Factors related to gizzard shad and the threadfin shad occurrence and abundance in Florida lakes

Gizzard shad Dorosoma cepedianum were collected in 23 and threadfin shad D. petenense were collected in 22 of the 60 Florida lakes sampled. Logistic regression equations were 94% effective for predicting gizzard shad occurrence from chlorophyll and lake surface area, and 84% effective for predicting threadfin shad occurrence from lake surface area and lake volume inhabited (PVI). Occurrence of both shad species was related positively to lake size. In lakes where gizzard shad or threadfin shad were collected, shad density and biomass of both shad species were related positively to chlorophyll. Gizzard shad populations were generally vulnerable to predation in lakes, with the per cent of gizzard shad ≤200mm L _T values exceeding 60% with few exceptions. Effects of gizzard shad and threadfin shad on fish community dynamics may be confined to relatively large (>100 ha) and fertile (chlorophyll >20–30μg l⁻¹) Florida lakes.     

Influence of gizzard shad on phytoplankton size and primary productivity in mesocosms and earthen ponds in the southeastern U.S.

Gizzard shad (Dorosoma cepedianum), a filter feeding omnivore, can consume phytoplankton, zooplankton and detritus and is a common prey fish in U.S. water bodies. Because of their feeding habits and abundance, shad have the potential to affect primary productivity (and hence water quality) directly through phytoplankton grazing and indirectly through zooplankton grazing and nutrient recycling. To test the ability of shad to influence primary productivity, we conducted a 16-day enclosure study (in 2.36-m³ mesocosms) and a 3-year whole-pond manipulation in 2–5 ha earthen ponds. In the mesocosm experiment, shad reduced zooplankton density and indirectly enhanced chlorophyll a concentration, primary productivity, and photosynthetic efficiency (assimilation number). While shad did not affect total phytoplankton density in the mesocosms, the density of large phytoplankton was directly reduced with shad. Results from the pond study were not consistent as predicted. There were few changes in the zooplankton and phytoplankton communities in ponds with versus ponds without gizzard shad. One apparent difference from systems in which previous work had been conducted was the presence of high densities of a potential competitor (i.e., larval bluegill) in our ponds. We suggest that the presence of these extremely high larval bluegill densities (20–350 larval bluegill m^–3; 3–700 times higher density than that of larval gizzard shad) led to the lack of differences between ponds with versus ponds without gizzard shad. That is, the influence of gizzard shad on zooplankton or phytoplankton was less than the influence of abundant bluegill larvae. Differences in systems across regions must be incorporated into our understanding of factors affecting trophic interactions in aquatic systems if we are to be able to manage these systems for both water quality and fisheries.     

Impact of the removal of gizzard shad (Dorosoma cepedianum) on nutrient cycles in Lake Apopka,Florida

1. The St. Johns River Water Management District removed over 5.4 million kg of gizzard shad (Dorosoma cepedianum) from Lake Apopka, FL during 1993–2005, as a means of reducing lake phosphorus and phytoplankton concentrations and improving water clarity. Other steps included reduction of external nutrient inputs and operation of a treatment wetland. We measured nutrient excretion by Lake Apopka gizzard shad to quantify the nutrient effect of this biomanipulation. 2. Both N and P excretion were significantly affected by fish body mass and temperature. Larger fish had lower mass‐specific rates of excretion than smaller fish. 3. High water temperature increased P excretion to a much greater extent than N, resulting in a low N : P of nutrient excretion in midsummer. The N : P of excretion was lower than has been observed in other systems, probably because of higher water temperature. 4. Removal of gizzard shad >200 g prevented the annual release of 45 800 kg N year^?1 (3.46 kg N ha^?1 year^?1) and 7700 kg P year^?1 (0.62 kg P ha^?1 year^?1) on average. The actual impact on the P cycle varied substantially from year to year (range 7900–78 800 kg N year^?1; 1200–14 800 kg P year^?1), primarily because of fluctuations in the catch. 5. On an annual basis, the P directly removed in fish tissues was similar to that removed by the treatment wetland. The P excretion prevented by the removal of fish was approximately 20% of the reduction in external P loading achieved during 1993–2005. 6. In the short term, most of the P demand of planktonic primary producers is met through recycling of P, which greatly exceeds external P loading. Depending on population biomass, phosphorus excretion by the resident gizzard shad population was similar in magnitude to the P release by diffusive flux from the sediments.     

