Small in size but rather pervasive: the spread of the North American rotifer Kellicottia bostoniensis (Rousselet, 1908) through Neotropical basins

Kellicottia bostoniensis is a North American limnetic zooplankter and the single rotifer species within the group of 163 non-native species recorded in Brazilian inland waters by the Brazilian Environmental Ministry in 2016. This species is also the only non-native rotifer of the genus Kellicottia recorded in Brazilian basins. This paper reports the first occurrence of K. bostoniensis in thirteen Brazilian hydroelectric reservoirs with varying trophic levels and throughout two freshwater ecoregions. The abundance of K. bostoniensis was higher in oligotrophic reservoirs, indicating a possible preference of this species for clear waters. The distribution pattern detected in this study suggests a great spread potential within Brazilian basins, which could be due to water birds or fish gill transportation, but the exact vectors are still unknown. Moreover, its small-sized body, the presence of dormant eggs, long spines to avoid fish predation, and apparent tolerance to variable trophic conditions are the major life-history traits that might contribute to its spread and potential establishment. The high dispersal ability and overabundance of K. bostoniensis reported in our study, stress the invasive potential of this microzooplankton species and the importance of assessment programmes to evaluate possible deterioration of water quality and impacts on native plankton fauna.


Introduction
Studies of aquatic invasive species (AIS) are mainly focused on large-sized organisms, especially molluscs and fishes (Tricarico et al. 2016;Pereira et al. 2018), neglecting smaller sized groups, such as zooplanktonic organisms (Lopes et al. 1997;José de Paggi 2002;Simões et al. 2009). Since identification may follow questionable morphological and time-restricted data, the local status of non-or native species of rotifers, seems to be an arbitrary and sometimes unreliable finding. This often occurs because monitoring studies generally focus on other aquatic communities such as primary producers, benthonic and ichthyofaunal to evaluate water quality (Jeppesen et al. 2011;García-Chicote et al. 2018), neglecting the importance of zooplankton numerical abundance (García-Chicote et al. 2018) and its ecological attributes related to trophic status variation (Jeppesen et al. 2003;Pinto-Coelho et al. 2005). Moreover, invasive zooplankton species are still poorly documented due to identification problems (Peixoto et al. 2010), improperly employment of sampling methods (Macedo et al. 2018), especially regarding the micro zooplankton such as rotifers (Chick et al. 2010; Thomas et al. 2017). Thus, regarding the fact that non-native zooplankton species have the power to alter trophic cascades and impair water quality (Walsh et al. 2016), omissions of new invasion events may make difficult future ecological studies and the design of effective control and monitoring measures (Ejsmont-Karabin 2014).
Rotifers (Phylum Rotifera) are microscopic and diverse components of the zooplankton with importance in aquaculture, ecotoxicology and monitoring studies. This group holds a high diversity of life history strategies. They are known to have varied mechanisms of food acquisition, sexual to parthenogenetic reproduction, and most of them can persist as dormant stages. They are also integrators in aquatic food web inhabiting a range of systems (Wallace and Snell 2010) due to short generation times and high biomass turnover rates transferring energy from bacteria, microalgae and protists to their predators (e.g., larval fish), playing fundamental role in both microbial loop and classical aquatic food web. Generally, introduced rotifers are not known to lead negative impact on ecosystem functioning in lentic systems (O'Connor et al. 2008), and impacts via direct negative effect of this species on food web components are not conclusive (Oliveira et al. 2019). However, Kellicottia bostoniensis (Rousselet, 1908) has been interfering with the dynamics of the zooplankton community (José de Paggi 2002;Bayanov 2014;Bomfim et al. 2016), suggesting highly competitive features, local environmental suitability and effects on the use of food resources (Oliveira et al. 2019).
