delayed export of material towards the Orinoco. 5. Input of material from the Orinoco into the transient lake. 6. Input of material from the terrestrial system to the Mapire River. 7. Export of material from the Mapire into the Orinoco River. 8. Budget of material to the floodplain lakes from its drainage area. 9. Internal cycling and production of biomass of floodplain lakes. 10. Export of living biomass from the floodplain lakes into the transient lake. 11. Thightness of the spiralling length is gradually shortened changing to ‘in situ’ cycling.

both floodplain lakes were higher than in the Mapire River (Table 4); they tended to increase during the dry season. Major ion concentrations were highly variable throughout the sampling period (Table 4). At CHEN, Na and K concentrations tended to decrease during the dry season, while Ca increased. Mg values were relatively high at the beginning of the dry season, then slowly decreased. At MEAN, Na and K slightly increased at the beginning of the dry season, then decreased. Ca was below detection in January, but gradually increased, reaching values comparable to those of the flowing Mapire.

Variations in concentration of macro- and micronutrients, as a result of release and leaching processes in the flooded forest, were pulse-like in the transient lake. Water chemistry of the latter did not change significantly, when compared to that of the Mapire River water in the dry season. Hamilton & Lewis (1982) arrived at a similar conclusion by studying seasonality in the chemistry of lake Tineo, a floodplain lake of the Orinoco River further upstream. Only pH seemed to increase slightly, due to penetration of Orinoco water. Vegas-Vilarnibia & Herrera (1993) found that the Orinoco water affected conductivity, Calcium, Magnesium, alkalinity and total suspended solids of the transient lake basin during the peak flood. But such an influence was not observed or clearly distinguished in our case. Rather, variations seemed to be related to mechanisms at work on the Mapire floodplain. In principle, we expected higher dissolved PO:--P and NO,-N, as found during the rainy season, by decomposition and mineralization of particulate matter from the flooded forest, enhancing their supply to aquatic organisms. But, we found that macronutrients were generally low. PO:- -P values remained below 8 ,ug l- ‘, increasing in a pulse-like manner

39 Table 2~. NO,-N (pg l- ‘) at the transient lake and the Orinoco River Srainy season), numbers in brackets are the depths where NO,-N values are recorded. n.d. = undetectable. Station

Jul

Aw

Sep

Ott

Nov

DES PSR MM1 TER MAP OR1

n.d. 33.2 (3 m) 5.17 (1 m) 7.45 (4.5 m) n.d.

22.9 (15 m) n.d. n.d. n.d. n.d. n.d.

n.d. n.d. n.d. n.d. n.d. n.d.

n.d. 19.3 (11 m) n.d. 2.85 (0 m) n.d. n.d.

12.6 (11.7 m) n.d. 6.3 (7.7 m) n.d. n.d. 4.9 (0 m)

Table 2b. NO,-N

(pg l- ‘) at DES and OR1 (dry season), n.d. = undetectable.

Station

Jan

Feb

April

May

DES OR1

137.0 14.7

n.d. n.d.

n.d. n.d.

n.d. n.d.

to relative high concentrations at nearly all stations of the transient lake in November. This sudden increase might be due to decomposing dead leaves, suspended in the water. At the end of the rainy season, we observed an event of abcission of leaves from some emerging tree crowns. This represented a prompt input of allochtonous organic matter to the transient lake. Thus, the release of macro- and micronutrients from decomposing leaf tissues may cause a local nutrient increase in water. In fact, patches of higher K and Na concentrations in October, and of PO;--P, Ca and Mg in November may be explained in this way, because they cannot easily be attributed to penetration of Orinoco water into the Mapire River. This latter occurs only in August and September, and is limited to wedges near the mouth; it is unlikely that its effect could still be perceived over a month later. NO,-N and NO,-N increases were also pulse-like. NO,-N was only detectable at the peak flood in August, when D.O. stratification was stronger, with low values in deeper water. NO,-N occurred in measurable concentrations throughout the rainy season at different places. Release of macro- and micronutrients by decomposing leaves of Amazonian inundation forest (Franken, 1979; Irmler & Furch, 1980) and their effect on the chemical composition of the water (Furch et al., 1988) has been well documented. Experiments carried out by Furch et al.

