Palatability of mangrove leaves to invasive apple snails: the relation between feeding electivity and multiple plant characteristics

Mangrove forests worldwide have been subjected to biological invasion. Invasive apple snails ( Pomacea canaliculata ) have established populations in some mangrove forests. The feeding behavior of P. canaliculata in mangroves has been unclear until now. The feeding electivity of P. canaliculata to mangrove leaves, including leaves from Acanthus ilicifolius, Acrostichum aureum, Kandelia candel, Aegiceras corniculatum, and Sonneratia apetala , was studied through a selective test, a non-selective test, and a T-tube test. The growth indicators, the feeding ratio, the feeding rate, the feeding amount, the electivity indicator, and the electivity frequency were determined. The weight increase ratios of P. canaliculata that consumed leaves of A. ilicifolius and A. aureum were higher than those that consumed leaves of K. candel, S. apetala , and A. corniculatum . The electivity indicator showed that P. canaliculata preferred A. ilicifolius leaves and fed little on A. corniculatum leaves. Acrostichum aureum leaves were the second most preferred food for the apple snails. The feeding electivity of P. canaliculata to leaves from five species of mangrove trees was significantly differentiated by cluster analysis, redundancy analysis, and principal component analysis. Eight feeding indicators of the apple snails were positively correlated with the leaf characteristics of nitrogen content, protein content, leaf area, moisture content, and aspect ratio and negatively correlated with the lignin, phenolic, and tannin contents. Pomacea canaliculata could discriminate mangrove leaves through physical and chemical characteristics and shift feeding electivity among mangrove leaves under intraspecific competition. Mangrove forests composed mainly of A. ilicifolius and A. aureum might be at risk of invasion by P. canaliculata . A detailed survey on benthic animals is necessary to monitor and prevent P. canaliculata invasion in mangroves.


Introduction
Mangrove wetlands, which are among the most productive ecosystems in the world, provide diverse habitats for terrestrial and aquatic invertebrates, fish, and other wildlife (Alongi 2018;Celis-Hernandez et al. 2020). Mangrove forests play a key role in sustaining ecosystem services. Typically, functions of mangrove forests include the nutrient cycling (Taillardat et al. 2019), environmental protection (Torres et al. 2019), carbon sequestration (Ouyang et al. 2017), flooding resistance, and erosion prevention (Deb and Ferreira 2017).
The apple snail (Pomacea canaliculata Lamarck, 1822), which is native to South America, is a harmful gastropod that is listed as one of the 100 worst invasive alien species (Lowe et al. 2000). Pomacea canaliculata was introduced into Asian countries in the 1980s for human consumption (Joshi and Sebastian 2006;Lv et al. 2011). However, their economic values plummeted unexpectedly, leading to a failure in the local market. As a result, P. canaliculata escaped in various aquatic habitats, including paddy fields, lakes, rivers, canals, and ponds (Kwong et al. 2010;Yang et al. 2018). Pomacea canaliculata feeds on rice seedlings and has observably threatened rice productivity (Martín et al. 2008). Furthermore, consuming P. canaliculata may transmit rat lungworm disease to humans, as they are a crucial vector of the parasitic nematode (Angiostrongylus cantonensis) (Lv et al. 2011).
In addition to their high fecundity and strong resistance, P. canaliculata exhibit a voracious appetite for macrophytes in invaded wetlands, which constitute a major part of the daily diet of apple snails, irrespective of seasonal changes (Kwong et al. 2010;Yang et al. 2018). Over fifty plant species served as food sources for P. canaliculata, and the electivity of snails varied greatly (Horgan et al. 2014). The amount of elephant-head amaranth (Amaranthus gangeticus) consumed daily by P. canaliculata contributed 22% to the weight of individuals ). The grazing rates of P. canaliculata on diverse macrophytes were closely related to nutrient traits, including nitrogen, phenolic contents (Qiu and Kwong 2009), and dry matter .
