Fish Feed Quality Is a Key Factor in Impacting Aquaculture Water Environment: Evidence from Incubator Experiments

The effect of fish feed quality has gained increasing attention to alleviate the harmful environmental impacts induced by intensive aquaculture. In current research, we have conducted an incubator experiment to highlight the effect of fish feed quality on aquaculture water environment. Fish feed from three manufactures with two different dosages (0.1000 g, 0.2000 g) was added to the culture medium with and without Microcystis aeruginosa. Treatments with Microcystis aeruginosa were named as MHT, MHP and MZT; while the treatments without Microcystis aeruginosa named as HT, HP and ZT. Microcystis aeruginosa densities and nutrients concentrations were measured in the study. Results have shown that fish feed quality (manufactures) has a great effect on nutrients concentrations in the absence of Microcystis aeruginosa (P < 0.05). Meanwhile, fish feed can stimulate Microcystis aeruginosa growth that is also influenced by fish feed quality excluding lag phase (0~12 day) significantly in general (P < 0.05). The maximum Microcystis aeruginosa density (Nmax) is 1221.5, 984.5, 581.0, 2265.9, 2056.8 and 1766.6 1 × 104 cells mL−1 for MHT 0.1 g, MHP 0.1 g, MZT 0.1 g, MHT 0.2 g, MHP 0.2 g and MZT 0.2 g, respectively. In treatments with algae, fish feed quality affect total phosphorus (TP) concentrations (except the difference between MHT and MHP) and total nitrogen (TN) concentrations significantly (P < 0.05). For most of consumed nutrients, the obvious differences among all treatments were observed excluding lag phase in general (P < 0.05), which suggest that the nutrient utilization is also dependent on fish feed quality. Keeping in mind the above facts it is concluded that fish feed quality is a key factor in impacting aquaculture water environment.

experimental methods. Effects of different fish feed on nutrients release and algae growths were assessed using batch incubation experiments. In the experiment, 400 mL sterilized M-II culture medium without nitrogen and phosphorus was used, and weights of 0.1000 g and 0.2000 g of the three different fish feed (from different manufactures) were added into the media served as P and N sources with 1 L flask. Treatments without algae containing 0.1 g fish feed were named "HT 0.1 g", "HP 0.1 g", "ZT 0.1 g", and containing 0.2 g fish feed named "HT 0.2 g", "HP 0.2 g" and "ZT 0.2 g", respectively. Meanwhile, treatments with algae containing 0.1 g fish feed were named "MHT 0.1 g", "MHP 0.1 g", "MZT 0.1 g", and containing 0.2 g fish feed named "MHT 0.2 g", "MHP 0.2 g" and "MZT 0.2 g" conforming to the treatments' name, respectively. Duplicates were prepared. Flasks were shaken and their positions were changed at random three times a day. The initial algae density was 1.0 × 10 5 cells mL −1 .
During the experimental period (37 days), algal cell densities were counted every two days using a haemacytometer under a microscope 43,47 . Counting was performed three times per sample. Water sampling started 1 day after algae addition, and total phosphorus (TP), total dissolved phosphorus (TDP), total particulate phosphorus (TPP = TP-TDP), orthophosphate (PO 4 3− -P), total nitrogen (TN), total dissolved nitrogen (TDN), total particulate nitrogen (TPN = TN-TDN) and ammonia (NH 4 + -N) were also measured every two days. Concentrations of PO 4 3− -P, TDP and TP were determined via the persulphate digestion and ammonium molybdate spectrophotometric method 48 . NH 4 + -N was analyzed using the phenol-hypochlorite method 48 . TN and TDN were analyzed using the procedure of alkaline potassium persulfate digestion with ultra-violet light spectroscopy 49 . M. aeruginosa growth kinetics. Algal growth can be well described by (original) Logistic function 50-54 . However, this function does not satisfy the initial conditions of algal growth. A modified Logistic function was proposed by Huang et al. 49 and it is as follows: where N (1 × 10 4 cells mL −1 ) is the algae density at any time, N max is the maximum algae density (1 × 10 4 cells mL −1 ), r(d −1 ) is the intrinsic growth rate, N 0 is the algae density at 0 day, and N 0 is 10 × 10 4 cells mL −1 in the present study, t(d) is time and a (−) is a constant. N max , a and r can be obtained by fitting Eq. The growth rate reached its maximal value µ′ = cmax rN 4 max (1 × 10 4 cells (mL·d) −1 ) when N equals to half of N max 49,51,53,55 .