Nutritional importance of detritivory in the growth and condition of gizzard shad in an Ohio reservoir

Synopsis Gizzard shad,Dorosoma cepedianum, in Acton Lake, Ohio, ingested foods of varying nutritional quality during the 1981–1983 growing seasons. Adult (ages 2–4) fish fed on a mixed diet (ORG>30%; C:N<7:1) of zooplankton and organic detritus in early summer, and on detrital materials (ORG<16%; C:N>11:1) during the remainder of the growing season. Age 0 (<35 mm standard length) fish ingested only detrital materials. The nutritional quality (ORG = 10 – 20%; C:N<11:1) of these foods displayed little seasonal variation, but was higher than that of organic detritus taken by adult fish in late summer and autumn. Growth and condition of gizzard shad were poor when the diet consisted of detrital materials; however, age 4 fish (1983) grew rapidly and condition improved when zooplankton were consumed. These results suggest that ingestion of poor-quality detritus can reduce the growth and condition of gizzard shad in Acton Lake, whereas the seasonal inclusion of high-quality zooplankton in the diet can result in rapid growth and improved condition.     

Particulate and filter feeding in threadfin shad, Dorosoma petenense, at different light intensities

Previous workers inferred from stomach analyses that threadfin shad ( Dorosoma petenense ) ate plankton by both filter and particulate feeding. These inferences were confirmed in this study by laboratory experiments in which both types of feeding were observed. Threadfin shad consumed relatively small food particles (≤ 0.39 mm) by filtration, while larger prey (7.5 mm) were eaten individually. The shad were able to filter feed on small foods (brine shrimp nauplii and phytoplankton) at all light intensities from 0 to 9 × 10¹ fL. These data indicate that under natural conditions shad can filter feed at any time of the 24 h period if food conditions are sufficient to trigger feeding. Filter feeding is probably induced by chemoreception rather than vision. The particulate feeding rate decreased as light intensity decreased, reaching a minimum between 9 × 10⁻⁴ and 9 × 10⁻⁵ fL. From these data it is inferred that particulate feeding is a visual process in this species, requiring intensities equivalent to bright moonlight or greater. Filter and particulate feeding abilities allow threadfin shad to consume most of the different types of plankton and to change their diet with seasonal changes in the composition of the plankton.     

Cumulative stress-induced mortality of gizzard shad in a southeastern U.S. reservoir

Synopsis Starvation was apparently responsible for a large die-off of gizzard shad, Dorosoma cepedianum, in several east Tennessee reservoirs during the spring of 1983. Condition indices, calorific equivalents, lipids, and blood parameters of electrofished (control) shad from Watts Bar Reservoir were significantly higher than these parameters for recently dead shad and for stressed shad, indicating that the stressed and dead fish were at similar levels of physiological condition. We hypothesize that mortality due to starvation resulted from a year-long series of unusual environmental conditions beginning with an abnormally warm spring in 1982 which delayed spawning for some shad, a mild winter in 1982–1983 which increased metabolic demands, and an unusually cool spring in 1983 which delayed food availability. These events may have acted in a cumulative fashion, with each inducing additional increments of stress until lipid stores were depleted to a nonrecovery level, which appears to be about 4% of dry body weight. At least 10% of the adult gizzard shad died of starvation. Most predators were probably not adversely affected by the die-off because of the high availability of shad smaller than 16 cm (total length) and the vulnerability of stressed shad to predation.Energy Division, Oak Ridge National Laboratory     

The use of chemosenses by threadfin shad, Dorosoma petenense to detect conspecifics, predators and food

The abilities of threadfin shad ( Dorosoma petenense ) in using chemosenses to detect the scents of conspecifics, a predator and food were tested in the laboratory. Shad did not respond to the scents of conspecifics or a predator; possibly because both shad and their predators out-swim such cues as chemical scents that are transmitted slowly through the water. Shad showed a strong attraction to the scents of food. In feeding, chemosenses may be used for filter feeding on microscopic food and for locating food beyond the range of vision.     

