The present study reports the seasonal dynamics of the fish community structure in the presence of a non-native fish (Pterygoplichthys pardalis) in tropical waterbodies of southern Mexico. The studied floodplain existed in both connected pools and disconnected pools “to the Palizada river”. Local fish fauna showed consistent assemblages across seasons, but among the 17 local fish species recorded, the non-native species P. pardalis showed the highest percentage of contribution to the fish community (ca. 20%). Conversely, the frequency of occurrence and density of the non-native species showed variation in relation to the type of waterbody and seasonality; its density was approximately four times higher in disconnected pools than in connected pools in the rainy season and its predominant size class in both pool types was 21-30 cm in standard length. Interestingly, the diversity of the native species was negatively related to the density of the non-native species, indicating that the presence of P. pardalis can be a factor involved in the decline of native fish diversity in these local communities. Currently, the possible impacts of the presence of P. pardalis on local fish diversity at the regional scale are unknown, but our results suggest that hydrological dynamics regulate the establishment of the non-native species in this region. However, future studies are needed to provide insights into the actual scenario of P. pardalis distribution in Yucatan Peninsula and the possible impacts on the native fish fauna in southern Mexico.
The introduction of non-native species in aquatic environments has been recognized as a major threat to biodiversity (Elvira and Almodóvar, 2001; Kolar and Lodge, 2002). An important general gap in invasion science is information concerning how the increasing numbers of non-native species interact in human-altered ecosystems (Hobbs et al., 2006; Kuebbing et al., 2013). It has been suggested that non-native species change and deplete host ecosystems through parasite introduction (Crowl et al., 1992; Bruton, 1995), modify habitat and species-assemblage structure (Bain, 1993; Hinojosa-Garro and Zambrano, 2004; Mitchell and Knouft, 2009), change diet, survival, and foodweb interactions in the local fauna through predation (Beisner et al., 2003; Townsend, 2003) or competition pressure (Townsend, 1996; Strecker, 2006), and changes nutrient cycling and ecosystem services (Simberloff and Rejmánek, 2011). Furthermore, the success of non-native species has been related to environmental dynamics, where growth and invasion speed present an inverse response to environmental variability (Neubert et al., 2000).
Flood pulses promote physical and chemical changes in aquatic environments to which the species respond in relation to their morphological, physiological, and ethological adaptations, promoting temporal changes in fish community (Junk et al., 1989). Floodplains are important seasonal aquatic systems in relation to the temporal recruitment of individuals, species reproduction, feeding, refuge and nursery (Bayley, 1987; Freitas and Garcez, 2004). Unfortunately, anthropogenic pressure promotes changes in native species responses to disturbance, at different time and spatial scales, enhancing distinct phases in the invasion process, promoting the establishment of non-native species in these habitats (Melbourne et al., 2007; Clark et al., 2010). It has been suggested that the decline in native fish species since the early 1990s, is associated to the establishment of non-native species (Hoover et al., 2004). Examples of the establishment of non-native fish are those related to the Loricariidae family, which have been introduced to many freshwater ecosystems throughout the American continent since 1950 (Burgess, 1958). In Mexico, in the last two decades, the intentional introduction of non-native species of economic importance has been promoted as economic and food alternative for local people, without regard of the potential negative effects on local fish richness (i.e. Espinoza-Pérez and Ramírez, 2015).