The formula of the specific growth rate from the modified Logistic function as shown in Eq. (3), describing variations of specific growth rates with time is also better than that derived from Logistic function 49 : www.nature.com/scientificreports www.nature.com/scientificreports/ Statistical analysis. Experimental data was analyzed statistically by using Origin 8.6 and SPSS 19.0. Logistic model was examined for their fit to the experimental data using Origin 8.6. Origin 8.6 or SPSS 19.0 is used to determine correlation coefficients between the measured and predicted variables as well as between M. aeruginosa densities and nutrients concentrations. The statistical analysis is applied to identify the significant differences among groups with different fish feed by analysis of variance (ANOVA) with SPSS 19.0. Moreover, standard deviation was calculated and data was expressed in terms of means + SD of the two replicates.

Results and Discussion
Effects of different fish feed on nutrients concentrations without algae. Effects of different fish feed on phosphorus concentrations. Phosphorus is chemical compound found in fish feed 33 , its labile form (PO 4 3− -P) is a major form of released phosphorus from fish feed 43 . From Fig. 1, TP, PO 4 3− -P and TDP concentrations in treatments with HT, HP and ZT increase gradually in the first 10 days and then enter into a stable phase. Meanwhile, released concentrations of TDP and PO 4 3− -P from fish feed reached 85.39~90.00% and 75.23~89.91% of their corresponding maximal values at the first sampling day (or 24 hours). Akhan and Gedik's research results also indicated that release of nutrients from fish feed occurred rapidly, they believed that uneaten fish feed should be removed quickly to avoid nutrient enrichment 32 .
Under same fish feed dosage, TP (TDP or PO 4 3− -P) concentrations in treatments with HT and HP feed are 1.33~1.66 times higher than those of ZT, which is not consistent with their nutritional indicator of TP (in Table 1). This may be because the TP indicator in these feeds just follows "the lower limit rule". Calculated results shows that average TP concentrations are 1.97, 1.96 and 1.28 mg L −1 , average TDP concentrations are 1.75, 1.74 and 1.06 mg L −1 , average PO 4 3− -P concentrations are 1.60, 1.59 and 0.91 mg L −1 for HT 0.1 g, HP 0.1 g and ZT 0.1 g respectively, and these concentrations also doubles in treatments with 0.2 g correspondingly. This also implies that both HT and HP feed have much larger capacities in releasing phosphorus nutrients than ZT feed. In addition, significant analysis shows that there is a noteworthy difference in releasing phosphorus nutrients between HT and ZT and between HP and ZT (P < 0.001), while there is no significant difference between HT and HP (P > 0.05). www.nature.com/scientificreports www.nature.com/scientificreports/ Significant analysis also shows that fish feed dosage affects TP, TDP and PO 4 3− -P concentrations quite significantly (P < 0.001), which conforms to Wu et al. 's results 43 .
In Fig. 1(b,d), variations of TPP concentrations with time are quite different from those of TDP. In general, TPP concentrations in HT, HP and ZT groups are quite low and close to each other with the same dosage of fish feed, and all increase firstly and then decrease slightly. Fish feed quality does not have a significant effect on TPP concentrations in general (P > 0.05).
