Greenhouse Gas Emissions Associated with Nile Tilapia (Oreochromis niloticus) Pond Fertilization in Western Kenya

In the recent past, fish farming has gained great prominence in Kenya as the country straggles to meet food security. Nile tilapia (Oreochromis niloticus L.) farming has attracted the most demand, with the use of manure to enhance primary productivity in fish ponds being encouraged as a form of increasing productivity and returns on investment. The objective of this study was to understand the role of Nile tilapia farming in greenhouse emissions (GHGEs) in the region. Generally, there is paucity of such information originating from sub-Saharan Africa. Here, we report the levels of methane (CH4), carbon dioxide (CO2), and nitrous oxide (N2O) emissions from Nile tilapia fish ponds fertilized with organic and inorganic fertilizers. We also try to establish if there exists any relationship between GHGEs and physicochemical parameters (PCPs). The methane fluxes ranged from 0.001 to 0.043°mg·m−2h−1 in UF ponds, 0.005 to 0.068°mg·m−2h−1 in IF ponds, and 0.001 to 0.375°mg·m−2h−1 in OF ponds. The findings show that the fluxes were significantly different (P < 0.05). Mean fluxes of CO2 did not show significant difference among the treatments (P > 0.05), ranging from −0.180 to 1.40°mg·m−2h−1 in UF ponds, −0.020 to 1.101°mg·m−2h−1 in IF ponds, and −0.049 to 1.746°mg m−2h−1 in OF ponds. N2O mean fluxes were not significantly different (P > 0.05), ranging from −0.628 to 0.326°µgm−2h−1 in UF ponds, −0.049 to 0.187°µgm−2h−1 in IF ponds, and −0.022 to 1.384°µgm−2h−1 in OF ponds. UF had a mean flux of −0.003 ± 0.175°µgm−2h−1, IF had a mean flux of 0.032 ± 0.056°µgm−2h−1 and OF had a mean flux of 0.093 ± 0.324°µgm−2h−1. There was significant difference in the carbon to nitrogen (CN) ratio among the fertilization treatments (P < 0.05), whereas temperature, pH, dissolved oxygen, and conductivity showed no significant difference among the fertilization treatments (P > 0.05). The study observed that fertilization of Nile tilapia ponds significantly increases the release of CH4 emission and the CN ratio. Temperature, conductivity, and CN positively correlated with CH4, CO2, and N2O emissions. Dissolved oxygen showed a negative correlation with CH4 and CO2 emissions while negatively correlated with N2O emissions. The study identified the use of OF as a potential form of fish farming that promotes the emission of GHGEs and calls for adoption of sustainable technologies for the management of organic and inorganic fertilizers before their use in pond fertilization.


Introduction
Global warming has emerged as a major global challenge, and all governments in the world have been called to action [1]. One of the actions, governments are to undertake, is to cut greenhouse gas (GHG) emissions. To achieve this, governments need to understand which sectors are responsible for how much GHG emissions. In this study, we aimed at estimating the amount of methane (CH 4 ), carbon dioxide (CO 2 ), and nitrous oxide (N 2 O) gases emitted into the atmosphere from Nile tilapia farming. Generally, there is little information on how fsh farming infuences GHG emissions in the region. In the last 15 years, fsh farming in Kenya has increased 10-fold, increasing from 271 ha to over 2500 ha [2], therefore calling for a need to a better understanding of the types and quantity of the emission. Te agricultural sector has been estimated to be the largest source of GHG emissions of all sectors in Kenya [3], and about 40% of total national emissions in 2015 were from this sector alone [4].
Oreochromis niloticus remains the most widely cultured fsh species in the world because of its high consumer acceptance, high commercial value, and fexible eating habits that allow for the use of high levels of plant protein, making it easier to adapt easily to most fsh culture systems with increased intensifcation [5]. However, longstanding hurdles to enhancing Nile tilapia production in the supply of quality fsh feeds remain a challenge [6]. Supplementary feeds are the most expensive input in intensive and semi-intensive cultures [7]. Combining fertilization and supplemental feeding has been shown to reduce production costs [8]. According to [9], fsh pond manuring is used in fsh farming for the intensifcation of fsh production by balancing the ratio between carbon and other nutrients. Odinga et al. [10] demonstrated that fsh pond fertilization, whether with organic or inorganic fertilizer improved the growth of Nile tilapia in smallholder earthen ponds. At the same time, it has been shown that GHG emissions tend to increase as an aquaculture system moves from being extensive (untreated or partially fertilized) to semi-intensive (fertilized and/or partially fed) and intensive (completely fed and fertilized) [11]. Tese fertilizers are introduced in the fsh pond water, which forms the immediate environment for fsh and is very important as it forms the basis of fsh metabolism [12].
Te rapid expansion in aquaculture has received negative reputation for its associated environmental impacts [13]. Tere have been calls for sustainable intensifcation (SI) in order to produce more using less in a bid to increase productivity and environmental impacts arising from aquaculture [14]. According to [15], aquaculture remains one of the best innovative agricultural options for achieving food security under changing climate, but with the efects of climate change in aquaculture becoming more prevalent, research eforts are being directed towards developing and validating climate smart aquaculture (CSA) technologies, innovations, and management practices (TIMPs) for sustainable fsh production [16]. Terefore, in order for the region to mitigate GHG emissions, there is a need to understand how existing systems perform. Tis study therefore identifed and quantifed sources of GHG emission from aquaculture systems in Western Kenya.

