Interactive Effects of Abiotic and Biotic Factors Drive Aquatic Plant Colonization in Subtropical Mangroves

The colonization and spread of non-native species are recognized as a critical driver of environmental change in mangrove ecosystems. However, environmental factors that favor non-native plant colonization in mangroves are still poorly understood. To contribute to lling this gap, we investigated the effect of selected abiotic factors associated controlling aquatic macrophytes colonization in mangroves in Southeastern Brazil. Furthermore, we also assessed foliar attributes of native and non-native species to evaluate whether biotic attributes favored the colonization. We selected 18 plots in mangrove forests under different levels of anthropogenic N inputs, both colonized and non-colonized by aquatic macrophytes in the Estuarine-Lagoon Complex of Cananeia-Iguape, southeastern Brazil. We measured interstitial salinity, sediment nitrate and ammonium concentrations, and sediment physicochemical properties. We also measured foliar nitrogen (N) concentrations, foliar C:N, and foliar δ 13 C of both native and non-native species. We found that interstitial salinity at 10 cm depth followed by nitrate concentrations in sediment were the main factors associated with the occurrence of aquatic macrophytes in the studied mangrove areas. Moreover, non-native species had about 2-fold higher foliar N concentrations as well as about a 1.5-fold lower C:N ratios compared to native species. Low salinity and increased N availability in sediment allowed for the success of aquatic macrophytes into mangrove forests, also resulting in high amount of dead mangrove trunks (up 53% of basal area).


Introduction
Mangroves are the only forest formations that occur at the interface of terrestrial, freshwater, and marine ecosystems (Alongi 2002  Although there is wide recognition of the social, economic, and ecological relevance of mangroves, these ecosystems are still under strong human pressure (Valiela et al. 2001). One of the main drivers of ecosystem degradation is biological invasion. Despite the fact biological invasion is worldwide well recognized as one of the main vectors of global change promoting modi cation in the structure and functioning of ecosystems (Vitousek et al. 1996; Dukes and Mooney 1999; Mack et al. 2000; Vila et al. 2011), such driver has not received enough attention in mangrove ecosystems (Biswas et al. 2018). This is especially true considering the lack of investigations of environmental factors that favor the colonization of non-native species (Ren et al. 2014) and the functional attributes of non-native species that thrive in mangroves (Biswas et al. 2018).
Environmental factors that make an ecological system more susceptible to the colonization of non-native species are widely discussed, though there is a consensus that disturbance plays a key role in this process ( . For example, measures to recover stream ow to mangroves have been taken in Colombia. Rivers were dredged to restore the river ow that had been altered for the construction of highways. Nonetheless, the combination of large rain events at the same time of these interventions caused a salinity reduction which, in turn, led to the colonization of aquatic macrophytes (Typha domingensis Pers) in these mangroves (Röderstein et al. 2013). The occurrence of non-native species can have signi cant effects at the species, community, and ecosystem levels (Vila et al. 2011). This might occur especially if native species differ from non-native species in terms of functional attributes (Lee et al. 2017). Such differences may provide competitive advantages for non-native species (e.g., increased resource use e ciency, fast growth rate, and higher tness) since they generally have a greater performance compared to native species (Van Kleunen et al. 2010; González-Muñoz et al. 2013). Photosynthetic metabolism (e.g., C 3 x C 4 ) is a functional attribute that may differ between native and non-native species. The stable carbon isotopic signature (δ 13 C) indicates the type of photosynthetic pathway used by plants (Ehleringer et al. 2000). Moreover, the δ 13 C also has the potential to provide insight into plant water use e ciency (i.e., assimilation of C per unit of water lost in transpiration) and consequently on photosynthetic capacity, thus highlighting differing competitive abilities among C 3 plants. Foliar N concentration and carbon: nitrogen ratio (C:N) are commonly used to evaluated differences between native and non-native species, which may be associated with N use and in uences ecosystem process (Liao et al. 2008;Lee et al. 2017).
The Estuarine-Lagoon Complex (ELC) of Cananeia-Iguape, São Paulo State, Brazil is a Ramsar site (Ramsar 2017). Mangroves at the southern region of the complex are the widest and best conserved in São Paulo State (Cunha-Lignon et al. 2011). On the other hand, in the northern part of the complex, there was a signi cant increase in freshwater input due to the building of a canal in 1852 (locally known as "Valo Grande") to deviate ow from Ribeira de Iguape river (Mahiques et al. 2009(Mahiques et al. , 2013. The Ribeira de Iguape river is one of the largest rivers in the southeastern region of Brazil, draining more than 23,350 km 2 . This deviation, as well as the input of sewage and excess nutrients from agricultural activities in the Ribera de Iguape river basin, resulted in important changes in the estuary (locally known as "Mar Pequeno") and associated mangroves. This possibly favored the colonization by aquatic macrophytes in To shed more light on the controls of the colonization of mangrove ecosystems by non-native species, we investigated the effect of selected abiotic factors affecting the colonization of aquatic species in the ELC of Cananeia-Iguape. We also assessed functional foliar attributes of non-native and native species to clarify the role of biotic attributes to the success of non-native species in these systems.