Effects of different fish feed on nitrogen concentrations. Uneaten fish feed is probably the major input of nitrogen to the aquatic environment 35,[56][57][58] , and the nitrogen cycle in aquaculture ecosystem begins with the introduction of protein in fish feed and NH 4 + -N is a by-product of protein catabolism 26 . From Fig. 2, compared with the released process of phosphorus nutrients from HT, HP and ZT fish feed, nitrogen concentrations rise comparatively very slowly and the time to reach nitrogen nutrients equilibrium concentrations is much longer. TN, TDN and NH 4 + -N concentrations increase gradually in about 15 days, and then reach equilibrium in the following days. In addition, it is clearly observed from Figs. 1 and 2 that TN equilibrium concentrations are higher than TP equilibrium concentrations (1.40~5.04 mg L −1 ) in the present experiment. Fernandes et al. also observed that leaching loads of fish feed for the bluefin tuna were slightly high for nitrogen as 26 kg N tonne −1 , but significantly low for phosphorus as 4 kg P tonne −1 25 .
As shown in Fig  www.nature.com/scientificreports www.nature.com/scientificreports/ concentrations of treatments with 0.1 g fish feed. In reality, as shown in Table 1, ZT fish feed also contains the lowest crude protein, which may be due to the reason that ZT fish feed releases the smallest amount of nitrogen. In addition, similar to variations of TPP with time, TPN concentrations in Fig. 2(b,d) also fluctuate in low concentrations in all treatments during the whole period. Meanwhile, TPN concentrations are significantly different among the three different fish feed (P < 0.05).
As shown in Figs Table 2, for example, TDP is 84.48~91.95%, 88.80~94.90% and 80.91~90.93% of TP for HT, HP and ZT respectively. From the results in Table 2, the ratio of PO 4 3− -P to TP and NH 4 + -N to TN are obviously lower than those of TDP to TP and TDN to TN respectively because PO 4 3− -P and NH 4 + -N are only one part of them, respectively. Proportions of PO 4 3− -P and NH 4 + -N are in good agreement with Wu et al.'s results, and PO 4 3− -P and NH 4 + -N have high proportions of TP and TN 43 , respectively. Butz and Ven-Cappell 59 and Kibria et al. 35 also believe that fish feed contained major phosphorus fraction in a labile form, namely, the total phosphorus in fish feed, the more the water-soluble phosphorus. Thus, according to released P (TP, TDP and PO 4 3− -P) and N (TN, TDN and NH 4 + -N) concentrations, we believed that HT contains the most nutrients, HP is next while ZT is the lowest in a comprehensive view. It is consistent with crude protein indicators of fish feed in general, ZT fish feed has the lowest crude protein level at 20%. Thus, based on trade-offs among feed price, feed efficiency, feed cost, feed quality, environmental impacts and so forth in aquaculture operations, we could improve protein bioavailability and design reasonable ratio of protein to energy to save protein and reduce nutrients emission.
Effects of different fish feed on M. aeruginosa growth. Effects of different fish feed on M. aeruginosa densities. Fish feed contributes to abundant nutrient loads as discussed in the above, and it can effectively promote the growth of phytoplankton 28,43,60 . From Fig. 3, in the first few days of the experiment, algal cell densities increase very slowly due to their acclimation in fish feed medium with abundant nutrients in the medium. As    Not only fish feed dosage but also their quality affects algae growth greatly, and the algae densities' rankings in Fig. 3 are in agreement with those rankings of nutrients concentrations generally. The order of algae densities from the three different fish feed is MHT 0.2 g (MHT 0.1 g) > MHP 0.2 g (MHP 0.1 g) > MZT 0.2 g (MZT 0.1 g) during the whole experimental period (Fig. 3), and the corresponding measured maximum algae density is 2526.1 (1278.9), 2042.0 (1016.4) and 1757.2 (595.2) 1 × 10 4 cells mL −1 , respectively. Two kinds of significant difference analysis of algae densities are conducted, namely, including and excluding lag phase, which indicates that the algae densities of MZT are significant different from those of MHT and MHP when excluding lag phase (P < 0.05), while they are not significantly different when including lag phase (P > 0.05), and this may be because the algae density is low and close to each other during the lag phase among the three different fish feed. In addition, fish feed dosage also has a significant effect on algae densities (P < 0.05).
