Study area

The study was carried out in four ornamental fish farms located in the Apatlaco River sub- basin in the Morelos State, Mexico. The coordinates of each farm and how they were identified are as follows:

Farm A: 18° 38' 53.9" N and 99° 13' 13.2" W

Farm B: 18° 38' 52.3" N and 99° 13' 10.4" W

Farm C: 18° 38' 55.9" N and 99° 13' 09.4" W

Farm D: 18° 39' 20.4" N and 99° 12' 00.9" W

The study comprised two seasons in 2017, corresponding to the dry season during the months of February to March and the rainy season during June and July.

Sampling

Samples of water, feed and organisms of the species Poecillia sphenops, commonly known as molly, were collected in each of the units; this species was considered because it represents one of the most demanded and produced in the region. To evaluate the health status of the fish and determine the sample size, the tables proposed by the Manual of Diagnostic Tests for Aquatic Animals were used, considering fish with normal clinical appearance and the volume of production in each unit (around 2000 organisms); in addition to a prevalence of 10% and a reliability of 95%, so 27 organisms were included in each unit per visit (^{OIE, 2008}).

The organisms were collected randomly from each of the ponds and transported alive with continuous aeration, in plastic bags with water from the same ponds.

Water was collected from the P. sphenops production ponds at each of the farms; water samples were also collected from the Las Estacas ditch (supplying water to farms A, B and C) and from the Apatlaco River ditch (supplying water to farm D). Sampling followed the stipulations of the Mexican Official Standards ^{NMX-AA-003-1980} and ^{NOM-001- ECOL-1996}. Three single samples were obtained from the same point, collected aseptically using sterile bottles; they were transported and kept at 4 ºC (^{FAO, 2011}).

Finally, the collection of feed samples in each of the production units was done in triplicate and aseptically, using sterile bags for transportation (^{NOM-109-SSA1-1994}).

All samples were collected weekly during the seasons studied, and were transported to the Live Food Chemistry Laboratory of the Autonomous Metropolitan University, Xochimilco Unit, for analysis.

Microbiological analysis

The analysis of the organisms was approved by the Ethics Committee of the Biological and Health Sciences Division of the Autonomous Metropolitan University. Organisms were euthanized with an overdose of benzocaine, using the immersion method (^{AVMA, 2013}); subsequently, under aseptic conditions, they were dissected to expose the internal organs and extract the viscera, which were homogenized in sterile 0.89% saline solution (^{OIE, 2008}). For the case of feed and water samples, 1 g^-1 and 1 mL^-1 of each was analyzed, which were homogenized in 9 mL of sterile saline solution (0.89%) (^{Prado et al., 2013}).

In all cases, three serial dilutions were performed, which were inoculated in triplicate on BD Bioxon® brand brain heart infusion (BHI) agar (214700), BD Difco™ brand thiosulfate citrate bile sucrose (TCBS) (265020) and BD Bioxon ® brand eosin and methylene blue (EMB) (210600) plates and incubated at 35 ± 2º C for 48 hours under aerobic conditions. After this period, the colonial morphology was characterized and the colony forming unit (CFU) count was performed using a colony counter and selecting those plates with between 30 and 300 colonies (^{Velasco & Tapia, 2014}).

The following formula was used to determine the number of CFU mL-1 or CFU g-1 (^{NMX- AA-003-1980} and ^{NOM-001-ECOL-1996}):

Where:

UFC mLóg=N×IDV

N: average number of colonies ID: inverse of dilution

V: volume of inoculated sample

For bacterial identification, a differential selection of colonies was made according to morphology, so that they would be representative of the different types of colonies recovered. By successive reseeding, the bacteria were isolated and their purity was verified by Gram staining (^{Velasco & Tapia, 2014}). Bacteria were identified by biochemical tests and using a commercial bacterial identification system using API® galleries (20 E; 20 NE; 50 CH and 50 CHB) (BioMerieux, Mexico).