The distribution range of P. canaliculata was estimated to be between 40°N-40°S globally, and the species is known to have devastating effects in Asia, North America, and Oceania (Hayes et al. 2008). To clarify the invasion strategy leading to population establishment, many researchers have assessed the resistance of P. canaliculata to abiotic stressors, such as low temperature, pH, dissolved oxygen level, desiccation, and salinity (Seuffert and Martin 2009;Bernatis et al. 2016;Yang et al. 2017). As a freshwater gastropod, apple snails can grow in brackish water (5‰) (Yang et al. 2017). A P. canaliculata population was reported in brackish water (1-2.1‰) in a Chilean lagoon (Letelier et al. 2016). This finding demonstrated that P. canaliculata could survive brackish waters and its potential to invade estuarine habitats.
These fears recently became a reality following the discovery of P. canaliculata in mangrove forests. A routine benthic investigation reported that P. canaliculata had established populations in six mangrove wetlands in a coastal area (Ma et al. 2018). Here we report for the first time the presence of P. canaliculata in a mangrove wetland in Guangzhou, China. Given the voracious appetites of P. canaliculata for macrophytes in freshwater wetlands, we hypothesized that the mangrove leaves might provide a consistent food source for P. canaliculata in mangrove wetlands. Mangrove leaves contain complex secondary metabolites, including tannins, alkaloids, and diverse bioactive compounds (Behbahani et al. 2018). However, little is known regarding the feeding behavior of P. canaliculata on mangrove leaves. Therefore, to identify the mechanisms involved in food consumption by P. canaliculata in recently invaded mangrove forests, we asked two questions: 1) What was the electivity of P. canaliculata to mangrove leaves? 2) Was the feeding electivity of P. canaliculata influenced by the physical and chemical characteristics of the mangrove leaves? The study puts forward a new perspective to understand the influence of a recently invaded snail in mangrove wetlands. Answering these questions is beneficial for predicting the further spread of P. canaliculata in mangrove wetlands and assess the risk of P. canaliculata in diverse mangrove forests.
We used DNA Genomic Extraction Kits (AxyPrep) to extract total genomic DNA from the foot tissue of collected snails. The cytochrome oxidase subunit I gene was amplified by polymerase chain reaction (PCR) using the primers (LCO1490 and HCO2198) (Folmer et al. 1994). The reaction system (30 μL) composed of 2×PowerTaqPCRMasterMix 15 μL, 1 μL each of forward and reverse primers, ddH2O 12 μL, and 1 μL template DNA. Purified PCR product was sequenced using an ABI-3730xl (USA). Nucleotide sequences obtained were assembled and edited using BioEdit [7.2.6.1] (Hall 1999). The collected snails were identified as Pomacea canaliculata by sequencing of mitochondrial cytochrome oxidase I (GenBank Accession number: MW799954; MW799955). The apple snails were reared in a laboratory to obtain the eggs. Hatched snails were reared in artificial seawater (2.0‰) for two weeks before they were transferred in four sets of aquaria (80 cm × 60 cm × 60 cm) filled with artificial seawater (4.0‰). The salinity was monitored using a salinity meter (SX5051, Shanghai) and manipulated by adding artificial sea salt (Yanzhibao, Guangdong) to aerated tap water. The salinity (4.0‰) and pH (7.3-7.5) of the seawater were adjusted according to the water quality determined in the sampling locations and described in previous studies (Yang et al. 2017). Pomacea canaliculata were fed with lettuce (Lactuca sativa L.) daily at room temperatures (25 ± 2 °C). The water in the aquaria was changed with aerated seawater every three days. Healthy snails, characterized by smooth and intact shells and active movement in the aquaria, were used in further experiments. Before use in an experiment, the snails were starved for 24 h. All experiments were conducted at room temperature using artificial seawater at 4‰ with a 1:1 ratio of male and female snails.
Mangrove leaves, including those from A. ilicifolius, A. aureum, K. candel, A. corniculatum, and S. apetala, were tested to determine the feeding indicators of the apple snails. Fresh mangrove leaves were collected directly from the sampling sites, and their physical and chemical characteristics were determined. At the end of the non-selective/selective feeding and electivity experiments, the remaining mangrove leaves were cleaned using purified water to remove adhered salt. The tests were both performed in 6 replicates. Additionally, leaves used in the non-selective/selective feeding tests were oven-dried at 105 °C (4 h) and 80 °C (48 h) to obtain a constant weight.