Effects of gizzard shad on benthic communities in reservoirs

Effects of gizzard shad Dorosoma cepedianum on benthic communities in a large southern reservoir (Lake Texoma, U.S.A.) were examined during two field enclosure and exclosure experiments in which enclosures were stocked at high and low densities in 1998 and 1999, respectively. In both years, chironomid abundance significantly increased in treatments that excluded large fishes from foraging on sediments. Mean abundance of chironomids and ostracods were significantly higher ( P  < 0·05) in exclosures than enclosures stocked with gizzard shad at 1140–1210 kg ha⁻¹. In 1999, benthic invertebrate abundances did not differ ( P  > 0·08) between exclosure and enclosures stocked at 175–213 kg ha⁻¹. Per cent organic matter, algal abundance and abundance of other macroinvertebrates in sediments did not differ significantly among treatments in either year. Although chironomid abundance was reduced in gizzard shad enclosures in 1998, food habits from this and other studies showed that adult gizzard shad in Lake Texoma only consumed detritus and algae. It is likely that high sedimentation rates in Lake Texoma limit the ability of gizzard shad to regulate algae and detritus in benthic sediments. Thus, it is concluded that disturbance of benthic sediments by gizzard shad caused the observed reduction in chironomid abundance, rather than through consumption or competition for resources.     

Dietary shifts of a dominant reservoir planktivore during early life stages

We studied dietary shifts in the early life stages of gizzard shad Dorosoma cepedianum, a dominant forage species in North American reservoirs. Larval fish and zooplankton samples were collected weekly during spring in Sardis Reservoir, Mississippi, USA. Diet and prey electivity data suggested the existence of three dietary niches during early life stages: microzooplankton (larvae ≤10 mm total length) in which microzooplankters comprised over 90% by number; crustacean zooplankton (larvae 11–25 mm) in which larval gizzard shad consumed substantial numbers of crustacean zooplankton; and microplankton (larvae >25 mm) in which gizzard shad shifted to filtering protozoans, rotifers, and phytoplankton. There was a high overlap (84%) between the diet of larval gizzard shad and crappies Pomoxis spp. during early May. Larval gizzard shad can potentially reduce microzooplankton density through predation, then shift to crustacean zooplankton and drive their density to decline, then revert to filtration of microzooplankton and exploit phytoplankton. Although, gizzard shad have the ability to influence trophic interactions in reservoir ecosystems, their influence may sometimes be masked by the intensity of bottom-up and top-down effects, as well as population and community interactions. This revised version was published online in July 2006 with corrections to the Cover Date.     

Predation on exotic zebra mussels by native fishes: effects on predator and prey

SUMMARY 1. Exotic zebra mussels, Dreissena polymorpha, occur in southern U.S. waterways in high densities, but little is known about the interaction between native fish predators and zebra mussels. Previous studies have suggested that exotic zebra mussels are low profitability prey items and native vertebrate predators are unlikely to reduce zebra mussel densities. We tested these hypotheses by observing prey use of fishes, determining energy content of primary prey species of fishes, and conducting predator exclusion experiments in Lake Dardanelle, Arkansas. 2. Zebra mussels were the primary prey eaten by 52.9% of blue catfish, Ictalurus furcatus; 48.2% of freshwater drum, Aplodinotus grunniens; and 100% of adult redear sunfish, Lepomis microlophus. Blue catfish showed distinct seasonal prey shifts, feeding on zebra mussels in summer and shad, Dorosoma spp., during winter. Energy content (joules g⁻¹) of blue catfish prey (threadfin shad, Dorosoma petenense; gizzard shad, D. cepedianum; zebra mussels; and asiatic clams, Corbicula fluminea) showed a significant species by season interaction, but shad were always significantly greater in energy content than bivalves examined as either ash-free dry mass or whole organism dry mass. Fish predators significantly reduced densities of large zebra mussels (>5 mm length) colonising clay tiles in the summers of 1997 and 1998, but predation effects on small zebra mussels (≤5 mm length) were less clear. 3. Freshwater drum and redear sunfish process bivalve prey by crushing shells and obtain low amounts of higher-energy food (only the flesh), whereas blue catfish lack a shell-crushing apparatus and ingest large amounts of low-energy food per unit time (bivalves with their shells). Blue catfish appeared to select the abundant zebra mussel over the more energetically rich shad during summer, then shifted to shad during winter when shad experienced temperature-dependent stress and mortality. Native fish predators can suppress adult zebra mussel colonisation, but are ultimately unlikely to limit population density because of zebra mussel reproductive potential.     