Although more abundant than other Cladocera and Copepoda in freshwater systems, rotifers are rarely cited as non-native or invasive in species lists (Ejsmont-Karabin 2014; Coelho and Henry 2017), where K. bostoniensis is the only rotifer cited as invasive in Brazil (Latini et al. 2016;Bomfim et al. 2016). This North American rotifer was, apparently, restricted to its native range until ~ 1943, when it was first noticed in Sweden (Carlin 1943) and then in the Netherlands (Leentvaar 1961), but now is widespread to Finland (Eloranta 1988), France (Balvay 1994), Russia (Ivanova and Telesh 2004), South America (Lopes et al. 1997;José de Paggi 2002), and Asia (Pociecha et al. 2016), including Japanese water bodies (Sudzuki and Kawakita 1999). The first record of K. bostoniensis occurred in Brazil at the Segredo Reservoir, Iguaçu river basin (Lopes et al. 1997), where its introduction was associated with the transport of dormant eggs (Crispim and Watanabe 2000;Panarelli et al. 2008;Santangelo et al. 2014;Battauz et al. 2015). From this first record until now, there are several reports of populations of K. bostoniensis in lentic and lotic systems throughout Brazil which also indicate different vectors, such as fish stocking (Peixoto et al. 2010) and ballast water (Gray et al. 2007). Nevertheless, the introduction pathways and possible vectors of this invasive rotifer are poorly understood in Brazilian waters and should be investigated in further studies, especially in reservoirs.
Reservoirs are human-modified systems highly susceptible to biological invasions caused by changes in habitat structure, hydrological regime and physical and chemical water variables (Havel et al. 2005;Johnson et al. 2008). These dammed systems are also more readily invaded than natural lakes, due to physiochemical properties, and higher levels of environmental disturbance (Havel et al. 2005). Moreover, these aquatic ecosystems can act as stepping stones to the spread of invasive species via interconnected systems and basins (Johnson et al. 2008;Pereira et al. 2018). Multiple use reservoirs are one of the most important freshwater ecosystems in Brazil, since they are ubiquitous in most of hydrographic basins and many new reservoirs are at planning phase or under construction (Couto and Olden 2018). Most Brazilian reservoirs have been impacted by several ways of human intervention, namely deforestation, uncontrolled urbanization in the surroundings, excessive use of water for irrigation and industrial purposes, nutrient loads from agricultural and untreated sewage from the drainage basin, and fish stocking based on non-native species. Considering the level of impacts of these man-made aquatic ecosystems, the lack of knowledge on invasion patterns of organisms severely hampers the ability to successfully manage current invasions and prevent new introductions (Sakalidis et al. 2013). By investigating biology and autecology of K. bostoniensis and by following the path of knowledge as to its global presence, we can possibly gain a better view of its potential as an invader in tropical lakes and reservoirs.
Currently, there are records of K. bostoniensis populations in at least fifteen reservoirs in Brazil (Lopes et al. 1997;Landa et al. 2002;Serafim Jr. et al. 2010;Bezerra-Neto et al. 2004;Casanova et al. 2009;Padovesi-Fonseca et al. 2011;Coelho and Henry 2017;De-Carli et al. 2017). Still, studies that evaluate the invasion of K. bostoniensis in wide temporal and spatial scales along the Neotropical region remain scarce, especially considering the richness and extensive variability of aquatic systems of this province. Therefore, the aim of this study is to update the current distribution and establishment status of the non-native rotifer K. bostoniensis in Brazil through the assessment of its presence and abundance from data gathered of multiple-use reservoirs, harbouring different environmental and trophic conditions.

Study area
Occurrence data was obtained from long-term monitoring studies conducted in 13 Brazilian reservoirs with different morphological features (Supplementary material Table S1). Corumbá (COR) and Itumbiara ( (Table S1) was estimated through the measure of the Secchi disk depth multiplied by 2.7 (Esteves 2011). The number of sampling sites in each reservoir was selected in accordance to its size. Specimens were stored in 200 mL glasses with 4% formaldehyde solution. Water transparency was estimated using a 40 cm diameter Secchi disk. Samples were taken using a Van Dorn bottle at the subsurface layer for analysis of concentrations of chlorophyll-a, total phosphorus (TP) and soluble reactive phosphorus (SRP). Chlorophyll-a was determined by extraction carried out in heated (70-75 °C) 90% ethanol following Nusch and Palme (1975). These variables were further used for trophic state index calculations.
