Seasonal Distribution of Nutrients and Primary Productivity on the Eastern Continental Shelf of Venezuela as Influenced by the Orinoco River

Nitrogenous nutrients, dissolved silicate, and salinity were measured in surface waters and shallow hydrocasts along similar cruise tracks during the spring (dry season) and fall (wet season) of 1988. Both cruises transected the eastern Caribbean, transited the Gulf of Paria, ran parallel to the Orinoco Delta and into the main channel of the Orinoco River. Trends in primary productivity were also measured by daily carbon 14 incubations. In both seasons, samples covered the range from highly oligotrophic and transparent to highly productive and rich in biogenic and abiogenic particulate matter. Most of the Orinoco outflow appears to turn N to NW and remains in shallow waters off Venezuela and surrounding Trinidad, permitting benthic regeneration of river-borne nutrients. However, the role of the Orinoco and associated low-salinity coastal waters in fertilizing large areas of the eastern Caribbean basin, as suggested by satellite imagery, can be approximated crudely from the nutrient composition at Boca de Dragon, which is representative of the nutrient status of these waters as they flow into deeper Caribbean waters. Additional nutrients may be supplied to the area primarily from Amazon-derived water entering the Caribbean Basin further north, with some coastal upwelling along the continental shelf in the dry season.


INTRODUCTION
Coastal oceanic water masses, especially those located in the tropics, are among the most fertile and productive. This biological and organic richness may result from high loads of nitrogen, phosphorus and silicon supplied by hydrodynamic processes resulting from continental drainage, or from coastal upwelling processes during seasons of winds favoring this process. Within the context of coastal bioecosystems, estuarine-deltaic zones are notable as transition environments between the ocean and the continent. The Orinoco River system in Venezuela provides one of the largest-scale examples of these interactions. It has a major influence on the hydrographic and biological conditions of the estuarine system of the Gulf of Paria and the Caribbean Sea.
The Orinoco River begins near Brazil, at the junction of Sierra Parima and Sierra Unturan, and flows over an area of 2560 km 2 , of which 1670 km 2 are navigable. It irrigates a flood plain of 981,000 km 2 in Venezuela, Colombia, and Brazil, is the only South American river with a true delta, and has an ocean front of 300 km. Its principal affiuent is the Caroni River, with a brown-ochre color, which is mixed with the waters of the Orinoco, which are yellow. The Orinoco flow is turbulent and contains a large amount of suspended clay. These Orinoco-Caroni waters are rich in organic and inorganic matter which is transported to the Atlantic Ocean and the Caribbean Sea through the estuarine-deltaic system. The Orinoco River contributes a large amount of suspended sediments (86.3 x 10 6 tons yr -I), with an average water flow I "".' "". May to November. Evaporation is highest in February (129 mm) and lowest in November (76 mm), while precipitation is highest in August (127 mm) and lowest in April (19 mm). Thus the timing of the two cruises reported here nearly coincided with the two extremes. Westerly winds are prevalent, with an eastern component during most of the year. A semi-diurnal tide regime dominates in the estuarine-deltaic region [Salazar, 1989].
Here we summarize the"saline stratification, nutrient distributions, and primary productivity measurements from the spring (April 1988) and fall (September 1988) cruises. We focus on longitudinal transects NE-SW and E-W from the mouth of the Orinoco. This paper focuses on the near-shore data (all data S of =l1°N), as studies in that area were the most intensive and the riverine impact is strongest. For effects in the Caribbean N of 11 °N, see the accompanying paper [Zika et al., this issue]. In order to better understand the hydrochemical mechanisms of the delta, conservative and nonconservative processes resulting from the runoff of the Orinoco during high and low flow are considered.

Source of Samples
Water samples were collected in the estuarine-deltaic system of the Orinoco River, Gulf of Paria, and adjacent areas (Figure 1) during research cruises CI-8805 in the spring (April1988) and fall CI-8816 (September 1988) on board the RIV Columbus Iselin. Seawater samples were taken using a rosette sampler (General Oceanics, Miami, Florida) consisting of twelve 5-L teflon-lined polyvinyl chloride Niskin bottles. The rosette was fitted with a Neil Brown conductivity, temperature, and depth (CTD) probe. Additional surface (2m) samples were collected from the through-the-hull teflon bow pumping system.

Nutrient Analyses
Water samples were collected in 60-mL Nalgene bottles and refrigerated until analyzed. Samples were analyzed within 24 hours after collection. Turbid samples were filtered using 0.45 /Lm pore-size Acrodisc disposable filter assemblies (Gelman Sciences, Ann Arbor, Michigan).