In Mexico, the populations of the Amazon Sailfin Catfish (Pterygoplichthys pardalis Castelnau, 1855) have rapidly increased and successfully established in freshwater ecosystems, mainly in Río Balsas, Río Grijalva-Usumacinta, in wetlands in the state of Tabasco, and in the Laguna de Términos and Río Palizada in the state of Campeche (Amador-del Ángel and Wakida-Kusunoki, 2014; Contreras-Mac Beth et al., 2014). Some studies agree that Pterygoplichthys spp. cause damage to the local fish fauna and aquatic ecosystems in many ways (e.g. Hoover et al., 2004; Hastings et al., 2006). At high-densities (> 50 ind m−2), Pterygoplichthys spp. populations show patchy distributions across aquatic ecosystems in Chiapas, Mexico (see Capps and Flecker, 2013) modifying nutrient and re-mineralization cycling (e.g. nitrogen and phosphorous), through the enhanced abundance of feces, and the increased periphyton biomass and primary productivity (Scott et al., 2012; Rubio et al., 2016). Furthermore, Pterygoplichthys spp. increases competition for food with native fish species, especially for detritus and algae, altering the fish community structure in aquatic ecosystems (Mendoza-Carranza et al., 2010; Pound et al., 2011). By the other hand, it has been suggested that some invasive species take advantage of ecosystem changes, such those related to anthropogenic pressures, and are not directly related or act as drivers of biodiversity loss (Gurevitch and Padilla, 2004). Also, the success in the establishment of non-native species is related to the propagule pressure (number of individuals that are intentionally introduced), however, this relationship has been largely unexplored. Therefore, we used P. pardalis and the Palizada river floodplains as a study model to evaluate the effect of non-native species on the structure of local fish community in this seasonal ecosystem. Specifically, we hypothesized that the structure of local fish communities in pools with high densities of P. pardalis will differ from those pools with low densities of P. pardalis. This is the first study in southern Mexico which evaluate the impact of the introduced non-native species P. pardalis on native local fish communities. Thus, results will contribute to understand the biological implications of the presence of this non-native fish species in seasonal ecosystems, addressing the native fish community conservation.
Five pools in the Palizada River, Campeche, Mexico were chosen based on their proximity and accessibility from roads. Connected pools (CP) were those presented drain channels conducting water from the overflow of the river to the lower basin, meanwhile disconnected pools (DP) were those remained isolated from the river flow along seasons (Appendix 1: available in the online supplementary material). The aquatic vegetation is represented mainly by Typha angustifolia, Phragmites australis, Claudium jamaicense, Eichhornia crassipes; meanwhile the riparian species present are Manilkara zapota, Brosimum alicastrum, Haematoxylum campechianum, Swietenia macrophylla, Guaiacum sanctum, Cordia dodecandra and Tabebuia rosea; Pistia stratiotes and E. crassipes were dominant in the water column (ca. >50% coverage) while T. angustifolia and P. australis in the shore. CP had a mean depth of 4.17 ± 2 m while DP had ca. 3.7 ± 1 m. Sampling was carried during October 2013 and July 2014 for the rainy season, and during February 2013 and April 2014 for dry season.
Physicochemical variables in water
Water physicochemical variables including pH, temperature (°C), O2 (mg l−1), conductivity (m cm−1), total dissolved solids (g l−1), salinity and depth (m) were measured in situ recording four samples per pool in each sampling period using a multiparameter water quality sonde (YSI Professional Plus, Yellow Springs, O.H, U.S.A). Turbidity was determined using a portable turbidimeter (HACH, Model HQ40d, U.S.A.), while chlorophyll a (µg l−1) quantified using a portable fluorometer (Turner Designs, Aquafluor 8000-010, U.S.A.). In each pool, three 250-ml samples of water were taken 15 cm below the water surface and kept refrigerated for further analyses of total phosphorus (mg TP l−1), nitrite (mg NO2-N l−1), nitrate (NO3-N mg l−1), and ammonia (mg NH3-N l−1) using a spectrophotometer (DREL 2800 Complete Water Quality Lab, HACH, Loveland, C.O., Wavelength Range 340 nm - 900 nm). Sediment samples (0.5 kg) were taken in each pool using an Ekman dredge (15 x 15 cm), wrapped in aluminum foil and kept refrigerated. In laboratory, the sediment samples were dried at 80 °C for 48 h in an oven (CRAFT). Subsamples of 30 g from the first five cm in each sample were incinerated at 550 °C for 1 h using a muffle oven (Thermo Scientific, Benchtop F47910, 1200° C) to obtain the ash-free dry weight as a proxy for the organic matter content (based on Underwood and Paterson, 1993).