Sediment processing by gizzard shad, Dorosoma cepedianum (Lesueur), in Acton Lake, Ohio, U.S.A.

Gizzard shad are primarily detritivorous in Acton Lake, a 253-ha impoundment in southwestern Ohio, U.S.A. To determine the magnitude of sediment utilization by the gizzard shad population in Acton Lake. I used data on population density and age structure, daily ration, and feeding selectivity in estimating the mass of sediments processed by shad daily from April through November. At densities of 4595–10 645 fish ha⁻¹(wet weight biomass = 90–121 kgha ¹), gizzard shad could process 3.8–23.0 kg of dry sediments ha⁻¹ day ¹. On average throughout the growing season, gizzard shad could process a dry mass of sediments each day equivalent to 13% of shad wet weight biomass. Because of the high rate of sedimentation (> 700 kg dry sediment ha⁻¹ day⁻¹) in Acton Lake, gizzard shad can process < 4% of the freshly deposited sediments each day, and therefore are likely to have little effect on benthic community dynamics in the system.     

Movement into floodplain habitats by gizzard shad (<Emphasis Type="Italic">Dorosoma cepedianum</Emphasis>) revealed by dietary and stable isotope analyses

Floodplain habitats have been inferred to provide a variety of functions for aquatic organisms, yet few studies have documented movement between channel and aquatic floodplain habitats. We exploited spatial variation in stable isotope ratios of gizzard shad (Dorosoma cepedianum) to document movement between floodplain lakes and the main river channel of the Brazos River, Texas, during a period of frequent hydrologic connectivity. Additionally, we examined stomach contents of shad to determine if ontogenetic diet shifts or faunal exchange best explained variation in isotopic ratios. Regression analysis indicated significant relationships between gizzard shad size and isotopic ratios in oxbow lakes, whereas these relationships were not significant for the main channel. Plots of individual fish in each habitat suggested that adult shad migrated into oxbow lakes during floods whereas juveniles assimilated material produced in oxbows. Some adults in oxbows had signatures similar to juveniles, and these individuals were probably long-term oxbow residents. The proportion of adults with a “river” signature was greater in the oxbow with the shortest flood recurrence interval where opportunities for faunal exchange were more frequent. Analysis of stomach contents indicated almost total overlap between adult and juvenile diets indicating that movement between habitats having different isotopic ratios of basal resources rather than ontogenetic dietary shifts best explained patterns of isotopic variation in Brazos River gizzard shad.     

Feeding of Burbot, Lota lota, at Different Temperatures

Daily food intake of adult burbot, Lota lota, fed on vendace, Coregonus albula, were estimated experimentally at four different water temperatures (2.4, 5.1, 10.8 and 23.4°C). Mean daily food intake (MDI; g d^–1) and relative daily food intake (RDI; g g^–1 d^–1) increased with temperature from 2.4 to 10.8°C and decreased at 23.4°C. Temperatures of maximum daily food intake values were 13.6°C for MDI and 14.4°C for RDI. No correlation between food intake values and burbot weight was observed. RDI values were used to estimate annual food consumption of burbot population. Annual food consumption estimates were 9.7kg ha^–1 and 24.3kg ha^–1 when burbot biomass was 2.0 or 5.0kg ha^–1, respectively.     

Biomass and primary-production of herbaceous plant communities in the Amazon floodplain

The increase in biomass of different aquatic and terrestrial herbaceous plant communities was measured during various growth periods in the Amazon floodplain near Manaus. Maximum biomass varied from 4–11.2 t ha^–1 dry weight in mixed annual terrestrial communities to 6–23 t ha^–1 in aquatic annual species (Paspalum repens, Oryza perennis, Luziola spruceana and Hymenachne amplexicaulis) and 15.6–57.6 t ha^–1 in communities of the perennial species Paspalum fasciculatum. Cumulative biomass of 3 successively growing annual species reached 30 t ha^–1 a^–1. Net primary production is considerably higher than maximum biomass. Paspalum fasciculatum reached 70 t during a growth period of 8 months. If one considers for annual species a monthly loss of 10–25% of the biomass, then net primary production in areas with three successive macrophyte communities and a cumulative maximum biomass of 30 t ha^–1 is estimated to reach up to 50 t ha^–1 a^–1. Annual P/B ratio may reach about 3.     