Primary Productivity Measurements
Surface water ( "-"0-2 m) samples were collected in BOD bottles to which 0.95 mL of NaH 14 C0 3 (2.5 /LCi mL-1 , 42.37 nmol mL -I, Amersham Corp., Arlington Heights, Illinois) was added. The samples were irradiated in full sunlight in a flowing seawater bath for 3 hours (dark controls were covered with aluminum foil). After irradiation the samples were filtered through GF/F glass fiber filters (Whatman International Ltd., Maidstone, England), rinsed with 50 mL of 1 N HCl and stored for further analysis. Integrated photosynthetically active radiation (PAR) was measured with a LI-COR (model 1800) calibrated spectroradiometer. Radioactivity was determined with a Beckman LS 3801 liquid scintillation counter (Beckman Instruments, Fullerton, California). Primary productivity values were calculated as JLgC fixed per m 3 h-I.

Spring and Fall Salinity Distributions
The spring-fall salinity distributions (Figures 2-5) are dramatically different. In spring there was a sharp freshwater-seawater interface within the navigation channel and upstream of Pt. Barima, so that the NE-SW transect ( Figure  2) never encountered pure river water (S min = 8 ppt at R5 at ebb tide). The E-W transect (Figure 4), as close inshore as feasible navigationally, also encountered only relatively saline waters (S > 30 ppt) because the saltwater-freshwater gradient was centered close inshore in the nonnavigable waters of the Orinoco's extensive deltaic system and distributaries in spring.
In fall, both sections were within the core of the near-field freshwater-saltwater mixing zone (Figures 3, 5). The E-W transect and the Gulf of Paria was roughly 50% saltwater at that time, while the NE-SW transect encountered S = 0 ppt water from surface to bottom at R5. The surface distribution of salt (not shown) on these two cruises thus resembled that given by Rodriguez [1975] for April, but in September the E-W transect and Gulf of Paria waters were already almost twice as saline as those reported for August, suggesting that the peak flow was already well past and the gradual advection and mixing of this large input with seawater were underway. Where there was sufficient water for navigation by the R/V Iselin, the delta region of the Orinoco River exhibited a typical salt wedge circulation which governs hydrographic and chemical conditions in deeper waters. Thus the spring versus fall effects of the riverine inflow are limited primarily to a shallow surface layer and are hardly detectable even at 50 m. This factor can be seen clearly by comparing salt, silica, and nitrate seasonal differences along both transects: NE-SW (Figure 2, spring; Figure 3, fall) and E-W (Figure 4, spring; Figure 5, fall). The tendency to upwelling in the spring and its relaxation in the fall are clearly indicated in the NE-SW (onshore-offshore) sections by the sloping and horizontal (respectively) isolines for S = 36 ppt, and silicate or nitrate 1 JLM. The E-W transect comparison most clearly shows the much larger extent of freshwater influence: the spring top-bottom contrast is "-"31-36.5 ppt, while in the fall the range is = 13-36.6 ppt.
The sloping isolines indicate the dry season upwelling of intermediate water from the southeastern Atlantic in the platform at Boca Grande. Water of salinity 35.5 ppt at 30m (R1) ascends to the surface layers (2-10 m), pushing the freshwater mass into the river (Figure 2) from the surface (R3) with changes in salinity, delineating this area of the estuary from the salt wedge towards the bottom at R5, where the wedge's apex is formed. The mixing zone was between R5 and R6 near the platform at Boca Grande. At a depth of 90 to 140 m there was a cold, deep-water mass from the southeastern Atlantic which was higher in salinity (36.8-37.15 ppt) between R1 and R2, rich in nitrogenous compounds (nitrite, 0.18-0.16 p.M; nitrate, 0.47--6.2 JLM) and low in silicate (0.12-0.35 JLM}. Thus, during the dry season, when the Orinoco has its lowest flow, its influence is limited to a narrow fringe stretching along the estuarine-deltaic area (Figures 2 and 3).

Nutrients and Primary Productivity
The vertical nutrient distributions are also shown in Fig Table 1 presents a highly condensed summary of the data for both seasons in the vicinity of the Orinoco River.
Given the restrictions imposed in the spring by navigation requirements and the strong tendency of the river water to spread over the salt water in a thin difficult-to-sample sheet in both seasons, these data are obviously too sparse and unrepresentative of the river-delta-coastal system as a whole to permit a detailed characterization of the region in either season, or a detailed interseasonal comparison. The purpose of gathering these data was rather to provide ancillary information for the more detailed process-oriented studies reported in other papers in this Special Section.
Nevertheless, a few issues of general interest to the biogeochemistry of this fascinating but relatively littlestudied region can be addressed with this data set. They are discussed below. Some of these same issues are also addressed from different perspectives in some of the accompanying papers (e.g., the Si budget and Ra-Si relationships [Moore and Todd, this issue]; primary productivity, nutrients, and specific pigment distributions [Bidigare et al., this issue]).