Non-selective feeding test of P. canaliculata on mangrove leaves
Cleaned fresh leaves were used in the non-selective feeding test. Fresh leaves from each species (49.4-50.0 g) were placed separately in a chamber (40 cm × 30 cm × 23 cm). Ten healthy snails (height: 3.2 ± 0.2 cm) were randomly selected for each chamber, and their weights were recorded for further calculation. The snails were placed together with fresh leaves in a plastic chamber filled with artificial seawater (6 L). All the snails were alive at the end of the 120 h test.
The artificial seawater in the chambers was changed daily to maintain water quality. The feeding amount per snail weight (Fa), absolute feeding ratio (Ra), feeding rate per day (Rt), and Ivlev's electivity indicator (E), hereinafter referred to as the electivity indicator, were calculated to assess the feeding behavior of the snail (Ivlev et al. 1961, equations 1 to 6). i=1-5, representing five mangrove leaves; m i was the moisture of a type of leaf; FW i was the initial fresh weight of a type of leaf (g); DW i was the final dry weight of a type of leaf (g); Fa i was the feeding amount (dry weight) per individual weight (g g -1 ); W i was the total weight of ten snails (g); Ra i was the absolute feeding ratio; Rt i was the feeding rate (dry weight) per day per individual weight (g d -1 g -1 ); 10 was the number of individuals; T was the experimental duration (d); E i was the Ivlev's Electivity indicator; E i > 0 means the plant was selected, E i = 0 means random selection, E i < 0 means the plant was non-selective; P i was the ratio of weight of a type of leaf to total leaves weight; R i was the ratio of intake of snails on a type of leaf to total intake.

Selective feeding test of P. canaliculata on mangrove leaves
Fresh mangrove leaves were treated in the same way as for the nonselective feeding test. Each species of mangrove leaf weighed from 9.9 to 10.9 g. Leaves (50 g) from five species were placed together with healthy apple snails in a chamber (40 cm × 30 cm × 23 cm) filled with artificial seawater (6 L). The apple snails included in the test were 3.5 ± 0.1 m in height. The feeding test lasted for 120 h, and all the snails were alive in the chamber after the test. The artificial seawater in the chamber was changed daily to maintain water quality. Feeding amount per apple snail weight (Fa), absolute feeding ratio (Ra), feeding rate per day (Rt), and Electivity indicator (E) were used to assess the difference in food intake by the apple snails among the leaves from the five species of mangrove.

Electivity of P. canaliculata on mangrove leaves in a T-shaped Tube
The feeding electivity of the apple snail on leaves from five mangrove species was measured using a modified T-shaped tube (Takeichi et al. 2007; Figure S2). Six healthy snails 3.4 ± 0.2 cm in height were placed at one end of the tube, and mangrove leaves (9.8-10.4 g) were placed at the other two ends (right and left). Artificial seawater was added to the T-shaped tube before the test. Leaves from the five species of mangrove were placed in pairs on the left and right ends of the tube. . Paired combinations of leaves from the same mangrove species were also assessed in the T-shaped tube. After each test, the T-shaped tube was cleaned with purified water to remove leaf residue and snail metabolites to prevent possible interference with the next test. Each test lasted for 24 h. The electivity frequency (Ef) was the ratio of the number of snails recorded at one end to the total number of snails. The average Ef was used to measure the choices of snails between the same species of mangrove leaves.

Growth of P. canaliculata after feeding on mangrove leaves
The growth of P. canaliculata with mangrove leaves was assessed by examining the following indicators: weight increase ratio, survival ratio, and egg mass quantity. The fresh leaves were cleaned with a filter paper before the experiment. Each species of fresh leaf was placed separately in a plastic chamber (40 cm × 30 cm × 23 cm) together with ten healthy snails (height: 3.4 ± 0.2 cm). The growth test was performed in 6 replicates for each kind of leaf. The tested leaves (49.8-50.5 g) and the artificial seawater (6 L) in each chamber were replaced daily. The initial live weight (W1) and the final live weight (W2) of the snails were determined using an electronic balance, and the experiment lasted for 30 days. The weight increase ratio was calculated as (W2-W1)/W1 × 100%. Dead snails were removed from the chamber, and the final quantity (S) of dead snails was recorded (Martín et al. 2008). The survival ratio of the snails was calculated after 30 days using the equation of (10-S)/10 × 100%. The egg masses of the snails were collected every day, and the total quantity of egg masses was recorded as the total number of the clutches.