Study area
The study was conducted in 18 plots inserted in 8 fringe mangrove sites in the ELC of Cananeia-Iguape, southeastern Brazil in 2015 (Fig. 1). This study was carried out by the Integrated Monitoring of Mangroves Research Group recognized by the National Council for Scienti c and Technological Development (CNPq, Brazil). The study plots integrate a permanent plot network where mangrove vegetation structure has been monitored since 2001.
The study sites differ mainly in two aspects: (i) levels of anthropogenic N inputs (Reis et al. 2019) and (ii) presence of aquatic macrophytes. We selected two conserved mangrove areas under high marine in uence, free of aquatic macrophytes and with no in uence of anthropogenic N inputs. These areas were our control sites. We also selected two mangrove sites free of aquatic macrophytes but subject to medium N inputs from sewage discharge of a small urban area in the Cananeia Island (+ N). Lastly, we selected four mangrove sites close to the Valo Grande canal that were strongly affected by excess N inputs from the Ribeira de Iguape river basin. Two of these sites comprised mangroves areas without the presence of aquatic macrophytes and submitted to excess N inputs ranging from medium to high ( + + N).
The other two sites included mangrove areas colonized by aquatic macrophytes and subjected to high N inputs (+++NM) (

Sampling design
In July 2015, vegetation structure was measured and sediment and foliar samples were obtained in each plot. To avoid trampling and possible interferences in the vegetation structure inside the plots, all foliar and sediment samples were obtained immediately outside each plot. The interstitial salinity was also obtained in situ next to each plot.
The plots have varying sizes, according to the stem density (Schaeffer-Novelli and Cintrón 1986; Schaeffer-Novelli et al. 2015). In each plot, all trees taller than 1 m were identi ed and had their diameter at breast height (DBH) registered with a metric diameter tape, and their height measured using a telemeter or a telescopic pole. The condition of the trunks (alive or dead) was also registered (Schaeffer-Novelli and Cintrón 1986; Schaeffer-Novelli et al. 2015).
A sediment sample (0-10 cm depth) was collected per plot to quantify the concentrations of NH 4 + -N and NO 3 − -N. Extra sediment samples at 0-10 and 10-20 cm depths were also collected next to each plot to characterize sediment physicochemical properties. The sediment samples for physicochemical analyses from plots in the same site were pooled together, totaling two composite samples by treatment (i.e., Control, + N, ++ N, and +++NM). All sediment samples were kept refrigerated until analysis.
Next to each plot, mature foliar samples (i.e., green and fully expanded) from three mangrove trees with DBH ≥ 4 cm were obtained, as well as foliar samples of the aquatic macrophytes, one sample per species. Foliar samples were analyzed for the δ 13 C, and concentrations of C and N. The foliar samples were washed with tap water and then oven dried at 40°C for 48 hours immediately after eldwork.

