Variations in CO 2 fluxes from grass carp Ctenopharyngodon idella aquaculture polyculture ponds

Monthly and diurnal CO2 fluxes at the water−air interface were measured in 3 grass carp Ctenopharyngodon idella polyculture ponds using the closed-chamber method during the farming season from April to September 2013. The results showed that the mean CO2 emission rate from the 3 ponds was 97.8 mg m−2 h−1 (range 50.6 to 151.0 mg m−2 h−1). Apparent seasonal and diurnal variations in CO2 fluxes were observed; the lowest (1.73 mg m−2 h−1) occurred in April and the highest occurred in July and August (212.9 and 181.9 mg m−2 h−1, respectively). The night-time to daytime CO2 emission ratio was <1 (range 0.44 to 0.83) in May during the early part of the farming season. Night-time CO2 emissions in early August were higher than those during the daytime (range 1.06 to 1.37). The night-time to daytime CO2 emission ratios were 0.84 to 1.36 during late September. The CO2 fluxes from the freshwater aquaculture ponds were higher than those of lakes and reservoirs, whereas they were lower than those of marshes in previous studies. The increases in nutrient load and pelagic respiration from feeding the fish were significantly associated with CO2 emissions from the freshwater aquaculture ponds. Freshwater aquaculture ponds in temperate monsoon climate regions contribute an estimated 2.98 million t yr−1 CO2 based on the mean investigated CO2 emission rate, contributing 0.0088% of the current annual global CO2 emissions, and represent a small but previously unquantified CO2 emission source.


INTRODUCTION
The long-lived greenhouse gases (LLGHGs) carbon dioxide (CO 2 ), methane (CH 4 ) and nitrous oxide (N 2 O) have reached new high levels since the industrial revolution.CO 2 , which contributes about 64% of radiative forcing by LLGHGs, has increased from 280.0 ppm in 1800 to 389.0 ppm in 2010 (Tarasova et al. 2012).Global warming caused by increasing atmospheric LLGHG concentrations could affect ecosystem structure and function, and many studies on terrestrial and aquatic ecosystem CO 2 cycles have been performed (Butman & Raymond 2011, Keenan et al. 2013, Raymond et al. 2013).
Most studies on carbon cycles of freshwater ecosystems have focused on lakes, reservoirs and rivers.Total carbon sinks in lakes, including both inorganic and organic carbon, amount to 0.077 Gt yr −1 , of which the atmosphere accounts for 0.0532 Gt yr −1 (Downing et al. 1993).Furthermore, there is a higher retention rate (500 g m −2 yr −1 ) of carbon in reservoirs than in lakes, and the total carbon sinks will reach 0.2 Gt yr −1 in 2050 (Walsh 1991, Yan & Liu 2001).Inland freshwater ecosystems have huge carbon sequestration po tentials.However, the freshwater ecosystem is also considered a source of CO 2 to the atmosphere (Cole et al. 1994, Huttunen et al. 2002, Eugster et al. 2003, Huttunen et al. 2003, Jonsson et al. 2003, Xing et al. 2005, Repo et al. 2007).Both processes play important roles in the carbon cycle of the Earth system.
According to statistical data on aquaculture production maintained by the Food and Agriculture Organization of the United Nations, the combined water surface area of freshwater aquaculture ponds is estimated to be 25 669 km 2 in China and 87 500 km 2 worldwide (Verdegem & Bosma 2009).Freshwater aquaculture ponds sequester an estimated 13.1 Mt yr −1 of carbon globally, as organic carbon from uneaten feed, organic fertilizers, dead plankton and excreta from different species settle to pond bottoms (Boyd et al. 2010, Adhikari et al. 2012).However, no detailed studies have considered CO 2 sink/source functions of freshwater aquaculture ponds across the water−air interface.
Freshwater aquaculture pond ecosystems are generated semi-artificially by daily feeding and photosynthesis of phytoplankton.He (1993) reported that phytoplankton primary production accounts for 29.7 to 49.7% of the total organic carbon in aquaculture ponds and that feeding and organic fertilizer comprise 50.3 to 70.3%.However, a large quantity of nutrients from residual feed is retained in aquaculture pond water.The feed conversion rate in grass carp polyculture ponds is 1.69 to 2.86 (Zhang et al. 2011).Xia (2013) revealed that only a small portion of nutrient input into grass carp culture ponds is converted into fish production and that most of it (81.7%C, 65.6% N, 67.5% P) remains in pond sediment or is discharged to adjacent bodies of water.
Many studies have indicated that decomposition of allochthonous organic carbon from inland freshwater ecosystems is a significant source of atmospheric CO 2 (Jones et al. 1998, Jonsson et al. 2001, Karlsson et al. 2007).The accumulation of residual feed and feces increases nutrient load and plankton biomass in ponds (Zhao et al. 2002, Burford et al. 2003).Nevertheless, the effects of nutrient load on carbon sink/ source functions are controversial.Several studies have shown that carbon sequestration is enhanced with an in crease in nutrient load (Downing et al. 1993), or that carbon emissions decrease when primary production is high in freshwater (Xing et al. 2005(Xing et al. , 2006)).In contrast, other studies have reported that high primary production contributes to atmospheric carbon owing to the intense mineralization of organic matter de rived from primary production (Hamilton et al. 1994, Roulet et al. 1997, Huttunen et al. 2003, Ran takari & Kortelainen 2005, Aller & Blair 2006, Zheng et al. 2009).
Grass carp Ctenopharyngodon idella is a very popular freshwater aquaculture species, accounting for about 18.8% of freshwater aquaculture production in China and 12.2% worldwide (Fisheries Department of Agriculture Ministry of China 2013, FAO 2013).Pond-farmed grass carp are fed a low-protein pellet feed (Bostock et al. 2010).Generally, a number of filter fish species, such as silver carp Hypophythalmichthys molitrix and bighead carp Aristichthys nobilis, are stocked into aquaculture ponds to limit plankton biomass.In addition, some benthic fish species, such as common carp Cyprinus carpio and white leg shrimp Litopenaeus vannamei, are stocked to take full advantage of pond space (Dong 2011).As the role of freshwater aquaculture ponds in temperate monsoon regions within the global carbon cycle has not been investigated, we measured the CO 2 fluxes at the water-air interface from 3 representative grass carp polyculture ponds to determine the influence of aquaculture on CO 2 emissions from freshwater.