Study Area.
Te study was carried out in Kakamega County, as shown in Figure 1 (0.2827°N, 34.7519°E), found in Western Kenya. It covers an area of 1,395 km 2 with a population of 1,868,000 [17]. Kakamega is mainly tropical, with variations due to altitude with an average elevation of 1,523 meters [18]. It experiences heavy rainfall all year round, with two seasons, namely, the long rains (April to July) and short rains (August to November). Rainfall ranges from 156 to 663 mm/month with a temperature range of 9.95-26.0°C [18]. Te coldest month is July, with an average of 9.95-12.0°C, whereas the hottest season is experienced between December and February, with a temperature range from 24.5-26.0°C [18].
Kakamega County is well endowed with a vast water resource that can be harnessed for fsh farming. In 2018, Kakamega County had 7,939 fsh farmers operating 8,540 fsh ponds covering an area of 2,260,945 m 2 [19]. In the same year, 1,730,000 fngerlings of Nile tilapia and catfsh fngerlings valued at Kshs. 13 million were stocked in the county. Fish weighing approximately 1,600 t, valued at about KES. 500 million, were harvested and sold in the same year [19]. Te large numbers of smallholder fsh farmers that fertilize their fsh ponds using animal manure cannot be ignored, and this calls for the need to assess their contribution to the environment, apart from fsh production.

Study Design.
Our study focused on the emissions during pond post application. An experimental study involving three fsh farms in Kakamega County, each with three ponds measuring 300 m 2 and a depth of 1 m, was adopted. Te ponds were purposively selected in a randomized complete block design (RCBD). Te choice of fertilization method in this study was based on a survey conducted among the fsh farmers of Kakamega County, which revealed that 87.6% of fsh farmers fertilized their ponds, while 12.4% did not fertilize their ponds. Among the farmers fertilizing their ponds, 64% used animal manures (41% chicken manure, 49% cow dung, and 11% other manures), while 36% used inorganic fertilizers. On each of the farms, the three ponds consisted of an unfertilized pond (UF), inorganic fertilizer fertilized pond (IF), and organic manure fertilized pond (OF). Stocking and feeding of the fsh were done from June 2021 to December 2021. Fish ponds used in this study were earthen Nile tilapia ponds, which were well-constructed and retained water throughout the cycle, with initial pond flling at the beginning of the cycle, followed by weekly water top-up to compensate for evaporation.

Sampling
2.3.1. Sampling of Manure and Feeds. Te manure and feeds were aseptically sampled into 250 ml autoclaved propylene bottles by picking from 5 diferent points and mixing for uniformity. 100 g of each sample manure and compost were transferred into the bottles for analysis of the carbon to nitrogen ratio. Te samples were transported to the laboratory in a cooler box packed with ice. Tis was done in three batches during the study.

Sampling of Water.
Te water samples were collected aseptically using presterilized 250 ml glass bottles, which were submerged 15 cm to 20 cm below the water with the mouth facing upwards. Te glass bottles were thoroughly washed with detergent, rinsed with tap water and soaked for 6 hours in 20% (v/v) of hydrochloric acid, and rinsed with reagent water to remove all traces of organic materials which could cause CN contamination. Samples were taken by flling the bottles to the top to exclude air. Bottles with water samples were labeled accordingly. Water samples were placed in a cool box packed with ice and then transported to the laboratory for total carbon and total nitrogen analysis on a monthly basis.

