In the last decade, bacterial and viral diseases have affected white shrimp production in several countries of Latin America. This has made some shrimp farmers to use polyculture or crop rotation for tilapia production as an alternative production system ( Watanabe et al. , 2002 ). The polyculture or co-culture of shrimp-tilapia ( Fitzsimmons & Shahkar, 2017 ) has been implemented in many countries for production and the efficiency of this culture strategy has been studied in recent years ( Martínez-Porchas et al. , 2010 ; Yuan et al. , 2010 ; Shahin et al. , 2011 ; Bessa Junior et al. , 2012 ). This strategy has been used in commercial production systems because it improves water quality, controls the growth of phytoplankton, reduces organic matter in effluents and controls outbreaks of diseases (bacteria and viruses) in both tilapia and shrimp ( Yi & Fitzsimmons, 2004 ; Ye et al. , 2011 ). In Latin American countries such as Brazil, Honduras, and Mexico, tilapia are cultured in brackish ponds traditionally used only for shrimp farming ( Alceste et al. , 2001 ; Wang & Lu, 2016 ). The culture of shrimp and tilapia, contributes to improve shrimp' health, production and increases profits in a co-culture system ( Li & Dong, 2002 ; Hernandez-Barraza et al. , 2012 ). Tilapias are omnivorous and are mainly filter feeders, so they improve the quality of water in ponds ( Diana et al. , 1991 ; Ruan et al. , 1992 , 1993 ; Zhang et al. , 1999 ; Tian et al. , 2001a ). In co-culture when shrimp die or are moribund, the tilapias consume those shrimp, limiting cannibalism and mode of disease transmission. Tilapia also contributes to a decrease in the number of pathogenic vectors by consuming small crustaceans and promoting the increase of Gram-positive bacteria ( Watanabe et al. , 2002 ). Studies on ponds used for co-culturing of tilapia-shrimp have shown that with this method the concentration of the bacterium Vibrio harveyi that causes luminous bacterial disease is con­trolled, there is a sustainable stable environment for plankton and improvement in shrimp' survival ( Cruz et al. , 2008 ). In the co-culture of tilapia-shrimp, different aspects such as productivity, feeding, density, time of culture and age of different species, combination of species and culture model have been investigated ( Wang et al. , 1998 ; Hernandez-Barraza et al. , 2012 ; Bessa Junior et al. , 2012 ; Lopez-Gomez et al. , 2017).

In the monoculture of shrimp, the excess nutrients increase phytoplankton, ammonia concentrations, and modify oxygen dynamics ( Midlen & Redding, 1998 ). By incorporating tilapia into the co-culture system, the performance of the shrimp is increased by improving water quality (49.5%), reducing nutrients in effluents (22.6%) and reducing disease outbreak (11.8%) ( Yi & Fitzsimmons, 2004 ; Lopez-Gomez et al. , 2017). Most of the recent studies on tilapia-shrimp co-culture are done indoors in concrete or synthetic ponds, with some of the physical-chemical variables being controlled ( Hernandez-Barraza et al. , 2012 ; Simão et al. , 2013 ; Lopez-Gomez et al. , 2017; Sharawy et al. , 2017 ). However, for those done using rustic ponds under natural conditions in commercial farms, quantitative knowledge of the effect of this cultivation strategy on the aquatic environment is still required to better understand the underlying mechanisms of the dynamics of water quality. The aim of this work is to determine the effects of white shrimp ( Litopenaeus vannamei ) and tilapia nilotica ( Oreochromis niloticus var. Spring) in monoculture and co-culture on water quality variables and production in brackish low-salinity water earthen ponds during dry and rainy seasons.

When comparing water quality variables by sam­pling site and season it was found that there was a significantly ( p <0.05) lower concentration of TP, alkalinity, BOD ₅ and TSS in the inlet water than in the pond and outlet water and in nitrate, TN, BOD ₅ and TSS during rainy (Table 1) and dry seasons (Table 2), respectively. Although there were no significant differences ( p >0.05), there was a tendency to have a higher concentration of TAN in the ponds and water outlet and a lower BDO ₅ in the co-culture than in the monoculture ponds during rainy season (Table 1). During the dry season, the concentration of phosphates had a non-significant ( p >0.05) tendency to be more concentrated in the monoculture ponds.