Rose Bengal dye was added to samples containing high levels of inorganic material to facilitate taxonomic identification according to Koste (1978), including the trophic structure based on José de Paggi (2002). We identified and counted specimens from all 13 reservoirs using Sedgewick-Rafter chamber (1 mL) and an Olympus BX-50 optical microscope 400x. Resting eggs were identified, although not counted. We chose the reservoir with the highest abundance of K. bostoniensis to measure specimens (total body length, larger anterior spine and posterior spine lengths) for evaluation of taxonomic features. Total phosphorus and soluble reactive phosphate were analyzed by persulfate digestion according to APHA (2005). Samples for chlorophyll-a analysis were filtered with one replication in Whatman GF/C filters and extraction was carried out in cold ethanol (Nusch and Palme 1975).

Data analyses
We employed Trophic State Index (TSI) from Toledo Jr. et al. (1983) as it is a modification of Carlson's for tropical environments to analyze the trophic state of the sampled reservoirs. Thus, Secchi depth, total phosphorus, soluble reactive phosphate and chlorophyll-a were used in the estimation of the trophic conditions. Published data on TSI of the reservoirs not sampled were also retrieved to compose our trophic evaluation. A linear regression was used to investigate the relationship between the TSI and mean

Results
Kellicottia bostoniensis is largely distributed in a pool of freshwater reservoirs ( Figure 1) including lower, intermediate and higher trophic state levels (Table 1). Abundances were higher under lower trophic statuses recognizing the tropical reservoirs studied (Table 1, Figure 2) and total body size showed a significant reduction across the ascendant eutrophication considering RLA, VIG, and FUN Reservoirs ( Table 2). The highest average abundance of K. bostoniensis was recorded in RLA reservoir (940 ind. m -3 ), followed by SBR (930 ind. m -3 ), both oligotrophic systems (Table 1). Meso-eutrophic and eutrophic water bodies presented lower average abundances, TOC (20 ind. m -3 ), SAN (25 ind. m -3 ), IPO (80 ind. m -3 ), and FUN (16 ind. m -3 ). Mean abundance of K. bostoniensis was significantly and negatively correlated with the trophic state of reservoirs (p < 0.01; R² = 0.91; Figure 3). Higher abundances of K. bostoniensis were found in systems with lower trophic levels. The higher abundances were recorded in RLA and SBR, both systems having oligotrophic conditions. Resting egg densities were not estimated although appeared in large amounts,  Table 1. bound to the body or not, during counting and identification analysis at RLA samples (Figure 4). The longest total body size (366.20 to 436.29 μm) and spines (131 to 170 μm) (Table 2) were found in RLA. As the trophic level rises to Meso-Eutrophic and Eutrophic, total body length reduces to 298-325 μm and 275-308 μm, respectively in VIG and FUN. Variability in total body size and length of the anterior and caudal spine was observed among specimens in all sampled reservoirs. The presence of resting eggs among the analyzed specimens was only observed in RLA. Statistically, we found a strong negative response of mean total body size to eutrophication (p < 0.001, R 2 = 0.77) (Figure 3).

Discussion
Our study reveals that the geographic distribution of the non-native K. bostoniensis has increased in Brazil, considering the four new records for Upper Parana (reservoirs 1 to 4, Figure 1) and the first reported occurrences for Paraiba do Sul ecoregions (reservoirs 5 to 13). These findings also confirm the spread of K. bostoniensis across reservoirs of southeast Brazil, as the occurrences of this species were thought to be limited to reservoirs within the Upper Parana and Iguassu ecoregions. In comparison to Europe, the spread into Neotropical region seems to be faster and moving further north, as foreseen by José de Paggi (2002). The presence of K. bostoniensis was first reported in reservoirs and lakes in the southern part of Brazil, north of Argentina (Lopes et al. 1997;José de Paggi 2002) and later in the southeast and central part of the first country (Maia-Barbosa et al. 2008;Bomfim et al. 2016). Kellicottia bostoniensis was recently found further north, in a river in the Parnaiba ecoregion (Picapedra et al. 2017) in Northeast Brazil. The distribution may be even wider as rotifers are tiny animals with a scarcity of literature on their distribution in South America and a lack of specialists. Since it invaded large basins in Brazil, including São Francisco River (Ferraz et al. 2009) and those smaller ones such as Doce River (Maia- Barbosa et al. 2008), K. bostoniensis presents high dispersal ability compared to our mapping capacity. These were important findings, especially in reservoirs that had been sufficiently sampled in the past such as Paranoá Reservoir (Branco and Senna 1996;Starling 2000) but only revisited recently with the occurrence of K. bostoniensis (Padovesi-Fonseca et al. 2011).