Determination of the chemical and physical characteristics of mangrove leaves
The contents of crude fiber, protein, total nitrogen, lignin, tannin, total phenol, wax, and chlorophyll were determined according to the published methods (Table 1). The hardness, thickness, and shape characteristics of mangrove leaves were measured using standard devices (Table 1). The hardness, thickness, and leaf shape characteristics were repeatedly measured 20 times on each leaf. Twenty mangrove leaves in total were used for each leaf characteristic.

Statistical analyses
All statistical analyses were performed in SPSS 19.0 (SPSS Inc., USA) and R software 4.0.3 (R Core Team 2020). A Levene test and Kolmogorov-Smirnov test were used to assess the homogeneity of variances and normality of the data. To compare the growth, feeding indicators (Ra, Rt, Fa, E, Ef) of P. canaliculata and leaf characteristics, we used ANOVAs (Duncan's or Tamhane's T2), nonparametric tests (Kruskal-Wallis with Dunn-Bonferroni), and T-test. To analyze the relationship between feeding indicators and leaf characteristics, we used linear fit through the function "lm" in the R package of "stats" (R Core Team 2020). To analyze correlations between typical feeding indicators and plant characteristics, we used Redundancy analyses (RDA) through the R packages of "vegan", "ggplot2", and "ggrepel" (Wickham 2009;Oksanen et al. 2019;Slowikowski 2020; R Core Team 2020). To detect the collinearity among the explanatory variables, we performed a collinearity test through the function "ordistep" of R packages "vegan" (Blanchet et al. 2008;Oksanen et al. 2019; R Core Team 2020). To analyze the differentiation of feeding electivity of P. canaliculata on mangrove leaf, we used Principal component analysis (PCA) through the function PCA, fviz_eig, in R packages of "FactoMineR" and "factoextra" (Lê et al. 2008; Kassambara and Mund 2017; R Core Team 2020).

T-tube test electivity of P. canaliculata on the mangrove leaves
The P. canaliculata exhibited significantly different values of electivity indicator (E) among the mangrove leaves ( Figure 1), with E value being highest for A. ilicifolius compared with A. aureum (p = 0.009), S. apetala (p < 0.001), K. candel (p = 0.006), and A. corniculatum (p < 0.001). The E value for A. corniculatum was lower than that of A. ilicifolius (p < 0.001), S. apetala (p < 0.001), and A. corniculatum (p < 0.001). The P. canaliculata showed a negative E value for the leaves of K. candel and A. corniculatum.
The E value of P. canaliculata for S. apetala was nearly zero. A distinct difference existed in the average electivity frequency (Ef) and pairwise Ef of P. canaliculata in the T-tube test (Figure 2). The average Ef value of P. canaliculata decreased as follows: A. ilicifolius > A. aureum > S. apetala > A. corniculatum, K. candel. There was a significant pairwise Ef value for A. ilicifolius leaves than for leaves of other species. The lowest Ef value was observed between K. candel and A. ilicifolius.
Cluster analysis showed the similarity in food electivity of P. canaliculata to the mangrove leaves (Figure 3). The five species of mangrove leaves were divided into three groups: A. ilicifolius and A. aureum; S. apetala; and A. corniculatum and K. candel. Acanthus ilicifolius and A. aureum leaves were both attractive to the apple snails, with a significant difference from other mangrove leaves.