Experimental ponds
The study was carried out at a freshwater fish farm in Gaoqing, Shandong Province, in the north of China (37°04' N, 117°33' E), which represents a typical temperate monsoonal climate with mean annual temperature of ~13.1°C.The mean annual precipitation is 539.4 mm and there is much more precipitation in summer than in other seasons.The study was conducted in 3 representative grass carp polyculture ponds.The GSB pond was stocked with grass carp, silver carp and bighead carp; the GSBL pond was stocked with grass carp, silver carp, bighead carp and shrimp; and the GSBC pond was stocked with grass carp, silver carp, bighead carp and common carp.Pellet feed was provided daily at 07:00, 10:00, 13:00 and 16:00 h.There was no water ex change during the farming season.The basic information of the 3 ponds is shown in Table 1.

Sample collection
Water samples were collected with a horizontal sampler monthly during the farming season from April to September 2013.The samples were stored in clean plastic 1 l bottles and were transported imme-diately to the laboratory.At the same time, CO 2 gas samples were collected at the water−air interface at 09:00 to 11:00 h on sunny days using the floating closed-chamber method (Xing et al. 2005(Xing et al. , 2006)).Diurnal CO 2 variations were measured during different times of the culture season on 16 to 17 May, 1 to 2 August and 28 to 29 September.The floating chambers (50 cm in height and 30 cm in diameter) were made from Plexiglas and covered by aluminum foil to avoid high temperatures in summer resulting from direct sunlight (Duchemin et al. 1999, Huttunen et al. 2003).A small vertical vent stopped by a silicone septum on the top was used for sampling, and a 4.5 V dry battery-driven fan was used inside the chamber to mix the air but not disturb the water−air interface.Three replicate chambers were placed on each pond.Prior to sampling, the syringe was pumped several times to mix the air inside the chamber.Four samples were drawn from the chamber via 100 ml Tygon syringes 0, 10, 20 and 30 min after deployment.These air samples were driven into dedicated sample bags and transported to laboratory in a cool box.During sampling, the air temperature inside and outside the chamber was measured and noted.