Coliform analysis

Coliform counts were determined following the methodology proposed by ^{NMX-AA-042- SCFI-2015}, using Bioxon® lactose broth (211700) and Difco™ brilliant green bile broth (273000) as culture medium. The incubation temperature was 35º C for 48 hours for the presumptive test; while the confirmatory test for total coliforms was incubated at 37º C for 48 hours, and for fecal coliforms at 44º C for 24 hours. The analyses of each sampled point were performed in triplicate and the results were expressed as MPN per 100 mL-1 of water.

Physicochemical parameters

Parameters such as pH, temperature, dissolved oxygen (D.O.), salinity and total dissolved solids (TDS) were measured in situ (^{NMX-AA-008-SCFI-2011}), using a multiparameter probe model HI 98194 (HANNA®, Mexico). Chemical parameters such as nitrogen (N) and phosphorus (P) were analyzed with a multiparameter photometer model HI 83203 (HANNA®, Mexico), following the manufacturer's instructions for each case.

Biological oxygen demand analysis

The determination of biological oxygen demand at five days (BOD5) was carried out using the Winkler method determined by ^{NMX-AA-028-SCFI-2001}, performing the tests in triplicate. The initial and final dissolved oxygen concentration was measured with a HI 9146 model oximeter (HANNA®, Mexico).

Statistical analysis

The data were subjected to a descriptive analysis; subsequently, to determine differences between sampling sites and monitored seasons, multiple mean comparison tests were applied by analysis of variance and Student's t-test for minimum significant difference (P<0.05), after verifying that the data complied with the assumptions of normality. In the case of microbiological results (CFU mL^-1), the data were transformed to Log 10 to meet the assumptions of normality. All statistical tests were performed using the SYSTAT 13.0® statistical program.

RESULTS

Microbiological water analysis

The microbiological analysis of water allowed the isolation and identification of a total of 10 bacterial species belonging to the group of Proteobacteria and Firmicutes, where both water tributaries presented the greatest bacterial diversity and abundance during the seasons analyzed, including significant differences for most of the bacteria identified in Las Estacas ditch, with respect to the farms that are supplied by it, as can be seen in Table 1. Likewise, a change in the composition of the microbiota was observed during the rainy season in Las Estacas ditch and the farms supplied by it, displacing species such as Burkholderia gladiolii and Bacillus thuringiensis; on the contrary, the species Vibrio parahaemolyticus and Aeromonas hydrophila were recorded. It should be noted that B. gladiolii was only recorded in Las Estacas ditch during the dry season, while Pseudomonas luteola was reported only in farm A during the rainy season. In relation to the group of Enterobacteriaceae, Klebsiella sp. and Escherichia coli showed a slight increase in the CFU mL^-1 count during the rainy season in Las Estacas ditch; however, no statistical difference was determined for E. coli (p>0.05) from one season to the other. Similar behavior was observed in the Apatlaco River ditch, where high bacterial counts were recorded; however, no statistical difference was determined with respect to farm D (p >0.05). Nevertheless, it could be seen that as bacterial counts such as Klebsiella sp. and Bacillus cereus increased, they could be recorded in farm D during the rainy season. In contrast, E. coli was only detected in the Apatlaco river ditch, while A. hydrophila was only present in farm D (Table 1).

Microbiological analysis of fish

From the microbiological analysis of the organisms, a total of seven bacterial species were identified for the dry season and five for the rainy season, with the greatest diversity in farm B during both seasons (Table 2). Of the total species recorded, only Enterococcus faecalis and Vibrio alginolyticus were found to be different from those reported in the water analysis.

The highest bacterial counts were recorded in farms A and D, with significant differences as shown in Table 2. These data suggest that there is a direct influence of the microbiota of the water on the bacteria present in the organisms, since during the rainy season, it was observed that as bacterial concentrations of species such as Bacillus cereus and Burkholderia cepacea were reduced in the water, these were no longer recorded in the fish, and on the contrary, the species Pseudomonas luteola and A. hydrophila were reported (Table 2).