Abiotic factors 2.3.1 Physical and chemical properties of sediment
The sediment samples at 0-20 cm depth were analyzed for concentrations of organic matter, Na + , K + , Ca 2+ , Mg 2+ , and available P, sum of exchangeable bases, and cation exchange capacity according to Embrapa (1997Embrapa ( , 2009). Sediment texture was analyzed using the hydrometer method (Bouyoucos 1932) and classi ed using the U.S. Department of Agriculture textural triangle. Organic matter (OM) was extracted with potassium dichromate in sulfuric medium and quanti ed by titration using ammonium ferrous sulfate. Available exchangeable bases were extracted with ammonium acetate 1M.
Concentrations of Na + were analyzed by ame photometry, K + by atomic emission, and Ca 2+ and Mg 2+ by atomic absorption spectrophotometry. Available P was extracted with Mehlich 1 solution and quanti ed with ammonium molybdate using a spectrophotometer. These analyses were carried out in the Department of Soil Science of the Luiz de Queiroz School of Agriculture, University of São Paulo.
Also, in each plot, we used a Napoleon-type auger and an optical refractometer to measure in situ the interstitial sediment salinity at 10 cm (Sal10) and at 50 cm (Sal50) depth.

Statistical analyses
The normality of raw data and residues was tested using the Shapiro-Wilk W-test. The homoscedasticity of the variances was veri ed by the Barlett test.
Multiple comparisons between the treatments (i.e., Control, +N, ++N, and +++NM), regarding the sediment physicochemical properties and biotic and abiotic data, were evaluated using the Permutational Multivariate Analysis of Variance (PERMANOVA), using Euclidean distance matrix with 999 permutations after standardization of the data with the function "scale" (Anderson 2001). A Pearson correlation matrix was used to evaluate the correlation between abiotic and biotic variables in order to eliminate highly correlated variables from the analyses (r > 0.9) (Supplementary Fig. 1). Given that BADT and BALT were highly correlated, only BADT was used for PCA and further analyses ( Supplementary Fig. 1).
The study plots and the abiotic and biotic variables were ordered through Principal Component Analysis (PCA) to verify the environmental distance between treatments (Legendre and Legendre 2012).
Univariate comparisons of abiotic and biotic variables between treatments were performed using the oneway ANOVA F-test followed by pos-hoc Tukey HSD test. Comparisons of foliar attributes between mangrove and aquatic macrophytes at the +++NM plots were tested using Student's t-test or the nonparametric Mann-Whitney U-test.
The variance in ation factor (VIF) values of the predictor variables were evaluated. Variables with VIF > 20 were excluded because they presented high collinearity. We conducted a redundancy analysis (RDA) to evaluate the effect of the selected abiotic factors (Sal10 + NH 4 + -N + NO 3 − -N) on the biotic variables that presented signi cant differences between treatments on univariate comparisons (BADT, foliar C:N and foliar δ 13 C) and also foliar N, which is an important foliar attribute. The statistical signi cance of the RDA axes and groupings were tested by the analysis of similarities (ANOSIM). A partial RDA (pRDA) was performed to evaluate the individual effect of each predictive abiotic variable on the set of response biotic variables (Legendre and Legendre 2012). The signi cance of the effect of each predictor variable on the response variables was tested by an analysis of variance (ANOVA).
All analysis was performed using the software R (R Core Team 2017) at p < 0.05.