Sample measurements
Water temperature and dissolved oxygen were measured with a YSI dissolved oxygen meter (Model 5000, 230 V), and pH was measured with a PHS-3C acido meter (Shanghai REX Instruments) in situ when the water samples were collected.The quantity of feed supplied to the ponds was determined monthly, and the feed conversion rate was calculated as the total quantity of feed provided (kg)/total weight gain of fish (kg) during the farming season.Or ganic carbon content of the feed was determined using a PE-24 CHN analyzer (Heraeus) after acidifying a feed sample for 4 h with 1 N HCl to remove inorganic carbon (Holmer et al. 2007).Chl a was extracted with acetone (90%) in the darkness for 24 h after filtration of water samples through GF/F glass microfiber filters and analyzed according to the method of Wu (2006).The absorbance values of filtrate were measured at 750, 663, 645 and 630 nm with the blank of 90% acetone, and the concentration of chl a was determined by the given equation (in Wu 2006).Water samples for dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) concentrations were analyzed by a multi-2100s TOC analyzer (Analytik Jena) after being filtered through pre-combusted (450°C, 2 h) Whatman GF/F filters.Primary production and pelagic re spiration were measured using the light−dark bottle method (Eaton et al. 1995).Measurements were made by recording the change in O 2 concentration during 24 h at water depths of 0.05, 0.5 and 1.5 m, but the depths at which the bottle samples were collected were adjusted according to water transparency.After collecting the samples to determine the initial oxygen concentration, 2 light and 2 dark bottles were suspended at each water depth in each pond.Oxygen concentrations were measured using the Winkler method.Primary production and pelagic respiration were determined by cumulative mean values at each depth using the differences between the light and dark bottles, and initial values and dark bottles, respectively.
Gas samples were analyzed as soon as possible using the GC-2010 Plus gas chromatograph (Shimadzu) connected to a MGS-4 gas sampler and a MTN-1 methanizer.The gas samples were driven into the MGS-4 gas sampler, then converted into CH 4 in the MTN-1 methanizer by a nickel catalyst at 375°C after being separated on a 2 m × 2 mm stainless steel column at 40°C packed with poropak Q 60-80 mesh.The CO 2 concentrations in gas samples were determined with a flame ionization detector at 220°C.The carrier gas (N 2 ) flow rate was 22 ml min −1 , and flame gases (H 2 and compressed air) were set at 20 and 30 ml min were measured every 4 gas samples in order to determine the sample concentrations and check the errors.The CO 2 fluxes were calculated from the linear changes in time inside the chamber.

Data analysis
The data were analyzed using the statistical software package SPSS (Ver.18.0).Linear regressions were used to identify the relationships between seasonal CO 2 fluxes and other environmental factors (the concentrations of chl a and DOC, quantity of organic carbon supplied by feed, pelagic respiration and water temperature).The relationships between diurnal variations of CO 2 and environment factors were determined by bivariate correlation, and r-values were reported.

Characteristics of the experimental ponds
The variations in air and water temperature during the farming season are shown in Fig. 1a.The maxi-mum air temperature of 40.0°C occurred in July, and the minimum of 17.5°C occurred in April.The water temperature (range 22.2 to 32.8°C) varied similarly to the air temperature.
The yield and quantity of feed supplied to the ponds are shown in Table 1.The total quantities of feed added to the GSB and GSBC were 3.15 and 2.54 kg m −3 , respectively.The inputs of fish and shrimp feed into the GSBL were 1.94 and 0.30 kg m −3 , respectively.The feed conversion rate was 2.13 in the GSB, 2.07 in the GSBL and 2.08 in the GSBC.The mean ± SD organic carbon contents of the fish and shrimp feed were 42.7 ± 1.09 and 41.7 ± 2.02%, respectively.The monthly quantities of organic carbon supplied by feed, which reached a maximum in July or August (0.33 kg m −3 in the GSB, 0.24 kg m −3 in the GSBL and 0.28 kg m −3 in the GSBC), are shown in Fig. 1b.
The pelagic respiration results are shown in Fig. 1f and ranged from 9.13 to 19.6 g O 2 m −2 d −1 in the GSB, from 8.33 to 28.2 g O 2 m −2 d −1 in the GSBL, and from 10.3 to 31.7 g O 2 m −2 d −1 in the GSBC.The ranges of primary production were 9.06 to 12.8, 5.06 to 12.6 and 7.95 to 11.7 g O 2 m −2 d −1 in the GSB, GSBL and GSBC, respectively (Fig. 1f).The primary production/pelagic respiration (P/R) values of the 3 ponds were <1 during most of the farming season (Fig. 1f).