Silicate and (Nitrate Plus Nitrite) in Spring and Fall
Coastal nitrogen cycling is intense and complex [Carpenter and Capone, 1983]. The available data provide strong qualitative confirmation of the expected intense biological cycling in this region, as discussed previously [D'Armas, 1986; Salazar, 1989] for November conditions in the water column of the delta region. Sedimentary conditions there have been discussed by Milliman et al. [1982], Eisma et al. [1978], and Kennicut et al. [1987] and conditions in the Gulf of Paria have been described by Bonilla and co-workers, as referenced in the introduction.
In addition to salt, other semiconservative tracers are useful in estuarine studies, most classically dissolved silicate   [ Burton et al., 1970;Liss, 1976], as exemplified recently for the Mackenzie Shelf Estuary by MacDonald et al. [1989]. However, silicate distributions must be interpreted with care because biological uptake and subsequent sinking and redissolution of biogenic opal in the water and on the bottom can lead to seriously nonconservative behavior on spatial and temporal scales that need to be defined empirically in each system [e.g., Edmond et al., 1981]. However, use of Si as a tracer has a history in the greater eastern Caribbean basin with respect to understanding the ...     influence and seasonal variability of the great river systems to the southeast (Amazon and Orinoco) versus rain inputs. Froehlich et al. [1978] first used Si sections to suggest that rivers, not rainfall, were the principal cause of the seasonal surface freshening as far away as Puerto Rico. More re-cently, Moore et al. [1986] used _Ra-Si relationships to construct quantitative, sensitive tracers of the remote riverine inputs to this region. The radiochemistry approach was also applied on CI-8805 and 8816; the combined use of Si and radiochemical data to determine the "estuarine end mem-  ber" by this approach is discussed in their paper [Moore and Todd, this issue]. Hence it is of interest to examine its behavior also in the estuarine region which may alter this water before it reaches the eastern Caribbean. Figure 8 shows silicate versus nitrate plus nitrite plots for both seasons along with the mixing lines expected for spring and fall Orinoco river water plus oligotrophic seawater (nutrient-salinity plots cannot be used in spring, when all the low-salinity samples were acquired in rapid succession from the pumping system and could not be accurately matched with salinities). In spring (solid symbols) the data are from the vicinity of R4-R5, whereas in fall the same [Si] range Was covered over much of the E-W transect outside of the Gulf of Paria. As noted above, the highest Si values in both seasons closely approximate those expected from the riverine data of Lewis and Saunders [1989]. In contrast, in both seasons N0 2 -+ N0 3 -clearly cycles dynamically in waters of lower salinity (higher Sl). The dominant cycling mechairisms appear to differ seasonally, as described below. · Proceeding from high to low Si values in spring, we initially encounter waters with very low N0 2 -+ N0 3 -/Si compared to that expected from the river-seawater mixing (Figure 8). These samples were from within the turbidity maximum, where photosynthetic uptake cannot be the nitrate sink. Rather, the associated high methane values iii some samples (square symbols) strongly suggest that inter" actions With anoxic bottom and pore water environments were intense. Loss of NOi + N0 3 -by denitrification (and possibly Si injection by redissolution) are likely the dominant processes redefining the N0 2 -+ N0 3 -/Si ratios here.
At Intermediate [Si] the data fall on the high NOi + DON in the riverine end-member. Displacement of NH 4 + I DON bound to clays by seawater cations in this zone of intense mixing would provide a mechanism to free new substrate for subsequent oxidation to nitrite by bacteria [Carpenter and Capone, 1983], while the high turbidity would minimize light inhibition [Olson, 1981;Vanzella et al., 1989]. Finally, at <20 JLM Si this high N0 2 -+ N0 3 -/Si water appears to mix conservatively with oligotrophic seawater, perhaps with some photosynthetic uptake of N at the very lowest [Si]. Thus these data show that in spring, nonconservative behaviors indicating biogeochemical cycling and input of N to coastal waters were intense in this very active zone, generating both relatively N-poor and N-rich waters (N/Si ratios =0.1-5 of the riverine ratio). However, our sampling was far too sparse and unrepresentative to permit estimating the relative amounts of these two types of water input to the coastal zone over any significant area, and the absence of ammonia and DON data prevents a comprehensive mass balance approach.
In fall (Figure 8, open symbols) the much larger river flow pushed high-Si water out well beyond the core of the turbidity maximum; waters with strongly enriched NOi + N0 3 -/Si ratio were absent. The most Si-rich waters were also higher in N0 2 -+ N0 3 -than expected from the data of Lewis and Saunders [1989] but in the expected ratio. The fall data are plotted against salinity in Figure 9, revealing estuarine Si consumption and strong N0 2 -+ N0 3 -uptake, especially near =7 ppt salinity. This presumably reflects photosynthetic uptake by silicoflagellates (see Edmond et al. [1981] concerning this same situation in the Amazon plume).
At higher salinities, Si is more nearly conservative and N + N is exhausted. These trends suggest remineralization of Si and continued uptake of N + N by nonsiliceous primary producers at higher salinity. However, in both seasons these interpretations must be tempered by the realization that the system was probably not at steady state, that significant vertical gradients which may have originated by nonconservative processes are ignored, and that the "riverine" (near Barrancas, at the head of the deltaic region) endmember data of Lewis and Saunders [1989] or our samples at Boca Grande (R4-R5) may not typify the myriad other unsampled distributaries to the NE and entering the Gulf of Paria. ·