In each pool, fish were caught using a cast net (2.5 m diameter; 3 cm mesh size) deployed at 30 randomly points near the shore in each pool from a 3-m long boat (Bass Hunter, UBH-II). Each fish was measured for standard length (Ls) to the nearest 1 mm using a portable fish board (0-36 cm, Wildco model 118-E40). The fish were identified to the lowest taxonomic level following Schmitter-Soto (1998), Page and Robins (2006), and Miller et al. (2005). For P. pardalis, two size categories were established: ≤ 250 mm (juveniles), and > 250 (adults) according to Gibbs et al. (2008), Samat et al. (2008), and Rueda-Jasso et al. (2013). After measurements, native fish were released in situ and non-native species were euthanized and transported to Laboratory. Finally, fish individuals difficult to identify in situ were anesthetized using 2% of chloral hydrate for five minutes and placed into a cooler box for further identification following Iwama and Ackerman (1994). Laboratory analyses and fish identification were carried out at the Laboratorio en Ecología Acuática y Monitoreo Ambiental (LEAMA), CEDESU-UAC.
To determine the differences in physicochemical variables, data were ln (x + 1) transformed to meet the assumptions of normality and homoscedasticity using Shapiro-Wilks and Barlett test respectively (Shapiro and Wilk, 1965; Hartley, 1950); two-way analyses of variance (ANOVA) regarding pool-type (CP and DP) and seasons (rainy and dry) as factors. If data achieve normality and homoscedasticity, ANOVA was performed with unbalance sample sizes. The differences in the fish community composition between CP and DP were illustrated using non-metric multidimensional scaling (NMDS) based on Bray-Curtis dissimilarity distances using arcsine square root transformation. These analyses were performed in R v3.4.1 (R Core Team, 2014; www.R-project.org) and in the library vegan (Oksanen et al., 2013). In the NMDS analysis, small values have a strong influence on the simulation tests (McCune et al., 2002); therefore, fish abundances that comprised ≤ 5% of the total abundance were considered as accidental and excluded from the analyses. Analysis of similarities (ANOSIM) performed with 1000 permutations based on Bray-Curtis dissimilarity distances was used to test for significant differences between the fish communities in CP and DP between seasons by using the software PRIMER 6 (PRIMER-Ltd, Plymouth, U.K.; http://www.primer-e.com/primer.htm). To identify the percentage of the contribution of each fish species to each community, similarity percentage analysis (SIMPER) was performed with the library vegan in R environment (R Core Team, 2014; www.R-project.org). Additionally, the fish species abundance log (x + 1) transformed versus the frequency of occurrence (%), were plotted by type of pool and season using an Olmstead-Tukey test for association (Olmstead and Tukey, 1947) using R v3.4.1 (R Core Team, 2014; www.R-project.org). An average was calculated on both axes that resulted in four species categories: dominant, occasional, common, and rare.
A generalized linear model (GENMOD; SAS 2017) with binomial distribution, logistic-link function and descending option (i.e. modelling the probability of individual presence) was used to evaluate the frequency of occurrence of P. pardalis. The presence of P. pardalis was used as the dependent variable; season, pool-type and the interaction between pool-type and season were fixed as independent variables. In addition, the density of P. pardalis was evaluated with a generalized linear model (GENMOD) with Poisson distribution and log-link function (calculating a type-III and using Pearson error estimate). The model used density as dependent variable, and season, pool-type and the interaction between pool-type and season as independent variables. To determine the establishment of P. pardalis, the distribution of the occurrence and density of different class sizes of was evaluated by using the GENMOD, above mentioned. In both cases (density and frequency), class size and the interaction class-size*pool*season were fixed as independent variables. Finally, to evaluate the effect of P. pardalis density on the diversity of native fish (Simpson’s diversity index), we performed a simple linear regression model using R v3.4.1 and the library stats (R Core Team, 2014; www.R-project.org/).
Temperature and depth presented statistical normality and homoscedasticity (Shapiro test, p > 0.05; Barlett test p > 0.05) as well as statistical differences between seasons. Specifically, DP were less deep and warmer than CP. In addition, higher temperatures were obtained in DP during the rainy season, and the lowest temperatures occurred in the dry season (Factor pool-type, p = 0.56; Factor season, p = 0.007; interaction pool-type*season, p = 0.967). The depth of the CP showed seasonal fluctuations, being significantly deeper during the rainy season than in the dry season. DP showed significant differences in depth between seasons (Factor pool-type, p = 0.53; Factor season, p = 0.01; interaction pool-type*season, p = 0.37); the other variables passed normality and homoscedasticity tests, however, did not show significant differences between seasons and pool-type (Appendix 2: available in the online supplementary material).