This study showed that greater abundance of K. bostoniensis was found in oligotrophic reservoirs (Table 1, Figure 3), thus contrasting with the findings of Landa et al. (2002) that associated this species to more eutrophic conditions. However, as discussed by Bezerra-Neto et al. (2004), the occurrence of K. bostoniensis in reservoirs takes place in a variety of trophic statuses, possibly reflecting a high adaptability in tropical environments. The high abundances of K. bostoniensis in Ribeirão das Lajes reservoir (the most oligotrophic system) might also be related to the long-water retention time of this reservoir (300 days), since this genus is mostly related to lentic waterbodies (Bomfim et al. 2016). On the other side, the upstream Tocos reservoir might act as source for the spread of K. bostoniensis to other interconnected reservoirs in the Paraíba do Sul river basin.
K. bostoniensis was considered by José de Paggi (2002) an euryecious species, with the ability to produce dormant eggs, broadening thus its spatial and temporal propagation. Resting egg prevalence may be linked to reservoir's morphological characteristics as retention time, which is more than the double of days in RLA Reservoir (Table S1), were those were mainly found. The resting eggs usually accumulate in the sediments until the optimal conditions for hatching occur, allowing recolonization of the same environment (Hairston 1996;Cáceres 1998). However, differently from what is expected dormant eggs were found in high densities in the pelagic zone, suggesting that a representative part of the eggs are receiving the triggers needed to hatch before they reach the sediment, being wellestablished in oligotrophic water of RLA Reservoir. Moreover, Cáceres and Tessier (2004) have suggested that in permanent lakes diapause is more related to biological factors such as competition and predation. In these systems the effects of biotic interactions in structuring communities may be more relevant than abiotic factors (Shurin 2000), due their higher hydrological stability, a feature recorded for RLA Reservoir (Guarino et al. 2005). In addition, consistent planktivorous fish communities have been detected in Ribeirão das Lajes Reservoir including Kellicottia as part of their natural diet (Dias et al. 2005). Furthermore, Nielsen et al. (2000) and Ślusarczyk (2001) showed that fish predation also appeared to positively influence dormancy in zooplankton communities. Finally, in RLA Reservoir the clear transparent waters favor visual predators as fish which may intensify eggs production rates in epilimnion as an adaptive strategy for population maintenance. Higher densities of eggs in surface water may also intensify the dispersal rates of this rotifer by fish, water birds and wind (Arnemo et al. 1968;Zhdanova et al. 2016;Lopes et al. 2016).
Despite the extensive body of research published on K. bostoniensis, few studies have evaluated morphometric features of the species. When considering the maximum total body length (average 398 μm) found in this study, the size of K. bostoniensis was larger than any already reported in literature ( Table 2). The total body size recorded for K. bostoniensis in this study were larger than the measurements of José de Paggi (2002) for total body size (260-310 μm), anterior spine (90-115 μm), and posterior spine (80-120 μm) for several Reservoirs in Iguazú and Uruguay rivers in Argentina. Body-morphometric features alongside population characteristics can provide information in relation to the condition of the species and are important to confirm the success of colonization within a new environment. Zhdanova and Dobrynin (2011) associated the larger size and spines found in K. bostoniensis to the presence of the predators Asplanchna and Chaoborus larvae, and to changes in abundance and viscosity of water at greater depths. The effects of water density cannot be inferred by the present study, as all specimens were collected within the epilimnion, but the presence of predators probably explains the larger lengths observed for K. bostoniensis in RLA Reservoir. Research on major characteristics of the zooplankton community in some of the studied reservoirs recorded the common presence of Chaoborus larvae (Branco and Guarino 2012) and the rotifer Asplanchna (Rocha et al. 2002).