Sampling of Gases.
Every time gases were sampled, supporting measurements of chamber temperature and ambient temperature were performed using Wertheim EN 13485 thermometer and atmospheric pressure using a phone-installed barometer. Static chambers were used to capture concentrations of CO 2 , CH 4 , and N 2 O. Concentrations were captured by deploying 3 chambers per pond on a monthly basis according to [20]. Te concentrations were captured at each of the three points at intervals 10 minutes for 30 minutes using syringes and pooled in 60 ml previously evacuated glass vials sealed with butyl rubber septum ( Figure 2), for transportation to the lab for analysis [21]. Te pooling was done to form a composite air thereby overcoming spatial heterogeneity as recommended in [22]. Four gas samples were taken, sequentially from time zero to 30 min (at intervals of 10 minutes) to be able to calculate the fux rate [23]. Sampling was done between 10 am and 12 am when the temperatures refected the daily averages [21].
Te chambers were locally fabricated according to recommendations in [24]; typically, chambers cover an area of 0.1-1 M 2 , should have a vent tube to maintain pressure equilibrium, using Styrofoam material (silver in color), and should be refective to reduce solar radiation and able to foat. Our chambers were rectangular, measuring 0.56 M length, 0.38 M width, and 0.1 M height ( Figure 3).

Pond Preparation.
Te nine 300 m −2 earthen fsh ponds were drained of their contents (water, fsh, and plants). Tey were then limed using quick lime (CaO) at a rate of 200 g·m −2 (one of) and left for 7 days before fertilization.

Fertilization.
Te ponds were flled with water as the fertilization regime was performed, i.e., UF being unfertilized, IF fertilized using di-ammonium phosphate, applied weekly at 2 g·m −2 /week whose NPK ratio was 18 : 46 : 0 (soaked in water and broadcasted in the pond water), and OF, whose organic manures used were chicken manure (carbon to nitrogen ratio was 13.76) and cow dung manure (carbon to nitrogen ratio was 13.83), applied at the rate of 20 g·m −2 /week. Te ponds were left for 7 more days before stocking with fngerlings, after which fertilization followed weekly.

Stocking of Fingerlings.
All the ponds were stocked with 1,000 all-male Nile tilapia fngerlings (Oreochromis niloticus L.) from the same hatchery, with an average weight of 0.5 g and an average length of 1.9 cm (3 fsh per m −2 plus 100 mortality allowance). Te fngerlings were stocked at 6 weeks of age after hatching ( Figure 4). Te fngerlings were sourced from Labed Cash Hatchery Ltd.     Te fsh were fed at 5% average body weight in the frst two months, 3% average body weight in the next two months, and 2% average body weight in the last three months with fsh feed with the following proximate compositions listed in Table 1. Te carbon to nitrogen ratio of the feed that was fed to fsh was 6.82.

Physicochemical
Parameters. Water quality parameters, including temperature, dissolved oxygen, pH, and conductivity, were measured in situ using a Hydrolab MSIP-REM-HAH-QUANTA (USA) at three points of each pond (inlet, middle, and outlet). Tis was done on a monthly basis in all the replicates between 9 am and 11 am.