The tropical storm that occurred in the month of September (Fig. 1) influenced the increase in the values and concentrations of temperature, salinity, TSS and alkalinity of the water in the ponds under study (Fig. 2), in October. There is a tendency to increase salinity and BOD ₅ and to decrease the TSS from the start to the end of the culture cycle during the rainy season. In addition, there was a tendency to have a higher concentration of BOD ₅ , TSS and alkalinity in co-culture ponds than in monoculture ponds. In dry season, there was a tendency to increase the salinity and BOD ₅ concentration towards the end of the culture and to decrease TSS concentration and DO from March to May (Fig. 3). There was a tendency over time to have a higher concentration of BOD ₅ , TSS and alkalinity in the co-culture ponds compared to the monoculture ponds.

In the rainy season, the highest correlation was directly proportional between the alkalinity with salinity and TSS (Table 3), and in the dry season, it was between TSS with salinity and temperature (Table 4). In the dry season, the greatest number of significant correlations of the water quality variables was presented. Temperature correlated significantly ( p <0.05) with most of the studied variables. DO only had a significant correlation ( p <0.05) that is inversely proportional to temperature and salinity in the dry season.

Cluster analysis showed that there are two relations of variables such as the interaction of nutrients and the physico-chemical variables of the water in the system under study (Fig. 4a). Three groups were found between the variables that describe the behavior of the water quality in the system and the BOD ₅ as determinants of the variability of the behavior of the water, through PC analysis (Fig. 4b). The first group showed the relation DO has with the oxidation of nutrients and the effect of its production on the pH. The second group showed the relation of the temperature with the process of dilution-concentration through the alkalinity and TSS variables. The interaction of the nutrients with the rainy and dry seasons was determined by the salinity of the system water.

Cluster analysis classified the sampling stations, in the dry and rainy seasons (Fig. 5a) into two groups. The discriminant analysis confirmed the two groups due to the culture period. Within these two groups in both seasons of the year (Fig. 5b), there were sampling stations of the inlet in the first group and the other two, pond and outlet in the second group. In the sampling seasons of the dry period, there were differences in the water quality of the monoculture ponds and the co-culture pond, whereas in the rainy season there were no differences. The sampling stations of the water outlet of the monoculture ponds were different from all dry and rainy sampling stations.

The shrimp' final weight and biomass were sig­nificantly ( p <0.05) higher in the rainy season than in dry season (Table 5). FCR was significantly ( p <0.05) lower in co-culture than in monoculture and survival was significantly higher in co-culture than in monoculture. No significant ( p >0.05) differences were found between the treatments in initial weight, total weight gain and SGR. Food consumption and FCR were significantly ( p <0.05) lower in tilapia in the rainy season. All the other production variables did not have significant ( p >0.05) differences in the production periods. There was a non-significant tendency for tilapia to have greater weight and survival in the dry season. The co-culture had a significantly ( p <0.05) higher biomass production in the rainy season than in the dry season and did not present significant differences in total weight gain, SGR and FCR. The discharge rates of TN and TP were higher in the monoculture ponds in all the treatments (Table 5). In the rainy season, the TP discharge rate was higher than in the dry season and the percentage, in general, was higher for nitrogen than for phosphorus.

The variables of water quality during the two production cycles were within the appropriate intervals for the growth of tilapia and shrimp in monoculture and co-culture ( Candido et al. , 2005 ). There was a higher concentration of oxygen in the rainy season than in the dry season and there was a non-significant tendency ( p >0.05) to present higher DO concentration and lower BOD ₅ in shrimp monoculture than in co-culture. This was probably due to the combined effect of the two species (tilapia-shrimp), as reported for tilapia with other fish ( Ibrahim & Naggar, 2010 ; Shoko et al. , 2014 ).

The concentration of nutrients in the water of the ponds did not present significant differences ( p <0.05) in all the treatments. This has been found in most studies with mono and co-culture tilapia-shrimp ( Tian et al. , 2001a ; Alam et al. , 2008 ). However, there was a tendency to have lower TN and TP and their compounds had lower concentration in co-culture than in the monoculture as has been reported in other tilapia-shrimp co-culture studies ( Tian et al. , 2001b ). The average discharge rate of TN and TP was lower in the co-culture than in the monoculture and was lower for TN (21.9%) than the one recorded in co-culture in Southeast Asia (53.1%) using earthen ponds ( Briggs & Funge-Smith, 1994 ). Several processes contribute to a lower concentration of nutrients in co-culture, such as the accumulation and trapping of bound ammonia in the sediments, rapid uptake of ammonia by cyanobacteria and progressive grazing of tilapia on natural food ( Chiu, 1988 ; Tendencia et al. , 2006 ; Yuan et al. , 2010 ). Regarding the coloration of the water in the ponds, it was found that in the tilapia-shrimp co-culture, in the first weeks green color appeared and later it remained brown until the end. On the other hand, the monoculture pond had green color throughout. This is in agreement with the findings of Brito et al. (2017) who recorded that there was a low concentration of total phytoplankton (Chlorophytes, Cyanophytes, and Pyrrophyta) in monoculture and in co-culture there was greater Bacillariophyta mainly.