(1988) on P seu d ob om bax mumguba, a typical tree of the varzea forest, show that dissolved K, Mg, Ca and PO:--P are among the major ions released to the water at different stages of leaf decomposition, while NO,-N and NO,-N releases are moderate. According to the above - cited works, tree species differ in both element’s content of their leaves (Klinge et al., 1983) and their decomposition behaviour, thus the release of nutrients of the inundation forest bordering the Mapire River mouth may be qualitatively and quantitatively different. The rather poor content of dissolved bioelements of the transient lake suggests that the inundation forest does not suffer considerable losses of dissolved nutrients during the intense, long-term flood. Nutrient export in the particulate phase might not represent a significant loss to the inundation forest either, since the latter persists apparently without damage year by year. In such severely flooded ecotones, mechanisms to avoid losses are likely to be present. In rainforests, Small (1972) and Vitousek (1982, 1984) reported evidence of nutrient retranslocation from senescent to younger leaves before abcission. Nutrient translocation has been proposed as an important ecological mechanism in high rain environments, because it partially reduces nutrient loss by rain (Marin & Medina, 1981). In the case of periodic floods, re-translocation of nutrients could also be

40 a successful mechanism to avoid losses from the terrestrial to the aquatic system through the decomposition of leaves. Other ecological mechanisms may also control nutrient losses from these systems during the emergent phase, like slow rates of decomposition (Irmler & Furch, 1980), and mycorrhizal fungi reabsorbing nutrients (Herrera et al., 1978). Floodplain lakes (CHEN and MEAN) show greater variation in water-chemistry. This could relate to concentration or dilution of elements by evaporation and supply of water to activity of aquatic organisms. Planktonic algal blooms have been observed in both lakes in April and May, when their water volume was minimum. However, their water composition did not significantly vary, if compared to that of the Mapire. A clear effect of prevention of discharge of the Mapire River into the Orinoco by natural damming is the vertical stratification of some variables (Vegas-Vilarrubia & Herrera, 1993) D.O. concentration were clearly clinograde at peak flood. This was due to oxygen demand and may be an indication of an active heterotrophic metabolism within the transient lake. No anoxic conditions near the bottom were observed. In Amazonian floodplain lakes, clinograde distributions of D.O. and anoxic conditions are related with depth (5 m). In these lakes, nutrient regeneration from sediments to the watercolumn seems to depend on 0, shortage in deeper layers (Rai & Hill, 1984). But in our case, the presence of D.O. close to the bottom probably lowers the transfer of redox-dependent nutrients from the sediment to the water (Mortimer, 1971), although the transient lake is 15 m deep. Neither CHEN nor MEAN presented anoxia because of their mixing by winds. Other tropical shallow floodplain lakes also show oxic conditions throughout the year (Rai & Hill, 1984; H amilton & Lewis, 1987). Dissolved oxygen in aquatic systems is related to the balance between autotrophic and heterotrophic activity. NPP, GPP and RESP were quite variable. This is not surprising, because the metabolism of the submerged terrestrial system overlaps with that of the aquatic system, and aquatic primary producers might profit from