A total of 748 individuals of eight families, 17 genera and 21 species were recorded. The species richness (S) was lower during the rainy season for both pool types (CP = 8 and DP = 9), in comparison to the dry season (CP = 17 and DP = 19). Cichlidae had the highest S, followed by Poeciliidae and Atherinopsidae. In both seasons, the species with the highest densities were P. pardalis (1.4 ind/m2; ±2.23 SD), A. aeneus (0.93 ind m−2; ±2.3 SD), G. sexradiata (0.66 ind m−2; ± 2.6 SD) and T. meeki (0.49 ind m−2; ± 1.1 SD) (Appendix 3: available in the online supplementary material). NMDS ordination procedure with two dimensions suggested different groups for the DP and the CP in both seasons (stress = 0.18), and the ANOSIM revealed significant differences between these groups (R = 0.65, P = 0.001). CP showed changes in community patterns in rainy and dry seasons, while DP showed differences in community composition in both seasons (Fig. 1). SIMPER analyses showed that the species with the highest percentage of contribution to both pool types were P. pardalis, Astyanax aeneus Günther, 1860; Gambusia sexradiata Hubbs, 1936; Thorichthys meeki Brind, 1918; Cribroheros robertsoni Regan, 1905; and P. pardalis. However, the species contributions changed between seasons in each pool type. In the rainy season, P. pardalis and G. sexradiata together accounted for 42.52% of the total fish community (Table 1).
The Olmstead and Tukey test showed that P. pardalis was the dominant species in both types of pools, with higher abundances during the raining season in DP (Fig. 2). Oreochromis niloticus was also dominant in CP and DP in the dry season. In the same way, the native species T. meeki, A. aeneus, G. sexradiata, Thorichthys passionis Rivas, 1962; Rhamdia guatemalensis Günther, 1864; Thorichthys helleri Steindachner, 1864; and Petenia splendida Günther, 1862; were dominant in the dry season mostly in CP. Meanwhile, Atherinella alvarezi Díaz-Pardo, 1972; Cyprinius carpio Linnaeus, 1758; Rocio octofasciata Regan, 1903; and Mayaheros urophthalmus Günther, 1862; were also occasional in CP. The results also showed that most of the native species (e.g. P. splendida, Dorosoma anale Meek, 1904; R. guatemalensis, T. passionis, T. helleri, Vieja melanura Günther, 1862) presented a transitional gap between seasons in each pool type, being rare in dry season but tending to end up as occasional or dominant in rainy season.
A total of 153 individuals of P. pardalis were collected. The results show that pool type (X2 = 10.21, p = 0.001) had a significant effect on the frequency of occurrence of P. pardalis, but season (X2 = 3.8, P = 0.051) and the interaction between pool and season (X2 = 3.8, p = 0.051) did not have a significant effect. The occurrence of P. pardalis was greater in DP than CP, and the density was significantly affected by pool type (p < 0.001) and by season (p = 0.004), but it was not affected by the interaction between pool and season (p = 0.022); the highest density was observed in the rainy season in DP. Class size (X2 = 48.85, p < 0.001) and the interaction class size* pool*season (X2 = 72.82, p < 0.001) significantly affected the frequency of occurrence of P. pardalis. Specifically, smaller individuals occur more frequently than larger in both type of pools and seasons, although the highest occurrence for the two class sizes was in the rainy season (Fig. 3a). The same pattern was observed for the density: class size and the interaction among class size*pool*season significantly affected (p = 0.027 and p < 0.001; respectively) the density of this no-native fish; fry showed higher densities compared with juveniles in both pools (CP and DP), but they were particularly abundant in DP during the rainy season (Fig. 3b). Interestingly, the linear relationship results indicated that native fish diversity (Simpson’s index D) was negatively affected (R2= 0.68, P < 0.001) by the density of P. pardalis (Fig. 4).