Eutrophication is a major environmental problem induced by aquaculture activities, and algae densities reflect the level of eutrophication. Generally speaking, the lower the algae densities simulated by fish feed, the better the water quality is. Algae densities are coherent with released nutrients concentrations from fish feed and also consistent with nutritional indicators of fish feed in general. Thus, the above results imply that in order to protect aquaculture water environment, "environmentally friendly feed" are needed to both stimulate fish growth greatly and to lessen their effects on the water environment effectually in a balanced way. Meanwhile, new method is greatly needed to decrease the uneaten fish feed when throwing feed to fish manually and the uneaten fish feed also should be removed quickly before it releases nutrients to water.
In our study, both Fig. 3 and Table 3 show that the modified Logistic function can describe M. aeruginosa growth with good accuracy (R 2 = 0.984~0.999) in agreement with the reported results 49 . Consistent with measured algae densities, N max and N ave (time-averaged algae density) of different fish feed are also in the order of MHT > MHP > MZT with the same fish feed dosage, and N max and N ave also increase with increasing dosages of fish feed. Specifically, the fitted N max are 2557. 32 Table 3.
Effects of different fish feed on the growth rate of M. aeruginosa. As shown in Fig. 4(a,b), both measured and computed growth rates in different groups all increase monotonously with time before they reached their maximal values, and then all decrease monotonously, which is consistent with Huang et al. 's study 49 . From Fig. 4(a,b) and correlation analysis, the computed growth rates agree reasonably well with measured ones with correlation coefficients (R) of 0.911, 0.954, 0.825, 0.970, 0.970 and 0.975 for MHT 0.1 g, MHP 0.1 g, MZT 0.1 g, MHT 0.2 g, MHP 0.2 g and MZT 0.2 g respectively, and all correlations are significant (P < 0.001). Although the analysis of significant difference shows that the fish feed quality does not have significant effects on growth rate (P > 0.05), maximal calculated growth rates (µ ′ cmax ) and averaged calculated specific growth rates of MHT are obviously the most, next those of MHP while those of MZT the smallest, as shown in Table 3.
Effects of different fish feed on the specific growth rate of M. aeruginosa. Correlation analysis between measured and computed specific growth rates is conducted, the correlation coefficients (R) between measured and computed specific growth rates in all groups range from 0.713 to 0.841 (P = 0.002~0.037) except in group MZT 0.1 g with R = 0.579 and P = 0.188. This indicates that Eq. (3) is reasonably well in describing specific growth rates of algae generally. In Fig. 4(c,d), the computed specific growth rates increase firstly, then decrease in general. In addition, both measured and computed specific growth rates among different qualities' fish feed are quite close with the same fish feed dosage, significant difference analysis also shows that fish feed quality does not influence  Table 3. Parameters of modified Logistic function describing algae growth. a, a constant; r (d −1 ), the intrinsic growth rate; N max (1 × 10 4 cells mL −1 ), the maximum algae density; N ave (1 × 10 4 cells mL −1 ), the average algae density; R 2 , square of correlation coefficient; µ ′ cmax (1 × 10 4 cells (mL·d) −1 ), the maximal growth rate; µ ′ cave (1 × 10 4 cells (mL·d) −1 ), the average growth rate; µ′ cmax (1 × 10 4 cells (mL·d) −1 ), the maximal specific growth rate; µ cave (1 × 10 4 cells (mL·d) −1 ), the average specific growth rate. Data were calculated according to corresponding equations.

Interaction of different fish feed and M. aeruginosa growth on nutrients concentrations.