CO 2 flux values at the water−air interface
Monthly CO 2 flux variations The mean CO 2 fluxes at the water−air interface of the 3 ponds were 91.9 mg m −2 h −1 (range, −4.95 to 200.3 mg m −2 h −1 ) in the GSB, 50.6 mg m −2 h −1 (range, −0.13 to 200.8 mg m −2 h −1 ) in the GSBL and 151.0 mg m −2 h −1 (range, 10.3 to 311.8 mg m −2 h −1 ) in the GSBC (Fig. 2).These data demonstrate that the investigated aquaculture ponds were a source of atmospheric CO 2 for most of the farming time, except for the GSB and GSBL during April.The CO 2 fluxes for the GSB and GSBL in April were −4.95 and −0.13 mg m −2 h −1 , respectively.Maximum CO 2 fluxes in the GSB (200.3 mg m −2 h −1 ), GSBL (200.8 mg m −2 h −1 ) and GSBC (311.8 mg m −2 h −1 ) occurred in July or August.
The correlation analysis showed that CO 2 emissions were positively correlated with chl a concentrations (Fig. 3), quantity of organic carbon supplied by feed, pelagic respiration, water temperature and DOC concentrations (p < 0.05) of the ponds (Table 2).There were no significant correlations between DIC and CO 2 fluxes (p > 0.05).

Diurnal CO 2 flux variations
Mean diurnal CO 2 fluxes in May were 19.9 ± 15.1 (GSB), 26.4 ± 10.7 (GSBL) and 31.3 ± 21.6 mg m −2 h −1 (GSBC).The night-time to daytime CO 2 flux ratios were 0.72, 0.83 and 0.44 in the GSB, GSBL and GSBC, respectively.Similar CO 2 flux trends were ob served; CO 2 fluxes were relatively low at 16:00 to 20:00 h, reached a peak at 24:00 h, decreased to lower values at 04:00 h with the decrease in temperature in the early morning, and thereafter increased gradually (Fig. 4a).The night-time to daytime CO 2 flux ratios in  early August were 1.09, 1.37 and 1.06 in the GSB, GSBL and GSBC, respectively.Flux maxima were observed at 04:00 h, and minima in the GSB and GSBC occurred at 20:00 h, whereas the minima occurred at 12:00 h in the GSBL.The mean diurnal CO 2 fluxes in the GSB, GSBL and GSBC were 162.4 ± 30.0, 189.6 ± 45.9 and 241.4 ± 38.0 mg m −2 h −1 , respectively (Fig. 4b).Similar CO 2 diurnal flux variations were found in the GSB and GSBC during late September, and the night-time to daytime CO 2 flux ratios were 1.24 and 1.36.Daytime CO 2 flux was higher than that at night-time in the GSBL, and the night-time to daytime CO 2 flux ratio was 0.84.The mean diurnal CO 2 fluxes were 141.9 ± 37.5, 4.34 ± 7.21 and 141.6 ± 35.3 mg m −2 h −1 in the GSB, GSBL and GSBC, respectively (Fig. 4c).The correlations between the diurnal fluxes and the environment factors are shown in Table 3.

Characteristics of CO 2 flux at the water−air interface
Only CO 2 fluxes measured using the closed-chamber method were compared between different inland freshwater systems, as CO 2 flux at the water−air interface can be affected by the measuring method.The mean monthly CO 2 fluxes (range, 1.73 to 212.9 mg m −2 h −1 ) in the 3 grass carp polyculture ponds are listed in Table 4.The CO 2 fluxes we measured in freshwater aquaculture ponds were higher than   those reported previously for lakes and reservoirs but lower than those in marshes generally.In addition, mean CO 2 emission rate from the grass carp polyculture ponds was 97.8 mg m −2 h −1 during the farming season, which was also higher than those of lakes and reservoirs, which range from −49.8 to 64.0 mg m −2 h −1 , but lower than those of freshwater marshes, which range from 281.1 to 428.8 mg m −2 h −1 .Cummins (1977) used the river continuum concept to describe the continuous ecological changes of a river from its headwaters, upstream to downstream.Furthermore, the nutrient levels of aquatic ecosystems increase along with the decrease in depth and water body size (Odum & Barrett 2005).Several studies have shown that carbon sequestration is enhanced with an increase in nutrient load (Downing et al. 1993), or that carbon emission decreases when primary production is high in freshwater (Xing et al. 2005(Xing et al. , 2006)).Moreover, other studies have reported that high primary production contributes to atmospheric carbon owing to the intense mineralization of organic matter derived from primary production (Hamilton et al. 1994, Roulet et al. 1997, Huttunen et al. 2003, Rantakari & Kortelainen 2005, Aller & Blair 2006, Zheng et al. 2009).The lakes or reservoirs, aquaculture ponds and marshes can also be regarded as the continuously integrated series with the grad-ual increases in nutrient levels.The results in Table 4 indicate that CO 2 fluxes from the freshwater aquaculture ponds were higher than those of lakes and reservoirs, whereas they were lower than those of marshes, which supports the deduction that carbon source functions of freshwaters were enhanced with the increase in nutrient levels.