Gulf of Paria and Inputs to the Caribbean
Although some Orinoco and coastally trapped Amazon water enters the Caribbean by way of Galleons Passage N of Trinidad, efflux out of the Boca de Dragon is thought to represent a major component of the estuarine input to the Caribbean, as shown strikingly by satellite photographs [Miiller-Karger et al., 1990].
The gyre circulation in this Gulf [Bonilla, 1977;D'Armas, 1986] tends to homogenize its waters, including the inputs of the northernmost distributaries discharging into the Gulf itself. Hence it is especially interesting to examine the behavior of Si on this scale. Taking our riverine end-member values and R1 as a seaward end-member for Atlantic water entering the Caribbean in the vicinity of the Grenada passage, conservative mixing of surface waters in spring predicts = 18 JJ.M silicate at P1; <7 JJ.M was present. Thus at low discharge, the residence time of water in the Gulf is apparently long enough that substantial biological removal occurs.
In contrast, in fall Si appears to be conservative along the coast and in the Gulf of Paria (Figure 10 inset), despite evidence for localized Si removal (Figure 9). Presumably, either this Si was regenerated and re-entrained in the flow (perhaps in the course of turbulent mixing in the Boca de Serpente), or the fraction of the total flow thus affected was minor. This mixing line thus defines the Gulf of Paria source water to the Caribbean (at this time) as having an endmember value of = 120 JLM Si representing river water (Orinoco plus coastally trapped Amazon flows) little modified by dilution by rainfall or biological removal. This differs from Froehlich et al. [1978] estimate of average freshwater input to the eastern Caribbean (for the entire eastern Caribbean, a significant amount presumably being Amazon water entering with the North Equatorial and Antilles currents through more northern passages) had an apparent endmember composition of=60 JJ.M Si, corresponding mostly to three-component mixing of Amazon water (with nearly the same Si content as Orinoco water, =BO JJ.M), rainfall, and oligotrophic surface seawater. Figure 10 is an Si-salt plot with our end-member mixing line (dashed), that of Froehlich et al. [1978] (heavy solid), and the surface Si data for the entire east. Caribbean cruise track (Figure 1) in fall1988. Much of the data lie near the line of Froehlich et al. [1978], especially if adjusted to assume an end-member with 0.5 J.LM Si at 36.2 ppt rather than the value of 1 at 36.4 they used (light line). However, to the north (up to about B 0 N) many points. lie in between these two or along our mixing lines. If Si is nearly conserved in the E Caribbean, our Si data suggest a modified view of the threecomponent mixing concept to account for the fact that the Gulf of Paria component is apparently little diluted by rain, whereas the "offshore Amazonian" inputs supply raindiluted riverine Si.
Thus our Si data more or less directly demonstrate the existence in fall of two freshwater/Si inputs and show that the source of high Si water is at least partly associated with the Orinoco input through the Gulf of Paria, up to = B 0 N. These findings thus reinforce the conclusion from 228 Ra data by Moore et al. [1986] that two discrete routes of delivery for . o extst [see also Moore and Todd, this Spring elite/ Fall D. . 'b . lstn lllums of Primary Productivity On a h\I'Qcr ~ 1 n.. Amazon Wl · ca e, n,\'ther et al. [1967] suggested that dilutinft '!\.I .\tcr actually depressed productivity off Brazil by ~~ t"' nutrients wh .. suggest· ,~ . • e.reas Muller-Karger et al. [1990] .,. n\1\SSIV ~ 'I' · Orin"""-: · e ertt tzatton of the eastern Caribbean by ......... nutrtcnts A · . ammonia ~ 1 0 · s menttoned previOusly, the absence of nitrogen.,'"~~ ON data do not permit us to evaluate the in fall ~ ~ t~puts to the e~\stern Caribbean, though clearly waters e~~t~: ~assive ni~rnte depletion ( Figure 9) before producth·\t , ulf of ~a. The absence of small primary normally \~t\lues close mshore in either season shows a light ~-.\need coastal productivity, despite the fact that data bv ~tty • nnd salinity stresses must be strong. The from ·.