The feeding test of the apple snail on mangrove leaves Pomacea canaliculata showed significantly different feeding behaviors on the five species of mangrove in the non-selective and selective feeding tests with respect to absolute feeding ratio (Ra), feeding rate per day (Rt), and feeding amount per snail weight (Fa) (Figure 4) based on ANOVA (Duncan's or Tamhane's T2). In the non-selective feeding test, the Fa and Rt decreased as follows: A. ilicifolius > A. aureum, K. candel > S. apetala > A. corniculatum. The Ra decreased in the following order: A. ilicifolius > A. aureum > K. candel > S. apetala > A. corniculatum (p < 0.05). The three indicators all showed that the snails fed mainly on A. ilicifolius leaves. Regarding A. ilicifolius, the Ra reached 0.80 after 120 h. The Fa reached 0.08 g/g, indicating that A. ilicifolius accounted for up to 8% of snail individual weight. The Ra was less than 0.05 when snails were fed A. corniculatum.
In the selective feeding test, the Ra decreased in the following order: A. ilicifolius > A. aureum > K. candel, S. apetala > A. corniculatum (Figure 4). The highest Rt and Fa were observed for A. ilicifolius, and higher Rt and Fa values were found for A. aureum and S. apetala than for A. corniculatum. The P. canaliculata tended to feed on A. ilicifolius leaves when leaves from different species were offered simultaneously. The patterns followed by Rt and Fa were similar to those observed in the non-selective test. The P. canaliculata still fed on very few A. corniculatum leaves.

The growth of P. canaliculata on mangrove leaves
The weight increase ratio, survival ratio, and quantity of egg masses of P. canaliculata were significantly different among the five species of mangrove ( Figure 5). Based on ANOVA with Tamhane's T2 (df = 29, F = 36.686, p < 0.001), the weight increase ratios of the snails that consumed leaves of A. aureum were higher than those of the snails that consumed leaves of K. candel (p = 0.001), S. apetala (p < 0.001), and A. corniculatum (p < 0.001). The weight increase ratios of the snails that consumed leaves of A. ilicifolius were also higher than those of the snails that consumed leaves of K. candel (p = 0.003), S. apetala (p = 0.026), and A. corniculatum (p = 0.001). Based on ANOVA with Duncan's (df = 29, F = 10.441, p < 0.001), the survival ratios of the snails that consumed A. aureum, A. ilicifolius, and S. apetala leaves were significantly higher than those of the snails that consumed K. candel and A. corniculatum leaves. Based on Kruskal-Wallis . Feeding rate per day (Rt, g d -1 g -1 ), absolute feeding ratio (Ra), feeding amount per snail weight (Fa, g g -1 ) of Pomacea canaliculata on mangrove leaves under the selective (A, C, E, red box) and the non-selective (B, D, F, yellow box) conditions. ANOVAs (Duncan's or Tamhane's T2) were used to compare the indicators among mangroves, including A. ilicifolius (Acanthus ilicifolius), A. aureum (Acrostichum aureum), K. candel (Kandelia candel), A. corniculatum (Aegiceras corniculatum), and S. apetala (Sonneratia apetala).
A significant relationship was observed between the leaf characteristics and feeding electivity indicators. There was a significant linear relationship between the chemical characteristics of the mangrove leaves, including the lignin, nitrogen, phenolic, protein, and tannin contents, and the feeding electivity indicators (Table S1). The lignin, tannin, and total phenol contents were negatively correlated with the electivity indicators, and higher r 2 values were observed for the electivity indicators in the selective test than in the non-selective test. The nitrogen and protein contents were positively correlated with all the electivity indicators. High r 2 values were observed in the two tests. The total chlorophyll content was also positively related to absolute feeding ratio and feeding rate per day in the non-selective test. The leaf area and moisture characteristics both showed a significant Figure 6. Redundancy analyses (RDA) on the correlations between feeding electivity indicators and plant characteristics. Fa and FaN (Feeding amount per snail weight in the selective and non-selective test, g g -1 ); Ra and RaN (absolute feeding ratio in the selective and non-selective test, g g -1 ); Rt and RtN (feeding rate per day in the selective and non-selective test, g d -1 g -1 ); Ef (Electivity frequency); Ei (Ivlev's Electivity indicator).
linear relationship with the feeding electivity indicators (Table S2). The aspect ratio of the mangrove leaves was significantly correlated with E, Ef, and Ra, Rt, and Fa in the non-selective test. The liner regression resulted in higher r 2 values for the moisture content than for the other characteristics.