Factors affecting CO 2 emissions
The P/R values of the 3 ponds in the present study were usually <1 (Fig. 1f), indicating that primary production by phytoplankton was exceeded by pelagic respiration in the water column.Therefore, heterotrophic organisms in ponds must receive an external supply of organic carbon to maintain the ecosystem, and feed was the main organic matter source in the pond ecosystems.Continuous input of feed resulted in an increased nutrient load and plankton biomass (with maximum chl a of 0.49 mg l −1 ), leading to the increased pelagic respiration.A regression analysis showed that CO 2 flux was positively correlated with pelagic respiration (Table 2), indicating that intensive pelagic respiration may have significantly affected CO 2 emissions from the aquaculture ponds.CO 2 emissions from the present ponds were positively corre-  2), whereas CO 2 flux is negatively correlated with chl a and water temperature in highly autotrophic lakes (Xing et al. 2005(Xing et al. , 2006)), further indicating that pelagic respiration rather than algal activity is a significant factor regulating CO 2 fluxes.Furthermore, CO 2 flux was positively correlated with the quantity of organic carbon supplied by feed (Table 2), and the decomposition of the allochthonous carbon has been revealed to be the main reason for the CO 2 emissions from the inland waters (Jones et al. 1998, Jonsson et al. 2001, Karlsson et al. 2007).A recent study on the shrimp nutrient budget in farming ponds showed that 40.9% of the carbon input into ponds is released to the air as CO 2 (Su et al. 2009).DOC concentrations accumulated only slightly in the water column during the farming season (Fig. 1d), and mean DOC concentrations in the ponds were similar to those of shrimp farming ponds supplied with artificial feed (Liu et al. 2002).DIC is the largest aquatic carbon pool in ponds and increased continually during the farming season (Fig. 1e).DIC concentrations in our ponds reached 105.0 mg l −1 in the GSB, 105.3 mg l −1 in the GSBL and 101.3 mg l −1 in the GSBC at the end of the farming season, which were higher levels than those reported for natural bodies of water (range, 6.6 to 49 mg l −1 ) (James et al. 2000).The DOC and DIC concentrations were significantly correlated (p < 0.01).We speculated that the organic carbon from residual feed allowed bacteria to flourish (Eugster et al. 2003), leading to intensive mineralization of DOC and the accumulation DIC in the water (Casper et al. 2000, Nakano et al. 2000).The CO 2 flux was significantly and positively correlated with water temperature and DOC concentrations (Table 2).Similar correlations were reported in 75 to 95% of the lakes in northern and temperate regions where the gas emissions are supported by inputs of organic matter (Del Giorgio et al. 1999).Therefore, the decomposition of DOC caused by daily feed supply is also supposed to accelerate CO 2 emissions of the ponds.The CO 2 flux of the GSBL was significantly lower than the GSB and the GSBC in the last 2 mo, which could be explained by the significant decrease in chl a concentrations (Fig. 1c) and the lower feed supply after the yield (Fig. 1b) of 60% of fishes in GSBL.