The specimens body size reduction in a gradient of trophy from the oligotrophic RLA Reservoir to VIG (Mesotrophic) and FUN (Eutrophic) (Figure 3) may be supported by studies that have found that feeding and growth of cladoceran species may be lowered where cyanobacteria are dominant (Lampert 1981;Elser and Goldman 1990), which may allow for the dominance by smaller zooplankton species (Lampert and Sommer 1997). Moreover, the detritivores and small particle filtrator Kellicottia bostoniensis (Pourriot 1977) was recently recognized for directly affecting the heterotrophic and autotrophic flagellates accompanied by negative effects on ciliates (Oliveira et al. 2019). Morphological variability between populations from different Russian habitats were recently analyzed in Zhdanova et al. (2019). They found that lengths of lorica and spines of rotifers from eutrophic waterbodies (N = 998, median = 342) were significantly smaller when compared to mesotrophic water bodies (N = 341, median = 375). Minimum sizes were recorded in dystrophic waterbodies but no data on morphological variability were found for oligotrophic conditions. Kellicottia bostoniensis are small-sized in South America when compared to rotifers from temperate waterbodies (Arnemo et al. 1968;Eloranta 1988;Bezerra-Neto et al. 2004). In our opinion, trophic status and biological pressures as predation by Asplanchna may be more linked to the small sizes of K. bostoniensis then the effect of physical variables such as water temperature. Thus, the studies on abundances and sizes of rotifers from tropical environments should be now intensified since differences in features of waterbodies such as trophic status, depth and temperature can be important. Finally, as the invader is functionally linked to oligotrophic or lower nutrient-rich systems, we emphasize the harmful impact of K. bostoniensis on clean freshwater ecosystems since this species can reduce the amount of matter and energy to higher trophic levels.
An increasing occurrence of K. bostoniensis in Brazilian reservoirs can be expected, regardless of the systems trophic conditions, due to: (i) ubiquity of these systems in most of the watersheds of Brazil, (ii) the apparent adaptation of K. bostoniensis to lotic and lentic conditions, (iii) tolerance to high and varying eutrophication and low content of oxygen in stratified hypolimnion, (iv) reproduction throughout the year, (v) the increase in size as a possible response to predation by invertebrates, and (vi) spreading through upstream-downstream or vice versa by waterfowls or exotic fish aquaculture. As with many other rotifers, K. bostoniensis has developed various strategies in population maintenance and establishment, which allow it to be an excellent colonizer in pelagic habitats, thus explaining their dominance in zooplankton communities in tropical freshwater bodies (Matsumura-Tundisi 1999).
Notwithstanding, wider surveys are required before further considerations on K. bostoniensis distribution and possible introduction pathways in the various aquatic ecosystems in a country as large as Brazil. Moreover, the mechanisms involved in biological invasions should be further investigated in order to foster actions on the conservation of species diversity in tropical aquatic ecosystems. This spreading into a biodiversity hotspot in the Southwest Atlantic Basins is alarming since biological invasions have been largely discussed as the second major cause of biodiversity loss (Sala et al. 2000). The scarcity of pre-invasion data in most Brazilian freshwater ecosystems and autecological studies of invasive rotifers hamper the efforts to measure the level of those impacts against a background of temporal variation and multiple stressors. Thereby, rotifers are generally pointed as not harmful (O'Connor et al. 2008), and impacts on foodweb or disruptions on trophic cascade are still inconclusive even though predation pressure of K. bostonienes on microbial loop communities have been evidenced (Oliveira et al. 2019).
Despite not being reported social-economic impacts, which are largely spread considering many aquatic invasive species (Eilers et al. 2007;Walsh et al. 2016), a possible decrease in zooplankton native populations may result in blooms of harmful toxic algae (Irigoien et al. 2005), including the non-native ones already registered and established in tropical systems e.g. Cylindrospemopsis (Branco and Senna 1991;Huszar et al. 2000). More available data on the effects of non-native species could be considered and isolated from anthropogenic perturbations, thus providing strong support for the decision-making process on avoiding more biological invasions. Finally, we also highlight the need of molecular study to infer the origins of K. bostoniensis invasions, if populations vary in genetic structure and even if new occurrences result from new donor regions.