As discussed in 2.1, different quality of fish feeds has markedly different influence on released nutrients concentrations in general, that further affect algae growth. Wu et al. believe that in the presence of both algae and fish feed, nutrients releases were mainly controlled by fish feed dosage and algae utilization 43 . In the present study, not only fish feed dosage and algae utilization but also fish feed quality is taken into account to study the interaction of different fish feed and M. aeruginosa growth on nutrients concentrations. Figure 5 shows variations of TP, TDP, TPP and PO 4 3− -P concentrations with time in treatments with algae. From Fig. 5(a,b), some fluctuations of TP concentrations in treatments with algae were observed during the whole experimental period, and TP concentrations is not related to algae growth (R = −0.  -P concentrations in the three different fish feed, they are actually quite different, and the most appears in MHT, and MHP is next while MZT is the smallest in general.

Interaction of different fish feed and M. aeruginosa growth on phosphorus concentrations.
As shown in Fig. 5(b,d), TPP concentrations increase rapidly in the first 13 days then increase slowly in the following days. This is mainly related to initially released large quantities of phosphorus nutrients and uptake of PO 4 3− -P nutrients by algae. In Huang et al. 's 28 study, TPP concentrations are closely related to the algae biomass, namely, variations of TPP concentrations with time are similar to those of algae biomass. Correlation analysis in the present study also shows that there are positive correlations between TPP concentrations and algae densities in most groups (R = 0.710~−0.917, P < 0.002) expect group MZT 0.2 (R = 0.349, P = 0.192). This is because TPP concentrations do not increase and even decrease since day 11 in group MZT 0.2. Meanwhile, consistent with algae density, the order of TPP concentrations is also MHT 0.2 g (MHT 0.1 g) > MHP 0.2 g (MHP 0.1 g) > MZT 0.2 g (MZT 0.1 g), and the corresponding average TPP concentrations is 1.94 (0.89), 1.78 (0.70) and 1.47 (0.52) mg L −1 . However, quality or dosage has no significant effect on TPP concentrations in general (P > 0.05), which maybe because the difference of algae density among different quality of fish feeds are not significant especially during lag phase.
In addition, it is needed to point out that TP includes both extracellular P and intracellular P in treatments with algae, thus variations of TP concentrations with time in treatments with and without algae should be similar. However, we noted that, influenced by algae utilization and algae deposition, TP concentrations in groups with algae fluctuate and are lower than those in group without algae 43,62 .
Interaction of different fish feed and M. aeruginosa on nitrogen concentrations. From Fig. 6(a,c), TN concentrations in treatments with algae increase gradually in the first 15 days and then keep stable in the following days, the variations are consistent with those in treatments without algae. Meanwhile, algae densities are also related to TN concentrations released from fish feed in general (R = 0.616~0.908, P < 0.011), while the correlation coefficients are low in group MZT 0.2 with R = 0.357 (P = 0.175). Fish feed quality has significant influence on TN concentrations (P < 0.05), and the order of TN concentrations in groups is MHT > MHP > MZT in Fig. 6. Maximal TN concentrations are 11.00, 7.56 and 6.09 mg L −1 , the average values are 9.10, 6.09 and 4.57 mg L −1 for MHT 0.1 g,  (Fig. 6(a,c)).
In Fig. 6(b,d), TPN concentrations increase gradually in the first 20 days and then reach stable concentrations with time going in MHT, MHP and MZT. Consistent with TPP, TPN concentrations also have positive correlation with algae densities during the whole experimental period (R = 0.744~0.920, P < 0.001). Also, the order of TPN concentrations at the same time among different treatments is MHT 0.2 g (MHT 0.1 g) > MHP 0.2 g (MHP 0.1 g) > MZT 0.2 g (MZT 0.1 g), and the corresponding average TPN concentrations are 10.95 (6.77), 9.30 (4.50) and 7.19 (3.61) mg L −1 . However, fish feed quality has no significant influence on TPN concentrations among all treatments with algae (P > 0.05), and this may be also because fish feed has no significant influence on algae densities when including the data in the lag phase (P > 0.05, n = 16).