Population dynamics,production and angling catch of brown trout,Salmo trutta,in a mature upland reservoir in mid-Wales

Synopsis The brown trout in Llyn Frongoch, a mature upland reservoir, and its nursery stream was sampled during 1983. The stream stock consisted largely of the 1983 and 1982 year classes, with fish reaching mean lengths of 7.0 and 11.6 cm at one and two years of age. The size and biomass of the stream stock at the beginning of 1983 and 1984 were estimated to be 120 and 125 (1.20 and 1.25 fish m^–2) and 1.41 and 0.69 kg (14.1 g m^–2 and 6.9 g m^–2) respectively. Annual stream production ranged from an estimated minimum of 2.49 kg (24.9 g m^–2) to an estimated maximum of 4.59 kg (45.9 g m^–2). Both downstream and upstream movements of 0+ juveniles were recorded. The adult spawning stock was estimated at 79 males and 32 females, a sex ratio of 2.5:1, with most spawners belonging to the 1980 yearclass. The average size of the lake stock over the year was estimated to be 1 650 (229 fish ha^–1) or 250.8 kg (34.8 kg ha^–1). The 1980 yearclass was predominant; there were few fish older than five years. Seasonal variations in netting catches suggested movements to and from the littoral region. Growth in the lake was moderately fast, with fish reaching mean lengths of 21.7 and 27.2 cm by three and four years of age. Fish entering the lake after one year appeared to grow faster than fish which remained in the stream for two years. Annual production in the lake was estimated at 136.7 kg (19.0 kg ha^–1). The total angling catch for the season was estimated to be 62.6 kg (8.7 kg ha^–1).     

Response of corn (Zea mays L.) to manganese application on Atlantic Coastal Plain soils

Field research was conducted on four Atlantic Coastal Plain soils in the United States to evaluate response of corn (Zea mays L.) plants to Mn application. The soils under study were classified as either Aeric or Typic Ochraquults. Manganese application increased corn grain yields by an average of 1195 kg ha^–1 on the four soils. The average grain yields on the soils were 7955 kg ha^–1 for the control and 9150 kg ha^–1 for the +Mn treatment. A Mitscherlich plant growth model was used to establish relationships between percent maximum grain yield and Mn concentration in the ear leaf at early silk (r=0.87, =0.01) and in the mature grain (r=0.58, =0.01). Based on 90% of maximum yield as the definition of the critical deficiency level, the critical Mn deficiency levels calculated with parameters from the Mitscherlich model were 10.6 mg kg^–1 in the ear leaf and 4.9 mg kg^–1 in the grain.     

Response of field-grown chickpea (Cicer arietinum L.) to phosphorus fertilization for yield and nitrogen fixation

Application of phosphorus at 40, 60, 80 and 100 kg P₂O₅ ha^–1 in the presence of a uniform dressing of nitrogen (N) and potash (K₂O) each applied at 20 and 24 kg ha^–1 to chickpea (CM-88) grown in sandy loam soil in a replicated field experiment improved the nodulation response of the crop, increased its grain yield (ka ha^–1) by 18, 59, 40 and 14 percent, biomass yield (ka ha^–1) by 32, 32, 54 and 14 percent, biomass N (kg ha^–1) by 31, 48, 49, 19 percent, and biomass P (kg ha^–1) by 26, 40, 41 and 11 percent, respectively. The effect of phosphorus on the nitrogenase activity of the excised roots of chickpea was, however, inconsistent.     