pulse-like nutrient releases in the same way. These situations promote a mosaic of biogeochemical states within the transient lake. Thus, NPP, GPP and RESP values can only be interpreted as a result of biological activity of communities coexisting at a determined place and time. The ratio P/R shows that heterotrophic activity tends to prevail at the transient lake. The input of allochtonous organic matter from the inundation forest surely favours bacterial activity, and the presence of detritivorous organisms on suspended particles. Schmidt (1962, 1970), Rai (1979) and Rai & Hill (1981 a,b., 1982) studied heterotrophic bacterial activity in floodplain lakes of the Brazilian Amazon, underscoring its importance in nutrient mineralization. Floodplain lakes CHEN and MEAN offered a different picture, as primary production prevailed over respiration in nearly all samples. Environmental conditions favour the growth of phytoplankton, since planktonic blooms of Chlorophytes at MEAN and of Cyanophytes at CHEN have been observed. We estimate that the input of organic matter by the surrounding vegetation is reduced, at that time. Consequently, in these lakes P/R was > 1 during the dry season, although RESP was not negligible. Sudden, pulse-like increases of macro- and micronutrients as well as quite variable primary production and respiration might be significant indicators of events occurring in this type of transient systems. For these reasons, the search for regularities in such highly variable environments represents a first step to determine the mechanisms at work. For a better understanding of the Mapire River system as an example of an aquatic/ terrestrial and water/water ecotone, we developed a simple conceptual model. It subdivides the ecotone in subsystems representing functional units and their likely coupling by fluxes of energy and matter. During the rainy season, flow restrictions and the formation of the transient lake implied a delay in horizontal, unidirectional transport and export of materials. The flooded terrestrial system yields organic matter to the lake, generating chemical energy by degradation and mineralization. The temporary stagnation of water enhances

41 the possibility of nutrients to be recycled at the same place and diminishes their loss to the Orinoco. In terms of nutrient spiralling (Webster & Patten, 1979; Newbold et al., 1981, 1983), this means that the ‘thigtness’ of the spiral length, which is associated with the ability of streams to utilize nutrients, is gradually shortened, once the river lowers its flow. The loss of the horizontal component leads spiralling to change temporarily to ‘in situ’ cycling with a strong vertical component. Fluxes of energy and matter among terrestrial and aquatic subsystems may be favoured by the transient lake, allows a more intense physical contact among them, and acts as an interface. During the dry season, floodwaters return to the river channel, and water starts to flow again. Floodplain lakes, now disconnected from the main river, store the chemical energy invested during the rainy season as living biomass. This biomass is exported to the transient lake at the beginning of the next rainy season. The physical contact among subsystems is reduced with the disappearence of the transient lake, and the river flow imposes the direction of flux of energy and matter among them. This model is inspired by the interpretationofsimilarphenomenabyotherscientists (Hasler, 1975; Margalef, 1978; Rai & Hill, 1982, 1984; Howard-Williams, 1985). Further elaboration of the model requires more knowledge of the biogeochemical and hydrological processes of each functional unit, and of their couplings .

Acknowledgements We thank Ramon Margalef and Valenti Rull for critical review of the manuscript, and Karin Furch, Wolfgang Junk and Hans Klinge for helpful comments. Remarks of an unknown referee were useful in improving the manuscript. Our gratitude also to Jose Luis Valles and to Alejandro for their assistance during field trips, to Saul Flores, Clara Gomez and Gladys Escalante for technical work, and to Berta Sanchez for typing. Prof. Rafael Herrera and Dr H. Dumont improved the English. This work received partial

support from CONICIT and was carried out as part of the Mapire Project in the Laboratory of Soil Ecology, IVIC.