In relation to E. coli and Klebsiella sp. species, the microbiological analysis of the fish revealed a reduction in their concentration during the rainy season; the highest concentrations were reported in the organisms collected in farm D, with a statistical difference (p=0.0212 and p=0.0009) with respect to the rest of the units (Table 2).

Microbiological analysis of feed

The data obtained from the microbiological analysis of feed are presented in Table 3, which shows the isolation of six bacterial species. Of the total bacteria recovered, Bacillus megaterium was only recorded from feed samples, where the highest counts were found in farms C and D, for which a statistical difference (p=0.0272) was determined with respect to the rest of the farms; likewise, P. luteola and B. cereus were only isolated from samples from farm B during the rainy season.

In general, the analysis of feed from farm D showed the highest CFU mL^-1 counts, including significant differences with the rest of the production units, as shown in Table 3.

Determination of coliforms

Figure 1, section a, shows that in the case of the farms supplied with water from Las Estacas ditch, the highest concentration of total coliforms is reported in farm C during the dry season, while during the rainy season there is a reduction in their concentration; in contrast, Apatlaco ditch and farm D show an increasing trend.

Figure 1 Mean coliform bacteria count values. a) Total coliforms; b) Faecal coliforms * Significant difference (p < 0.05) 

With respect to fecal coliforms, the highest densities were recorded at farm B during the dry season, while Las Estacas canal showed a significant increase (p=0.0482) during the rainy season, which even exceeds the maximum permissible limit (1000 NMP per 100 mL^-1) according to NOM-001-ECOL-1996. On the other hand, farm D and Apatlaco river ditch showed a reduction in the bacterial load of this group during the rainy season, with a significant difference for the Apatlaco river ditch (p=0004) (Figure 1, section b).

Physicochemical parameters

The data obtained for the physicochemical parameters are presented in Table 4, which shows that the water temperature remained between 21.9 ºC and 25.7 ºC during the dry season and between 23.2 ºC and 29.5 ºC during the rainy season, with the lowest temperatures recorded in both canals with respect to the farms; a statistical difference was found for Las Estacas ditch during the dry season (p=0.0360). In contrast, the highest temperatures were recorded in farms B and D. In the case of pH, once again the tributaries had the lowest values, while farms D and A had the highest alkalinity with values of 9.8 and 9.6 units during the dry and rainy seasons, respectively.

In general, the concentration of O.D showed a decrease from the dry to the rainy season (Table 4), with the lowest concentration recorded in farm C; while the highest concentration was recorded in farm B during the dry season (10.2 mg L^-1) and in farm A during the rainy season (7.9 mg L^-1). In no case was there a statistical difference.

Regarding STD, the highest values were reported in farm A during both seasons, even determining a statistical difference in the rainy season (p=0.0007); in contrast, the lowest values were recorded in farm D (0.43 ppt) and in Apatlaco river ditch (0.37 ppt). On the other hand, the nitrogenous forms showed an increasing trend from the dry season to the rainy season in both canals, with the exception of nitrates (NO3) in Apatlaco ditch. As for phosphorus concentrations, a decreasing trend was observed during the rainy season compared to the dry season; however, the opposite occurred in Las Estacas ditch (Table 4).

Finally, DBO5 values are presented (Table 4), where the highest concentration was reported in farm A with 34.5 mg L-1 during the dry season, showing a significant difference (p=0.001), with respect to the rest of the farms supplied by Las Estacas ditch and the same ditch; as well as a reduction in its concentration (17.3 mg L-1) during the rainy season. On the contrary, the lowest concentration was registered in Las Estacas ditch and in Apatlaco ditch during the rainy season, determining statistical differences for the latter (p=0.0279), with respect to farm D which it supplies.