Biomass,Carbon, and Nitrogen Pools in Mexican Tropical Dry Forest Landscapes

Tropical dry forest is the most widely distributed land-cover type in the tropics. As the rate of land-use/land-cover change from forest to pasture or agriculture accelerates worldwide, it is becoming increasingly important to quantify the ecosystem biomass and carbon (C) and nitrogen (N) pools of both intact forests and converted sites. In the central coastal region of México, we sampled total aboveground biomass (TAGB), and the N and C pools of two floodplain forests, three upland dry forests, and four pastures converted from dry forest. We also sampled belowground biomass and soil C and N pools in two sites of each land-cover type. The TAGB of floodplain forests was as high as 416 Mg ha^–1, whereas the TAGB of the dry forest ranged from 94 to 126 Mg ha^–1. The TAGB of pastures derived from dry forest ranged from 20 to 34 Mg ha^–1. Dead wood (standing and downed combined) comprised 27%–29% of the TABG of dry forest but only about 10% in floodplain forest. Root biomass averaged 32.0 Mg ha^–1 in floodplain forest, 17.1 Mg ha^–1 in dry forest, and 5.8 Mg ha^–1 in pasture. Although total root biomass was similar between sites within land-cover types, root distribution varied by depth and by size class. The highest proportion of root biomass occurred in the top 20 cm of soil in all sites. Total aboveground and root C pools, respectively, were 12 and 2.2 Mg ha^–1 in pasture and reached 180 and 12.9 Mg ha^–1 in floodplain forest. Total aboveground and root pools, respectively, were 149 and 47 kg ha^–1 in pasture and reached 2623 and 264 kg ha^–1 in floodplain forest. Soil organic C pools were greater in pastures than in dry forest, but soil N pools were similar when calculated for the same soil depths. Total ecosystem C pools were 306. The Mg ha^–1 in floodplain forest, 141 Mg ha^–1 in dry forest, and 124 Mg ha^–1 in pasture. Soil C comprised 37%–90% of the total ecosystem C, whereas soil N comprised 85%–98% of the total. The N pools lack of a consistent decrease in soil pools caused by land-use change suggests that C and N losses result from the burning of aboveground biomass. We estimate that in México, dry forest landscapes store approximately 2.3 Pg C, which is about equal to the C stored by the evergreen forests of that country (approximately 2.4 Pg C). Potential C emissions to the atmosphere from the burning of biomass in the dry tropical landscapes of México may amount to 708 Tg C, as compared with 569 Tg C from evergreen forests.     

Evaluating short-term effects of omnivorous fish removal on water quality and zooplankton at a subtropical lake

We evaluated a biomanipulation program to test for short-term changes in water quality (chlorophyll a, Secchi depth, total phosphorus) and macrozooplankton biomass following partial removal of omnivorous gizzard shad Dorosoma cepedianum. The removal occurred at a eutrophic subtropical lake, and responses were compared to an unmanipulated control lake using a before-after-control-impact paired series analysis. The removal reduced the biomass of large (>300 mm) gizzard shad by 75% over 2 years via a subsidized commercial gill net fishery. However, the total population biomass of gizzard shad was reduced by approximately 32% from an average pre-manipulation biomass of 224 kg ha⁻¹ due to the size selectivity of the gear, which did not effectively capture small fish (<300 mm). No significant short-term changes in chlorophyll a concentration, Secchi depth, total phosphorus concentration or macrozooplankton biomass were detected following biomanipulation. The partial removal may have fallen short of the biomass reduction required to cause ecosystem responses. Our results suggest that moderate omnivore removals (i.e., <40% biomass reduction) will have little short-term benefits to these lakes, and future manipulations should use a less size-selective gear to achieve a larger total biomass reduction.     

Carbon balance of a tropical savanna of northern Australia

Through estimations of above- and below-ground standing biomass, annual biomass increment, fine root production and turnover, litterfall, canopy respiration and total soil CO₂ efflux, a carbon balance on seasonal and yearly time-scales is developed for a Eucalypt open-forest savanna in northern Australia. This carbon balance is compared to estimates of carbon fluxes derived from eddy covariance measurements conducted at the same site. The total carbon (C) stock of the savanna was 204±53 ton C ha^–1, with approximately 84% below-ground and 16% above-ground. Soil organic carbon content (0–1 m) was 151±33 ton C ha^–1, accounting for about 74% of the total carbon content in the ecosystem. Vegetation biomass was 53±20 ton C ha^–1, 39% of which was found in the root component and 61% in above-ground components (trees, shrubs, grasses). Annual gross primary production was 20.8 ton C ha^–1, of which 27% occurred in above-ground components and 73% below-ground components. Net primary production was 11 ton C ha^–1 year^–1, of which 8.0 ton C ha^–1 (73%) was contributed by below-ground net primary production and 3.0 ton C ha^–1 (27%) by above-ground net primary production. Annual soil carbon efflux was 14.3 ton C ha^–1 year^–1. Approximately three-quarters of the carbon flux (above-ground, below-ground and total ecosystem) occur during the 5–6 months of the wet season. This savanna site is a carbon sink during the wet season, but becomes a weak source during the dry season. Annual net ecosystem production was 3.8 ton C ha^–1 year^–1.