RDA analysis was performed on significantly related traits and electivity indicators ( Figure 6). The results showed that the protein content, aspect ratio, and leaf area were positively related to the indicators of Ef, E, Ra, Rt, and Fa. The leaf area and total chlorophyll content were both positively associated with indicators RtN (feeding rate per day in the non-selective test) FaN (Feeding amount per snail weight in the non-selective test), and RaN (absolute feeding ratio in non-selective test). The lignin and wax contents were significantly negatively related to all the electivity indicators. RDA confirmed that A. ilicifolius leaves were the preferred food for P. canaliculata. A significant positive association was observed between A. ilicifolius leaves and protein content, aspect ratio, leaf area, total chlorophyll content, and the electivity indicators. A. corniculatum leaves were closely related to lignin and wax contents, making them a non-preferred food for the P. canaliculata.
PCA was performed for the significantly correlated chemical and physical characteristics of the five species of mangroves and feeding electivity indicators ( Figure 7). The two observed principal components (PCs), PC1 and PC2, explained 70% and 14% variations of the variation, and their eigenvalues both exceeded 1. PCA showed that mangrove leaves were significantly clustered into five groups, indicating that P. canaliculata can discriminate the mangrove leaves through chemical and physical characteristics.

Survival and feeding on mangrove leaves under salt stress
Although P. canaliculata is a freshwater species, a recent study confirmed that P. canaliculata can tolerate salinity stress of 4‰ with the alligator weed (Alternanthera philoxeroides) as a food source (Yang et al. 2018). Given that other Pomacea snail egg masses can survive and produce hatchlings after being periodically submerged by tides (Martin and Valentine 2014), the use of various macrophytes as a food source could alter food webs and compromise mangrove conservation efforts. The removal of dead individuals in the chamber led to a change in snail density. However, such change did not affect the feeding of remaining snails as we provided enough leaves and maintained the water quality. As a result, our study demonstrated that P. canaliculata can survive by feeding on mangrove leaves, although mangrove leaves are not as rich in nutrients as duckweed, as the latter is a common food source (Liu et al. 2012). This phenomenon was related not only to the presence of necessary nutrients in the mangrove leaves, which supported the normal metabolism of the snails, but also to the starvation tolerance of the snails, as they feed very little when only A. corniculatum leaves were provided. Pomacea canaliculata can survive for 11 months under dry conditions and 29 months under moist conditions without feeding (Yusa et al. 2006). Mangrove litter is an essential component in the carbon cycle of mangroves, and the production of litter reached 11.8 Mg•ha -1 •yr -1 (Kamruzzaman et al. 2017). As a result, the mangrove litter is a potential food source of P. canaliculata in low-salinity mangrove habitats, consistent with field observations of P. canaliculata in brackish wetlands near the shore (Letelier et al. 2016).

The feeding electivity of P. canaliculata on mangrove leaves
Little research has discussed the feeding electivity of P. canaliculata on mangrove leaves, and reports have mainly focused on freshwater macrophytes (Fang et al. 2010;Wong et al. 2010;Horgan et al. 2014). Pomacea canaliculata can compete with native benthic snails by monopolizing food resources in freshwater habitats (Fang et al. 2010). We found that P. canaliculata could feed on various mangrove leaves, indicating they have a competitive advantage over native mangrove animals, as they also selectively feed on mangrove leaves by discerning differences by smell (Fratini et al. 2001). Based on the water temperature in winter (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011) (Liu 2013), the mean winter water temperature in this region where we sampled was over 25 °C. By contrast we found that water temperature ranged from 15.5 to 25.5 °C at the sampling location in 2018. A previous study found that food consumption of P. canaliculata lowered when the water temperature decreased from 25 °C to 15 °C (Bae et al. 2021). We speculated that P. canaliculata could feed on mangrove leaves in Guangzhou wetlands but grazing may be reduced from slight low-temperature stress in winter. Considering the cold tolerance of P. canaliculata after acclimation (Yoshida et al. 2014), the P. canaliculata could outcompete native benthic animals by utilizing mangrove leaves. Benthic snails play an important role in litter decomposition in mangrove ecosystems. Adult Terebralia palustris was shown to consume 10.5 mangrove leaves at a density of 10.5 individual m -2 in a Rhizophora mucronata forest (Fratini et al. 2004). Similarly, P. canaliculata could function in litter decomposition, altering the benthic food web and accelerating the nutrient cycle. The P. canaliculata possibly showed a comprehensive impact on mangrove wetland.