References Blanco, L. J. & L. Sanchez, 1984. Contribution al estudio taxonbmico de las diatomeas de1 Orinoco Medio, Bajo Caroni y algunas lagunas de inundation (Venezuela). Mem. Sot. Cienc. Nat. La Salle 121: 35-94. Blanco, L. J. & L. Sanchez, 1986. Contribution al estudio taxonbmico de las Euglenophyta, Cyanophyta, Chlorophyta y Chromophyta de1 Orinoco Medio, Bajo Carom, Uracoa y algunas lagunas de inundation (Venezuela). Mem. Sot. Cienc. Nat. La Salle 125-126: 7-48. Carbon M. J., 1991. Gemorfologia y estratigrafia de la cuenca de1 Rio Mapire y zonas adyacentes (sur de1 Edo. Anzokegui). CEA., IVIC. Caracas, 186 pp. Carpenter, J. H., 1965. The accuracy of the Winkler method for dissolved oxygen analysis. Limnol. Oceanogr. 10: 135140. Carpenter, J. H., 1966. New measurements of oxygen solubility in pure and natural water. Limnol. Oceanogr. 11: 264-277. Franken, M., 1979. Major nutrient and energy contents of litter fall of a riverine forest of Central Amazonia. Tropical Ecology 20: 211-235. Furch, K., W. Junk & Z. E. S. Campos, 1988. Release of major ions and nutrients by decomposing leaves of Pseudobombax munguba, a common tree in the Amazonian floodplain. Verh. int. Ver. Limnol. 23: 642-646. Golterman, H. L., R. S. Clymo & M. A. M. Ohnstad, 1969. Methods for chemical analysis of freshwaters. I.B.P. Handbook. 8th edn. Blackwell, Oxford, 172 pp. Hamilton, S. K. & Jr W. M. Lewis, 1987. Causes of seasonahty in the chemistry of a lake on the Orinoco river floodplain, Venezuela. Limnol. Oceanogr. 32: 1277-1290. Hasler, A. D., 1975. Coupling Land and Water systems. Springer Berlin, 305 pp. Herrera, R., T. Mtrida, N. Stark & C. F. Jordan, 1978. Direct phosporus transfer from leaf litter to roots. Naturwissenschaften 65: 208. Howard-Williams, C., 1985. Cycling and retention of nitrogen and phosphorus in wetlands: a theoretical and applied perspective. Freshwat. Biol. 15: 391-431. Howard-Williams, C. & W. J. Junk, 1977. The chemical composition of Central Amazonian aquatic macrophytes with special reference to their role in the ecosystem. Arch. Hydrobiol. 79: 446-464. Kaul, V., 1985. Primary productivity of inland aquatic ecosystems under varying climate- a review. Tropical Ecology 26: 164-178. Irmler, U., 1979. Considerations on structure and function of the ‘Central Amazonian Inundation Forest Ecosystem’ with

42 particular emphasis on selected soil animals. Oecologia 43: 1-18. Klinge, H., K. Furch, E. Harms & J. Revilla, 1983. Foliar nutrient level of native tree species from Central Amazonia. 1. Inundation forest. Amazoniana 8: 19-45. Junk, J. W. 1973. Faunistisch -okologische Untersuchungen als Mijglichkeit der Definitionen von Lebensraumen, dargestellt and Ueberschwemmungsgebieten. Amazoniana 4: 263-211. Junk, W. J., 1980. Areas inundaveis- Un desafio para a Limnologia. Acta Amazonica 10: 775-795. Junk, W. J. & C. Howard-Williams, 1984. Ecology of the aquatic macrophytes in Amazonia. In H. Sioli (ed.), The Amazon, Dr W. Junk, Dordrecht, 763 pp. Lewis, W. M. Jr., 1988. Primary production in the Orinoco River. Ecology 69: 679-692. Lewis, W. M. Jr. & F. Weibezahn, 1981. The chemistry and phytoplankton of the Orinoco and Caroni Rivers, Venezuela. Arch. Hydrobiol 91: 521-528. Mann, K. H., 1975, Patterns of energy flow. In B. A. Whitton (ed.), River Ecology, Blackwell Scientific, Oxford, 725 pp. Margalef, R., 1978. Ecologia, Omega, Barcelona, 1010 pp. Marin, D. & E. Medina, 1981. Duracibn foliar, contenido de nutrientes y esclerofilia en &rboles de un bosque muy seco tropical. Acta Cient. Venez. 32: 505-541. Mortimer, C. H., 1971. Chemical exchange between sediments and water in the great lakes-speculations on probable regulatory mechanisms. Limnol. Oceanogr. 16: 387-404. Mullin, J. B. & J. P. Riley, 1955. The calorimetric determination of silicate with special reference to sea and natural waters. Analyt. Chem. Acta 12: 162-176. Murphy, J. & J. P. Riley, 1963. A modified single solution method for the determination of phosphate in natural waters. Analyt. Chem Acta 27: 31-36. Newbold, J. D., R. V. Elwood, R. V. O’Neill & W. Van Winckle, 1981. Measuring nutrient spiralling in streams. Can. J. Fish. aquat. Sci. 38: 860-863. Newbold, J. D., J. W., Elwood, R. V. 0’ Neil1 & A. L. Sheldon, 1983. Phosphorus dynamics in a woodland stream ecosystem: study of nutrient spiralling. Ecology 63: 12491265. Rai, H., 1979. Microbiology of Central Amazon Lakes. Amazoniana 6: 583-599. Rai, H. & G. Hill, 1981a. Bacterial biodynamics in lago Tupe, a Central Amazonian black water ‘Ria Lake’. Arch. Hydrobiol. Suppl. 58: 420-468. Rai, H. & G. Hill, 1981b. Observations on heterotrophic activity in Lago Janauari: a ria/varzea lake of Central Amazon. Verh. int. Ver. Limnol. 21: 715-720.