Physical and chemical properties of sediment
Sediment texture and the active, potential, and exchangeable pH values were similar among study plots.
The PCA axes of abiotic variables explained 89.9% of the total variance in the data set ( Fig. 2A).
Ordination axis 1 was negatively correlated to NO 3 − -N concentrations and interstitial Sal10 and Sal50, which grouped Control and + N plots. The ordinations of +++NM plots was negatively correlated with the variables Sal10, Sal50, and NO 3 − -N. Ordination axis 2 was negatively correlated with the Nmin and NH 4 + -N that grouped the + + N 2-4 plots, which presented the highest concentrations of these variables ( Fig. 2A). The PCA axes of biotic variables accounted for 69.1% of the variance (Fig. 2B). Ordination axis 1 indicated that the invaded areas subjected to high N inputs (+++NM plots) were positively correlated with the foliar δ 13 C and N concentrations and with BADT, while were negatively correlated with foliar C:N ratios. The ordination of plots of Control, +N, and + + N was positively correlated with C:N and by the variables DAP and H (Fig. 2B).  (Fig. 4A). Foliar N concentrations of the aquatic macrophytes species were higher (26.0 ± 5.3 g/kg) relative to mangrove species (15.0 ± 3.9 g/kg) (t (19) =-5.4938, p = < 0.001) (Fig. 4B).
The foliar δ¹³C of the dominant aquatic macrophytes (C 3 and C 4 plants pooled together, n = 6) (median, and rst and third quartiles) (-28.5, -29.9 and − 16.9‰) was similar to mangrove species (29.3, -30.1 and 28.6‰) (U = 48, p = 0.290) (Fig. 4D). The PCA analysis indicated the association between the mangrove areas subjected to high N inputs and colonized by aquatic macrophytes (+++NM plots) with the concentrations of NO 3 − -N in the sediment, interstitial Sal10 and Sal50 and by the biotic variables BADT, foliar δ 13 C, N foliar and foliar C:N. The RDA and pRDA analysis indicated that the biotic variables BADT, foliar N concentrations, C:N ratios and foliar δ¹³C were signi cantly in uenced by interstitial salinity at 10 cm depth followed by NO 3 − -N concentrations in mangrove sediment. This suggests that colonization by aquatic plants is mainly controlled by these variables at our study sites. The Valo Grande canal opening has facilitated the entrance of propagules and banks of aquatic macrophytes into the estuary and mangrove areas in the Iguape region. Our results suggest that a salinity reduction to 1.7 PSU following excess freshwater inputs was the main factor contributing to the colonization and subsequent establishment of aquatic macrophytes in these mangrove forests. These results add to previous studies that indicated high salinity as the main environmental lter preventing non-native plant colonization into mangroves (Lugo 1998). The plants that invade mangrove forests are usually able to tolerate high salinity to some extent (Biswas et al. 2018). For this reason, species like aquatic macrophytes would not be able to colonize mangroves unless salinity was signi cantly reduced in these systems (Lugo 1998 Concentrations of NO 3 − -N in sediment were the second factor that mostly affected the colonization by aquatic macrophytes at our study sites. It can be explained by the fact that nutrient concentrations in sediment affect the growth of non-native species (Ren et al. 2014). For instance, while the occurrence of non-native species in mangroves and other coastal areas in south China was negatively correlated with sediment salinity, their biomass was positively correlated with total N content in sediments (Ren et al.