The coexistence of leaves from multiple mangrove species in the selective test changed the feeding behavior of P. canaliculata. The increase in Ra (absolute feeding ratio) showed that P. canaliculata tended to feed more on the preferred food under competitive conditions. The decrease in Rt (feeding rate per day per snail) and Fa (feeding amount per snail weight) indicated that the intraspecific competition adversely affected the food consumption and digestion of each individual. A different feeding phenomenon was observed for the leaves of the least preferred species (A. corniculatum). Pomacea canaliculata had no choice but to feed on A. corniculatum leaves when leaves from other species were not provided; this finding was supported by the increased Ra value and the unchanged Rt and Fa values.
Slight differences occurred in the electivity of P. canaliculata among the three kinds of feeding tests. Pomacea. canaliculata showed a higher Ef (electivity frequency) for A. aureum than for S. apetala and K. candel. There was no significant difference in the value of the indictor E (electivity indicator) of S. apetala, A. aureum, and K. candel. The P. canaliculata did not consume all the A. ilicifolius leaves in the non-selective test, whereas the feeding ratio of A. ilicifolius in the selective test reached 100.00%. This phenomenon indicated that P. canaliculata adjusted their electivity by feeding on a substitute species according to food availability. A possible reason for this pattern was that competitive intensity varied between the feeding tests, and strong intraspecific competition existed when the leaves of the mangrove species were mixed in the selective test. Native snails or crabs commonly have distinct feeding electivity on the mangrove leaves. The mangrove snail Terebralia palustris can consume seven species of mangrove leaves; they prefer Rhizophoraceae leaves and do not consume leaves of Xilocarpus granatum (Fratini et al. 2008). The intertidal crabs Neosarmatium smithi, N. asiaticum, N. malabaricum, and Muradium tetragonum tended to collect more leaves from Bruguiera spp. and Rhizophora apiculata than from Excoecaria agallocha (Cannicci et al. 2018). We speculated that P. canaliculata may adjust its feeding electivity and conflict with benthic animals in local habitats when food sources are limited. Invasion by P. canaliculata may have led to nutritional niche differentiation among native benthic animals.

Relationship between feeding patterns and the characteristics of the mangrove leaves
There was no significant correlation between the crude fiber contents of the mangrove leaves and the electivity indicators. The strong digestive capacity of P. canaliculata was related to endogenous cellulose (Imjongjirak et al. 2008). The cellulase activity of the mangrove snail (Cerithidea cingulate) was approximately 0.1 μmol•min -1 •mg -1 , which was far lower than the average cellulase activity of the apple snail (approximately 2.0 μmol•min -1 •mg -1 ) (Chen et al. 2013;Liu et al. 2014;Luo et al. 2015;Imjongjirak et al. 2008). Moreover, the cellulase activities (0.2-0.7 μmol•min -1 •mg -1 ) of the mangrove benthic snails Cerithidea rhizophorarum, C. cingulata, Batillaria multiformis, and B. attramentaria were also significantly lower than those of P. canaliculata (Chen et al. 2013;Liu et al. 2014;Luo et al. 2015). Mangrove leaves are often rich in cellulose, and cellulose reached 10.6% in Ceriops tagal leaves (Neilson and Richards 1989). Compared to the native benthic snail, P. canaliculata had an advantage in terms of their ability to digest the cellulose in mangrove leaves, which may have led to a competitive feeding pressure against the native benthic snails.