The introduction of non-native fish species is a recurrent reported cause of changes in community structure and loss of diversity of native local communities (Propst et al., 2008; Mitchell and Knouft, 2009). Overall, the results of this study indicate that the abundance of the non-native species P. pardalis has an important contribution to the local fish community in the Palizada river floodplain. Specifically, we found that DP showed less seasonal variation in water depth and temperature, and a more stable assemblage of local fish species. In these pools, the composition of native species did not change significantly among seasons compared to CP. In DP, most of the species found were cichlids and R. guatemalensis; whereas A. aenus and G. sexradiata showed the highest contribution percentage to the fish community in CP. In this regard, it has been reported that temporal variation in abiotic attributes, such as physicochemical water variables related with hydroperiod, can modify fish species presence and abundance according to life histories strategies of the species (Winemiller, 1995; Baber et al., 2002; Pazin et al., 2006; Escalera-Vázquez and Zambrano, 2010). This can explain the differences in the community structure we observed between DP and CP, since cichlids and R. guatemalensis that preferentially inhabit DPs have been related to stable aquatic ecosystems as a result of their seasonal-synchronized reproduction (Winemiller, 1995; Lourenço et al., 2012). On the other hand, A. aenus and G. sexradiata, the most abundant species in CPs, have the ability to seasonally inhabit new habitats, since they are effective colonizers with long spawning periods (January to July), early maturity, and the ability to cope with seasonal water fluctuations (Miller et al., 2005; Escalera-Vázquez et al., 2017). Indeed, A. aenus and G. sexradiata are the most abundant species in the Yucatan Peninsula (Schmitter-Soto, 1998). These results suggest that the connectivity of the Palizada river to adjacent pools is a key factor that regulates the presence and abundance of native species across the studied area (see Öhman et al., 2006). Our results also suggest that connectivity between the river and the adjacent pools may enhance the dispersion of the non-native fish species. According to the above mentioned and the richness index used in this study, we cannot conclude that the abundance of the non-native species is a direct cause of the loss on native fish diversity. By the other hand, native species diversity and temporal environmental changes in this aquatic ecosystem can act as buffer to prevent invasions. An alternative hypothesis regarding our results related to the aquatic systems studied, is that the dominance (in abundance) of the non-native fish species could be an indirect consequence of changes in habitat by anthropogenic activities, driving potential native species loss and facilitating this species establishment and invasion (Rejmánek, 1996; Didham et al., 2005). Also, according to MacDougall and Turkington (2005), the fish community studied is high related to seasonal changes in water conditions (hydroperiod), which may be primarily structured by noninteractive factors, and non-native species can dominate the invaded community acting as ‘‘passengers’’ with less constraining factors.
In particular, the frequency of occurrence and the density of P. pardalis was approximately twice and four times higher in DP than in CP respectively in the rainy season. Furthermore, the density of juveniles in DP in the rainy season was significantly higher than the density in CP, suggesting that P. pardalis may use DP for nursery in the rainy season. However, the frequency of occurrence (except for DP in the dry season) and the density of individuals <250 were similar between pools and seasons, suggesting that the juvenile and fry of P. pardalis can tolerate seasonal fluctuations, resulting in a high probability of inhabiting all pools in the Palizada river. Wakida-Kusunoki and Amador-del Ángel (2011) reported a density of 9.8 ind m−2 in the Palizada River, our results show a lower density (ca. 2.0 ind m−2). However, the introduction pressure and possible variation in rates must be considered, regarding that the success on establishment depends on propagule in the range of 10 ± 100 individuals (Cassey et al., 2018). Therefore, data related to intentional introductions programs should be confirmed, highlighting that this information in generally unknown (Veltman et al., 1996). On the other hand, personal observations in the field suggest that adult individuals drift away with waters discharged, explaining their absence in adjacent pools, although the presence of juveniles in the adjacent pools suggests that adults use these pools for reproduction. Given that DP were warm in the rainy season when temperatures are relatively cooler and homogenous in the floodplain, it is possible that water temperature determines the establishment success of the younger individuals of P. pardalis in the adjacent pools, suggesting that P. pardalis populations could be regulated by hydroperiod regimes. Previous studies have reported that the reproduction and onset of puberty of the closely related species (e.g. Pterygoplichthys disjuctivus Weber, 1991) is delayed by low temperatures (Rueda-Jasso et al., 2013).