Carbon to Nitrogen Ratio in Pond Water Samples.
Total dissolved C and N in water samples were determined by TOC/TN-analysis on a TOC-L CPN model Shimadzu analyzer. TDN was quantifed by high temperature catalytic oxidation (HTCO) at 680°C, and a platinum catalyst was used to complete the oxidative conversion of all forms of C to CO 2 and all forms of N to NO and NO 2 . NO and NO 2 then reacted with O 3 , producing an excited state of NO 2 (NO 2 * ). Upon returning to the ground state, light energy is emitted which is quantifed by chemiluminescence detection. Te content of total organic carbon (TOC) was then determined by the diference method (TOC � TC−TIC) as well as the addition method (TOC � POC + NPOC), where TOC was measured as nonpurgable carbon (NPOC) where after in-syringe acid addition of acid, the samples were purged with synthetic air to release inorganic carbon (TIC). Te 100 ml water samples for total C and N analysis were fltered on a GF/F flter (0.45 μm) using a vacuum pump (pressure 200 mm Hg). GF/F flter paper was frst heated in an oven at the temperature 105°C for 2 hours to attain constant weight. Te contents were then weighed on a scale (Mettler Toledo-XP205) and the weight was given as mgl −1 of total carbon or total nitrogen. Te carbon to nitrogen ratio was calculated by the following equation: where a � volume of the titre HCl for the blank, b � volume of titre HCl for the sample taken, and al � aliquot of the solution taken for analysis. For organic carbon, 0.3 g of the dried sample was digested using 7.5 ml sulphuric acid and 5 ml aqueous potassium dichromate (K 2 CR 2 O 7 ) mixture. Unused K 2 CR 2 O 7 was titrated against ferrous ammonium sulphate to the endpoint denoted by a color change from green to brownish. Te concentration of organic carbon was calculated according to [25] in the following equation: where Vb � volume in ml of 0.2 M ferrous ammonium sulphate used to titrate reagent blank solutions; Vs � volume in ml of 0.2 M ferrous ammonium sulphate used to titrate sample solution, and w � 12/4000 mili-equivalent weight of C in grams.  4 , and CO 2 was passed through a methanizer. Te courier gas used was N 2 with a fow rate of 20 ml·min −1 for the three gases (CH 4, CO 2 , and N 2 O). Te concentrations of the gases in the glass vials were calculated from the peak areas of standard gases with known concentrations. Headspace gas concentration changes over time were plotted to produce a slope, and the slope was used to calculate fux [23] in the following equation:  Te lowest pH of 4.8 was recorded in IF, while the highest (8.9) was recorded in OF. pH in UF (7.70 ± 0.66) remained highest and pH in IF (7.46 ± 1.26) lowest; however, there was no signifcant diference in pH based on the fertilization (P � 0.697).

Physicochemical Characteristics of Pond Water from
Te lowest dissolved oxygen of 2.20°mgl −1 was recorded in IF, while the highest (7.49°mgl −1 ) was recorded in OF. Te highest overall dissolved oxygen (4.79 ± 1.52°mgl −1 ) was recorded in IF treatments and lowest in UF (4.23 ± 1.36°mgl −1 ). Tere was no signifcant diference in dissolved oxygen among the treatments (P � 0.529).
Te lowest CN ratio of 10.06 and highest (29.33) were recorded in OF. Te highest average CN ratio of 19.22 ± 6.04 was recorded in IF ponds. Tere was a signifcant diference in the CN ratio based on fertilization (P � 0.025), with UF being signifcantly lower than IF and OF ( Table 2).
Tere was a steady increase in CH 4 ( Figure 5), CO 2 ( Figure 6), and N 2 O (Figure 7) with time. When the growth of GHG data was subjected to CH 4 , CO 2, and N 2 O, time regression, the greatest increase in emissions of CH 4, CO 2, and N 2 O with time was observed in the OF ponds.