The density used in this work was within the optimal range of stocking reported for shrimp co-culture with tilapia ( Wang et al. , 1998 ). The proportion of 10 shrimp/m ² and 4 tilapias/m ² used has been con­sidered as a high density of tilapia in these systems ( Thien et al. , 2004 ). It has also been used in other co-culture studies of shrimp (10 shrimp/m ² ), where there was good production and survival of L. vannamei (75.2 to 78.3%) and tilapia nilótica (87.6% to 91.2%) ( Jatoba et al. , 2011 ; Simão et al. , 2013 ). Our study showed that tilapia up to 86.7 g can be co-cultured with shrimp without a decrease in growth or increase in FCR. Shrimp' FCR is lower in monoculture than in co-culture in indoors clear water systems ( Hernández-Barraza et al. , 2013 ). However, in monoculture earthen ponds, it consumes more balanced feed and has higher FCR than in co-culture ponds. Shrimp and tilapia in rainy season consume less balanced feed and have lower FCR than in dry season. However, in monoculture earthen ponds, they consume more balanced feed and have higher FCR. In general, it has been found that the overall FCR is higher in clear water systems than in earthen ponds in monoculture and co-culture ( Hernández-Barraza et al. , 2013 ).

Most studies on tilapia-shrimp interaction have been reported in indoors controlled systems in which total production of fish and shrimp was found to be higher in co-culture than in monoculture ( Hernández-Barraza et al. , 2013 ); a similar behavior was recorded in earthen ponds as in our study. The greater weight and survival, and lower consumption of feed and FCA of the shrimp in the co-culture system can be attributed to the ecological role tilapia plays ( Wang et al. , 1998 ). This is because it directly affects the following: decrease of phytoplankton predation pressure ( Vinyard et al. , 1988 ), production of organic particles of balanced food, undigested food by tilapia and its fecal matter and indirect movement of water, activation of nitrogen and phosphorus cycle and the microbial biomass developed due to the bio-manipulation of tilapia ( Gonzales- Corre, 1988 ; Ross et al. , 1988 ; Yi et al. , 2002 ; Cruz et al. , 2008 ; Apún-Molina et al. , 2015 ).

The higher survival rate found in adding tilapia to the shrimp culture has already been reported in other shrimp species ( Akiyama & Anggawati, 1998 ). This was probably due to the inhibitory effect of tilapia on certain pathogenic microorganisms ( Akiyama & Anggawati, 1998 ; Tendencia et al. , 2004 ). In the co-culture of tilapia-shrimp, survival was greater for tilapia than for shrimp in our work, a relationship that has been found in most of the works of this same culture strategy ( Simão et al. , 2013 ). There were no differences in the survival of tilapia in monoculture (87.6%) and co-culture (91.2%), which has been reported in other studies ( Bessa Junior et al. , 2012 ) with the same species used in this study (84 to 100%). Natural earthen ponds with high quality and low salinity waters allow the sustainability of greater stocking density of tilapia (4 org/m ² ) than systems or ponds lacking substrate (1.2 org/m ² ) at the bottom (glass fiber or cement tanks). Natural systems maintain high rates of recovery of nutrients in the harvest and have less negative environmental impact ( Yuan et al. , 2010 ).

Our results indicate that the addition of tilapia at high density (4 org/m ² ) into white shrimp culture (10 org/m ² ) can improve productivity, nutrient utili­zation, and environmental friendliness. The highest weight of shrimp was recorded in the tilapia-shrimp co-culture in the rainy season, due to higher water temperature and better quality of live food. There was a trend towards greater concentration of nutrients in the water of the monoculture ponds in the dry season and not in the rainy season. There was lower DO concentration in dry and higher BOD ₅ in co-culture ponds. The average individual weight of the shrimp, biomass per pond and survival in the tilapia-shrimp co-culture in the two seasons studied were greater than those of the monoculture shrimp. In conditions of rural aquaculture sustainable in small farms, shrimp-tilapia co-culture should be promoted to improve the sustainability of shrimp culture in Mexico and Latin America. Finally, our results on the water quality of the systems under study showed that management strategies are required to improve the management of water in co-culture, such as the optimization of feeding frequency, management of trays, aeration system and the implementation of sequential bottom cleaning during the culture cycle.