Our results also indicate that P. pardalis accounted for the highest contribution to the fish community in Palizada river floodplain. Pterygoplichthys pardalis can trigger eutrophication by the increase in nutrients in the water column. Previously, Capps and Flecker (2013) observed an increase in ammonia (NH4) in systems with > 50 ind m−2 of Pterygoplichthys spp., increasing two-fold during species peak activity period at night (12-2 am). However, our results did not indicate a significant variation in ammonia, productivity (chlorophyll-a), and nutrient concentrations (TP). These results might be related to P. pardalis low abundance and the presence of small individuals which seem not to enhance nutrient enrichment in these waterbodies. However, we found that the density of P. pardalis had a significant negative effect on the native species diversity. Lourenço et al. (2011) observed that periods with low water levels increase fish species competition, reducing species richness and abundance in floodplain lagoons. Similarly, previous studies have also reported that the introduced non-native species enhance competition for space and food with native species in tropical waterbodies, resulting in diversity decline (Flecker, 1996; Nico and Martin, 2001). In the case of P. pardalis, competition for resources with native species has been suggested in wetlands in southwest Mexico (Mendoza-Carranza et al., 2010). Then, it is possible that the continuing increase of P. pardalis populations in Palizada river strengthen biotic interactions with native fish fauna. Also, species that readily exploits new food sources may be more preadapted to live in a novel environment that a more specialized one, representing a threat not only to species which share similar food sources, but also to the integrity of ecosystem services and maintenance (Yossa and Araujo-Lima, 1998; Sol and Lefebvre, 2000; Cohen, 2008).
In summary, our results provide insights into the current circumstances and possible future scenarios of P. pardalis presence in seasonal waterbodies of southern Mexico, since they reveal that P. pardalis is spreading from the Palizada river basin to secondary small waterbodies that support important local fish communities. At this moment, its specific interactions with native fish are not unknown, but it appears that rainfall-driven hydrology is a key factor in these dynamic seasonal tropical waterbodies which modify P. pardalis density, which in turn impacts the local diversity. However, further research is necessary to determine whether P. pardalis is able to colonize new habitats in the Yucatan Peninsula and its possible threats to the local fish fauna in southern Mexico.
Implications for conservation
In the Yucatán Peninsula (where the state of Campeche represents ca. 45% of the total area), the conservation of fish communities is necessary in order to maintain the integrity of ecosystem services at regional scales, where hydrology connects seasonally land and water. This includes the knowledge of the potential impacts of non-native species in aquatic ecosystems that depends on temporal connections to disperse and reproduce. On this matter, if non-native fish species are a real income and food resource for local people and fishermen, then, not only the implementation of real public policies to include more conservation areas that harbors many endemic species is needed, but also, include biological corridors that prevent invasion from non-native species in order to preserve one of the most valuable ecosystems regarding water problems projections worldwide for the next decade (i.e. Costanza et al., 1997; Vörösmarty et al., 2000), mainly, in an area where the human population has five-folded in the last three decades (INEGI, 2010).
The current study reports the presence of P. pardalis in seasonal pools of southern Mexico. Its abundance does not modify abiotic conditions in the studied pools. However, results suggest that the density of P. pardalis had a significant negative effect on the native species diversity. Further research is necessary regarding regional and global scales in order to gather more information about the mechanisms in which non-native species are capable of becoming invasive, passenger or local-adapted. This information is necessary to implement management strategies for local-native fish community conservation and ecosystem services that freshwater ecosystem provide.
This paper constitutes a partial fulfillment of the Graduate Program (MS in Science, Environmental Biology Program, Facultad de Ciencias Químico Biológicas) of the Universidad Autónoma de Campeche, for J. E. G. L. We are thankful to the farmers of Palizada region who allowed us to work in their fields and pools. We thank J. Tucuch-May, W. Cu-Peralta and J. Pali for their valuable assistance on fieldwork.
Supplementary material for this manuscript is available on-line at the publisher’s website.
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: J. E. G. L. was supported by a scholarship from CONACyT fellowship no. 353009 register 28677.