2014)
, underscoring the importance of N availability for the growth of non-native species. The Valo Grande canal also carries excess N inputs from human activities resulting in high N availability in mangrove sediment at the +++NM and + + N plots. Despite the + + N plots exhibited higher NO 3 − -N and mineral N concentrations in sediment relative to the +++NM plots, aquatic macrophytes were not recorded inside the mangrove stands where the + + N plots were located. Because these stands were located closer to an open ocean inlet, salinity ranged from 13.0 to 20.0 PSU at the + + N plots. Thus, the high salinity likely prevented the establishment of aquatic macrophytes species at these stands, despite propagules and oating banks of aquatic macrophytes being found in the surrounding estuarine waters. This highlights that salinity is the main factor controlling the establishment of non-native species at our study sites.
The co-occurrence of aquatic plants and mangroves has been found in ecotonal zones and bordering mangrove forests (Tomlinson 1986;Lugo 1998  The higher foliar N concentrations and lower C:N ratios of aquatic macrophytes relative to mangroves suggested a higher N demand by the non-native species. Accordingly, we have also found that aquatic macrophytes exhibit higher foliar δ 15 N than mangroves at the +++NM plots (Reis et al., unpublished data). Moreover, the lower NO 3 − -N concentrations in mangrove sediment at the +++NM plots compared to the others likely re ected a higher absorption of N by the species of aquatic macrophytes, along with higher N losses to the atmosphere via denitri cation as a consequence of excess inorganic N inputs and intensi ed N cycling (Reis et al. 2017b(Reis et al. , 2019. The differences in foliar N and C:N ratios between native and non-native species can also modify the quality of the organic matter in the sediment, which could also potentially affect N and C pools and cycling in the system (Liao et al. 2008;Lu et al. 2014;Lee et al. 2017). Further studies are needed to evaluate the consequences of the establishment of non-native species to the functioning of these mangrove forests.
While mangrove trees have C 3 photosynthesis pathways, the species of aquatic macrophytes studied here exhibited predominantly C 3 but also a few C 4 photosynthesis pathways, thus re ecting a change in the acquisition of C. For this reason, we observed a greater variability in foliar δ¹³C of non-native than native species at the +++NM plots. Due to the occurrence of C 3 and C 4 non-native photosynthetic plants and the similarity between the foliar δ¹³C values of the C 3 non-native and native species, this attribute was not a good indicator for differentiating mangroves and aquatic macrophytes in the case of this study.
In summary, the abiotic factors interstitial Sal10 and NO 3 − -N concentrations in sediment facilitated the colonization by aquatic macrophytes in the studied mangrove areas. We also veri ed the higher foliar N concentrations and lower foliar C:N ratios of non-native relative to native plants, possibly re ecting a greater N demand and assimilation of N by non-native plants. Considering that most non-native plant colonization and establishment in mangroves result from changes in abiotic conditions as a result of disturbances (Biswas et al. 2018) and that these changes may be di cult to reverse (Bradley et al. 2010), management efforts should be focused on the major disturbance factors that control the colonization of non-native plants by promoting restoration of abiotic conditions and recolonization of native species (Röderstein et al. 2013). In the case of this study, the rehabilitation of regional hydrological conditions coupled with the reduction of excess nutrient inputs, especially of N, should be the rst step in the management of aquatic macrophytes in mangroves of the ELC of Cananeia-Iguape. Therefore, the rehabilitation of mangrove forests near the mouth of the Valo Grande is a relevant topic due to the alarming amount of dead trunks observed in these areas and the many important ecosystem services that mangroves provide.

Conclusions
The increased freshwater ow and associated abiotic changes veri ed in the north sector of the ELC of Cananeia-Iguape were determinant for the species of aquatic macrophytes overcoming the geographic and environmental barriers, allowing their expansion and dominance in mangrove ecosystems. The reduction of salinity and increased N availability in the sediment allied to the functional attributes of the non-native plants related to a high N demand allowed the occurrence and success of the colonization aquatic macrophytes in the studied mangroves.
Declarations Figure 1 Location of mangrove study sites in the Estuarine Lagoon-Complex of Cananeia-Iguape, southeastern Brazil. Study sites included two non-N-enriched and free of aquatic macrophytes reference sites (Control); two sites free of aquatic macrophytes and subjected to medium N inputs (+N); two sites without the presence of aquatic macrophytes and subjected to medium to high N inputs (++N), and two sites colonized by aquatic macrophytes and subjected to high N inputs (+++NM). Because of the map scale, the two study sites at +++NM and at ++N were overlaid. Source of the shape les of mangrove forests and aquatic macrophytes areas: Cunha-Lignon et al. (2011) Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.