Due to the effect of algae growth, the fractional composition in treatments with algae, as shown in Figs. 5 and 6 and Table 2, is different from that without algae, as shown in Figs. 1, 2 and Table 2. For example, due to the algae utilization, the ratio of TDN:TN is 8.88~12.64%, 9.12~17.48% and 6.22~17.80% for MHT, MHP and MZT www.nature.com/scientificreports www.nature.com/scientificreports/ respectively (in Table 2), which are largely lower than those of HT, HP and ZT mainly because of selective uptake of nutrients by algae.
Effectsof different fish feed on nutrients utilization by M. aeruginosa. Nutrients releases from HT, HP and ZT fish feed are different as discussed in 2.1, which further affect algae growth and nutrients utilization. In order to study the interaction between different fish feed and M. aeruginosa growth, nutrients utilization by algae is also explored. In Huang et al. 's 49 and Goudar et al. 's 50 studies, Logistic function is also used to simulate nutrients consumption versus incubation time and as follows: in which t is the incubation time (d), ΔC (i.e. △TDP, ΔPO 4 3− -P, ΔTDN and ΔNH 4 + -N) is consumed nutrient concentrations (difference of nutrients concentrations between without and with algae) at time t (mg L −1 ), ΔC max is the maximum consumed nutrient concentrations,  r C is the consumed rate constant (d −1 ) and Δ a C is a constant.
As shown in Fig. 7, △TDP, △PO 4 3− -P, △TDN and △NH 4 + -N concentrations increase rapidly until it reaches their respective maximal consumed concentrations, then they remain stable. From Fig. 7 and Table 4 Table 4, it can also be founded that maximal calculated consumed TDP, PO 4  www.nature.com/scientificreports www.nature.com/scientificreports/ 3.33, 1.99, 1.39, 1.38 and 0.75 mg L −1 , respectively, this conforms to measured results. ΔC max increases with increasing maximum density of M. aeruginosa (N max ), which indicates that more algae need more nutrients to grow (Fig. 7). Correlation analysis also shows that there is a positive correlation between algae density and consumed TDP, PO 4 3− -P, TDN as well as NH 4 + -N concentrations with R 2 = 0.738~0.949, R 2 = 0.840~0.955, R 2 = 0.816~0.949, R 2 = 0.879~0.977, respectively. Meanwhile, fish feed quality has statistically significant effect on nutrient utilization if excluding the lag phase in general (P < 0.05) but no significant effect if including the lag phase (P > 0.05), and this is also because the algae density is close during the lag phase with different fish feed. In sum, the result implies that the nutrient utilization is dependent not only on the fish feed dosage but also on their quality.
In Tijani et al.'s study, both nitrogen and phosphorus utilization display a significant increase during the first 2~21 days, then enter a stationary phase on the 21st day and the utilization has an initial 48 h lag phase 64 . However, in the present study, as shown in Fig. 7  In treatments with M. aeruginosa (MHT, MHP and MZT), fish feed quality affects TP and TN concentrations significantly in general (P < 0.05). In addition, for most forms of consumed nutrients concentrations, the differences among all treatments excluding the lag phase are significant in most comparisons (P < 0.05), which suggests that the nutrient utilization is dependent on not only fish feed dosage but also fish feed quality. Maximum M. aeruginosa densities and growth rates in different fish feeds are also quite different, their orders are MHT > MHP > MZT with the same dosage.   Table 4. Parameters in Logistic function of consumed nutrients concentrations. Δ a C , a constant; Δ r C (d −1 ), the consumed rate constant; ΔC max (mg L −1 ), the maximum consumed nutrient concentrations; R 2 , square of correlation coefficient; ΔC ave (mg L −1 ), the average consumed nutrient concentrations;. MHT 0. www.nature.com/scientificreports www.nature.com/scientificreports/ In our study we experimentally studied the environmental effect of fish feed through incubator experiments without fish as a first try. Our preliminary results demonstrated that fish feed quality should be considered in terms of water environment protection.