Rai, H. & G. Hill, 1982. Establishing the pattern of heterotrophic bacterial activity in the three Central Amazonian lakes. Hydrobiologia 86: 121-126. Rai, H. & G. Hill, 1984. Primary production in the Amazonian aquatic ecosystem. In H. Sioli (ed), The Amazon, Dr W. Junk Publishers, 763 pp. Rosales, J., 1989. Analisis floristico-estructural y algunas relaciones ecologicas en un bosque inundable en la boca de1 Rio Mapire (Edo. Anzoategui). CEA, IVIC, Caracas, 233 pp. Schmidt, G., 1962, Vertical distribution of bacteria and algae in a tropical lake. Int. Revue ges. Hydrobiol. 54: 791-797. Schmidt, G., 1970. Numbers of bacteria and algae and their interrelations in some Amazonian waters. Amazoniana 2, 4: 393-400. Sioli, H., 1984. The Amazon, Dr W. Junk, Dordrecht, 763 pp. Small, E., 1972. Photosynthetic rates in relation to nitrogen recycling as an adaptation to nutrient deficiency in peat bogs plants. Can. J. Bot. 50: 2227-2233. Strickland, J. D. H. & T. R. Parsons, 1965. A manual of sea water analysis. J. Fish Res. Bd Can. 125: 202 pp. Vasquez, E., 1984. Estudio de las comunidades de rotiferos de1 Orinoco Medio, Bajo Caroni y algunas lagunas de inundacion (Venezuela). Mem. Sot. Cienc. Nat. La Salle 121: 95-108. Vegas-Vilarrubia, T. & R. Herrera, 1993. Seasonal alternation of lentic/lotic conditions in the Mapire system, a tropical floodplain in Venezuela. Hydrobiologia 262: 43-55. Vitousek, P. M., 1982. Nutrient cycling and nutrient use efficiency. Am. Nat. 119: 553-572. Vitousek, P. M., 1984. Litterfall, nutrient cycling and nutrient limitation in tropical forest. Ecology 65: 285-298. Vollenweider, R. A., 1969. A manual on methods for measuring primary production in aquatic ecosystems. I.B.P. Handbook 12, Blackwell, Oxford, 213 pp. Webster, J. R. & B. C. Patten, 1979. Effects on watershed perturbation on stream potassium and calcium dynamics. Ecological Monographs 19: 5 l-72. Welcome, R. L., 1979. Fisheries ecology of floodplain rivers. Longman, London, 3 17 pp. Whitton, B. A., A. Azis, P. Francis, J. A. Rother, J. W. Simon & Z. N. Tahmida, 1988. Ecology of deep water rice-fields in Bangladesh. 1. Physical and chemical environment. Hydrobiologia 169: 3-67. Wissmar, R. C., J. E. Richey, R. F. Stallard & J. M. Edmond, 1981. Plankton metabolism and carbon processes in the Amazon river, its tributaries and floodplain waters, PeruBrazil, May, June 1977. Ecology 62: 1622-1633.