Tannins and phenols are chemical defense substances in mangrove leaves, which have antioxidant and bacteriostatic activities (Nabeelah Bibi et al. 2019). Tannin negatively affected the feeding electivity of Sesarma plicata for mangrove leaves of K. candel, Bruguiera gymnorrhiza, and A. corniculatum (Chen and Ye 2008). We also found that the preferred leaves (A. ilicifolius) had the lowest total phenol and tannin contents, suggesting that the vulnerability of mangrove forests to P. canaliculata was closely related to mangrove species. The snails avoided or fed less on A. corniculatum leaves due to their high contents of tannin and total phenolic. However, a previous study of P. canaliculata reported no significant correlation between feeding rate and phenol content in wetland plants, in that study, the total phenol contents of the tested plants (over 75%) were 0.6-7.5% . Here, the total phenol content of the mangrove leaves was 60-140 mg•g -1 , which was higher than that in previously recorded for freshwater plants. We concluded that there was a negative correlation between the feeding electivity of P. canaliculata and the phenol contents of the mangrove leaves.
We found that P. canaliculata preferred the mangrove leaves rich in nitrogen, similar to the feeding electivity of the mangrove crabs. The feeding patterns of eight species of crabs were closely related to the nitrogen compounds in mangrove leaves (Nordhaus et al. 2011). Pomacea canaliculata preferred freshwater plants with high nitrogen contents ). Our result indicated that salinity stress did not change the demand of P. canaliculata for plant nitrogen.
There was a significantly positive correlation between the moisture of the mangrove leaves and the electivity indicators. In freshwater habitats, the feeding of P. canaliculata was negatively related to the dry matter of 21 wetland plants, indicating a positive correlation between the leaf moisture and feeding electivity . A mangrove crab (Perisesarma eumolpe) showed a preference to the high-moisture leaves of Ceriops decandra, partly due to the ability of the low-salinity fluid in these leaves to alleviate stress (Aminuddin and Arai 2017). We found that the leaf area and leaf aspect ratio both influenced the feeding of snails. Leaves with a larger area and a longer length could provide a convenient location for feeding and were conducive to adhering behavior, which indirectly promoted the feeding of P. canaliculata on the mangrove leaves.
RDA showed the feeding of P. canaliculata was strongly affected by the mangrove leaf characteristics. The five species of mangrove were clearly differentiated in the feeding test, as confirmed by PCA. The snails could distinguish among the species of leaves in the invaded habitat. These five mangroves of A. ilicifolius, A. aureum (Wang et al. 2003), Kandelia candel (Sun et al. 1998), Aegiceras corniculatum (Wei et al. 2008), Sonneratia apetala (Ren et al. 2008) are widely distributed in South China coast. We speculated that mangroves composed of A. ilicifolius and A. aureum might be vulnerable to P. canaliculata. Meanwhile, the mangrove composed of the other three species, if coexisting with A. ilicifolius or A. aureum, may also become the new invaded habitats due to the available leaf litter for the snails. The feeding electivity of P. canaliculata was both influenced by the physical and chemical characteristics of the mangrove leaves. The complex natural environment, fluctuations in water quality, and the presence of nonspecific natural enemies may cause changes in the feeding behaviors of P. canaliculata in a mangrove forest. As an omnivorous animal, Pomacea canaliculata can survive by feeding algae, phytoplankton, and plant materials (Horgan et al. 2014). Mangrove wetlands are rich in organic residues and algae, which possibly become food sources for P. canaliculata. Pomacea canaliculata could become a crucial primary consumer in the food chain of mangrove ecosystems. The spread of P. canaliculata in mangroves may pose a risk to native benthic animals by competing for food. However, the feeding behaviors of P. canaliculata could adversely improve the nutrients release of plant litter to the environment. Further studies including gut contents analysis and stable isotopes method can improve the understanding of the role of P. canaliculata in mangrove ecosystems.

Supplementary material
The following supplementary material is available for this article: Figure S1. The sampling location and the apple snails. Figure S2. T-shaped tube device used in the feeding electivity test. Table S1. Linear regression parameters of chemical traits of mangrove leaf and electivity indicators. Table S2. Linear regression parameters of physical traits of mangrove leaves and electivity indicators. Table S3. Chemical characteristics of mangrove leaf in the feeding electivity test. Table S4. Physical characteristics of mangrove leaf in the feeding electivity test.