Diurnal and seasonal variations in CO 2 flux
Significant diurnal CO 2 flux variations were detected in the farm ponds.The night-time to daytime CO 2 emission ratios were <1 (range, 0.44 to 0.83) in May, which was early in the farming season.CO 2 water solubility decreases with the increase in water temperature during the day (Parkin & Kaspar 2003).Roulet et al. (1997) also reported that CO 2 flux in beaver ponds is always higher during the day because of higher water temperature.Diurnal CO 2 fluxes are affected by biotic and abiotic factors, such as plankton biomass, instantaneous wind speed and underwater photosynthetically active radiation (Roulet et al. 1997), which may have led to the abnormally high CO 2 flux in the GSBC at 12:00 h.We found negative correlations between CO 2 flux at different air and water temperatures, chl a concentrations, pH and dissolved oxygen in early August, and maximum flux occurred in the early morning, indicating high gas exchange.Photosynthesis probably played an important role in removing dissolved CO 2 and releasing O 2 during the higher daytime temperatures, which resulted in a decrease in CO 2 flux and an increase in pH.Respiration at night presumably increased CO 2 emissions and the oxygen deficit with a lower water temperature (Guasch et al. 1998, Dornblaser & Striegl 2013).Higher night-time CO 2 fluxes occurred in the GSB and GSBC during September, and the correlations between diurnal CO 2 flux and the environmental factors were similar to those in August.However, the night-time to daytime CO 2 emission ratios were <1 in the GSBL and higher fluxes occurred at 08:00 and 12:00 h.It is possible that the decrease in chl a concentration was the result of a lower photosynthetic rate after part of fish yield.Variations in diurnal CO 2 flux values in aquaculture ponds deserve further study, as they are affected by artificial and natural factors.
CO 2 fluxes across the water−air interface varied significantly between seasons; lower CO 2 emissions were observed at the beginning of the farming season (mean 1.73 mg m −2 h −1 ), higher values were measured in July and August (means 212.9 and 181.9 mg m −2 h −1 , respectively), and lower values were observed again in September (mean 124.9 mg m −2 h −1 ).The seasonal variations of CO 2 fluxes were significantly and positively correlated with chl a concentrations and water temperature (Table 2).Based on the mean CO 2 emission rate (97.8 mg m −2 h −1 ), the investigated freshwater aquaculture ponds emit 0.43 kg m −2 yr −1 CO 2 at the water−air interface during farming season.The sampling site represents a typical temperate monsoonal climate, which covers the northeast part of China, the Russian Far East, the Korean Peninsula and north Japan (Wei 2005).Given that northeast China (including Beijing, Heilong jiang, Jilin, Liaoning, Hebei, Shanxi, Tianjin, Shandong, Shanxi, Henan and part of Jiangsu Province) contains the dominant freshwater aquaculture regions (6941.9km 2 ) (FAO 2013), freshwater aquaculture ponds emit an estimated 2.98 Mt CO 2 yr −1 during farming season in temperate monsoonal climate re gions.The estimated annual CO 2 emission rate is much lower than that of inland waters, which have been estimated to contribute 2100 Mt yr −1 (Raymond et al. 2013).Nevertheless, freshwater aquaculture ponds in temperate monsoonal climate regions, occupying 7.9% of the global freshwater aquaculture pond area, contribute 0.0088% of current annual global CO 2 emissions of about 34 000 Mt yr −1 (Olivier et al. 2012), and represent a small but previously unquantified source of CO 2 emissions, which will be useful in considerations of global carbon balance.Detailed analyses of carbon emissions by freshwater aquaculture ponds in other areas worldwide will be needed to ascertain whether freshwater aquaculture contributes to CO 2 emissions.

Table 1 .
−1 , respectively.Standard gases Basic information on the experimental ponds.The GSB was stocked with grass carp fingerlings of 0.07 to 0.08 kg ind.−1 , and the GBSL and GBSC were stocked with young grass carp of 0.30 to 0.35 kg ind.−1 .The other species stocked were the same size in all 3 ponds with young silver carp of 0.11 to 0.14 kg ind. - , young bighead carp of 0.32 to 0.37 kg ind. - .Fishes were stocked in mid-April, shrimps at the beginning of June.Fishes and shrimps were yielded at the beginning of October, except for 60% cultured fishes in GSBL, which were yielded in mid-August.Feeds input refers to the total feeds input during the culture period.
G: grass carp; S: silver carp; B: bighead carp; L: white leg shrimp; C: common carp.Means are given ± SD

Table 2 .
Regression equations showing the relationships between CO 2 fluxes (mg m −2 h −1 ; y) and some environmental factors (x)

Table 4 .
Comparisons of CO 2 fluxes (mg m −2 h −1 ) in different inland freshwater systems.Time given as yyyy.mmlated with water temperatures and chl a concentrations (Table