Discussion
Te study was set to investigate the efect of fsh pond fertilization on CH 4 , CO 2, and N 2 O emissions. Our study revealed that on average, the fsh ponds emitted CH 4 , CO 2, and N 2 O, except UF which did not emit N 2 O. Organic fertilization of fsh ponds appears to result in high GHG fuxes, because of the high nutrient load exhibited by CN. High organic matter decomposes, thereby utilizing oxygen and creating anaerobic conditions at the pond bottom [26], enabling methanogens to consume the matter, and releases CH 4 . At the same time, on the upper aerobic layers of the pond, methanotrophic bacteria oxidize CH 4 to produce CO 2. In this study, there is a strong positive correlation between CH 4 and CO 2 (r � 0.688; P � 0.002), pointing to the possible fact of methanotrophic bacterial activity in the ponds. Furthermore, the availability of more nutrients supplies extra substrates for N 2 O production [27], which is also important in primary production. Te UF, which was typically limited in substrates, may have infuenced the average N 2 O production, as these ponds acted as sinks and not as emitters of N 2 O.
Nevertheless, the study recorded lower daily fuxes for CH 4 m −2 d −1 than those recorded in tropical and subtropical surface water bodies in previous studies [28][29][30][31][32]. It is possible that this diference could be attributed to the quality of organic loads in the water bodies [33]. Tese fertilizers are introduced in the fsh pond water, which forms the immediate environment for fsh and is very important as it forms the basis of fsh metabolism [12]. Te breakdown of the diferent organic loads consumes oxygen, thereby creating anoxic environments that favor CH 4 emissions in the pond bottom as CO 2 is released on the upper aerobic pond layers when CH 4 is oxidized [26].
Water quality plays an important role in GHG emissions. Te increase in temperature has been shown to increase CH 4 emissions resulting from enhanced methanogen activity [26,34]. Additionally, increase in temperature reduces dissolved oxygen and eventually increases CH 4 emissions [35]. CO 2 fuxes signifcantly correlated positively with temperature, an indication that the mineralization of organic matter remains a key factor in determining the release of CO 2 as heterotrophy was higher than autotrophy [36]. Te negative correlation shown between pH and CO 2 is attributed to the fact that as CO 2 accumulates in water, it leads   Te Scientifc World Journal to a fall in pH levels [37]. According to [32], the acidic water favors a high fux of pond water to the atmosphere. Te addition of fertilizer led to an increase in NO 2. Tis could be as a result of the supply of more nitrogen substrate for decomposers leading to higher NO 2 emissions. However, when the fuxes of the three GHGs were compared, NO 2 had the lowest fuxes. Tis could be explained by the fact that there exist strict anoxic conditions in the pond, which enhance denitrifcation, leading to the end product of nitrogen gas instead of NO 2 [38]. N 2 O fuxes positively correlated with temperature and negatively correlated with dissolved oxygen. Higher water temperature enhances the denitrifcation process through oxygen demanding metabolic pathways [39,40].
Te CN ratio was noted to be a driver and determinant of GHG emissions owing to its strong signifcant positive correlation with CH 4 , CO 2, and N 2 O. Te CN ratio infuences the mineralization and immobilization processes.
It can be noted that the created wetlands (artifcial wetland in our case) provide food for fsh through aquaculture as explained in [41]. Tis creation of artifcial wetlands is a land use change, which leads to the loss of the soil's ability to sequestrate carbon, leading to more emissions. However, the emissions in our study are much lower than the emissions from crop felds. Te mean fuxes reported by [42] for most of the East African soils were 1.2-10.1 kg·C·ha, 3.5-15.9 g·C·m −2 h −1 , and 2-62 µg·N·m −2 h −1 for CH 4 , CO 2 , and N 2 O, respectively. Tis shows that fsh ponds have not been used long enough to accumulate a higher organic load compared to agricultural soils, which have been used for long periods. Tis is also supported by observations made by Singh et al. [43], who did a comparison of natural and artifcial ponds in India and noted lower CH 4 fuxes in artifcial ponds. Tis was because the artifcial ponds had lower nutrient concentrations and less sediment. Wetlands accumulate organic carbon in the soil due to anaerobic conditions leading to slower decomposition compared to crop lands [44]. During the cultivation of agricultural land, the soil organic carbon is subjected to more favorable conditions for decomposition (aerobic), hence enhancing more emissions [45].
Tough semi-intensive Nile tilapia production system using fertilization increases GHG emissions, it is important to note that the emissions produced in the production of other related animal protein foods are much higher compared to Nile tilapia which is 1.58 kg CO 2 [46]. 45.54 and 2.4 kg CO 2 were emitted in the production of 1 kg of mutton and milk, respectively [46], while 2.179 was emitted to produce 1 kg eggs [47]. Tese observations agree with [15]   Te Scientifc World Journal that aquaculture could ofer a good alternative for achieving food security under the changing climate.

Conclusions and Recommendations
Te study assessed methane (CH 4 ), carbon dioxide (CO 2 ), and nitrous oxide (N 2 O) emissions and physicochemical parameters arising from Nile tilapia fsh ponds fertilized with organic and inorganic fertilizers. Te three fertilization treatments signifcantly afected the CN ratio and CH 4 emissions. However, fertilization treatments had no signifcant efect on CO 2 and N 2 O emissions, temperature, pH, dissolved oxygen, and conductivity. Even though the fertilization treatments were not signifcant for CO 2 and N 2 O, there were much higher fuxes of CO 2 and N 2 O in IF and OF than in UF. Te authors conclude that fsh pond fertilization increases emissions of greenhouse gases. However, it is good to note that Nile tilapia production produces much less emissions compared to other animal proteins and remains a viable way to ofer food security, nutrition, and income to the smallholder fsh farmers in the changing environments.
Since a large number of farmers depend on organic pond manuring for improvement in production, there is a need for developing technologies of fsh pond manure processing and management to ensure fsh production units in the region.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
SAO designed the research methodology including fabrication of local gas chamber and analytical framework, collected samples, analyzed data, and wrote the paper. AS and HL provided guidance on the study design, methodology, and analysis and assisted in paper writing. GW provided advice on gas chamber designing and gas collection and analysis. Te submitted version of the manuscript has been approved by all the authors.