Abstract
Understanding the diet preferences and prey selection of invasive species is crucial to better predict their impact on community structure and ecosystem functioning. Limnomysis benedeni , a Ponto-Caspian invasive mysid shrimp, is one of the most successful invaders in numerous European river and lake ecosystems. While existing studies suggest potentially strong trophic impact due to high predation pressure on invaded plankton communities, little is known of its prey selectivity between phyto- and zooplankton, under different prey concentrations. Here, we therefore investigated the feeding selectivity of L .benedeni on two commonly occurring prey organisms in freshwaters, the small rotifer zooplankton Brachionus calyciflorus together with the microphytoplankton Cryptomonas sp. present in increasing densities. Our results demonstrated a clear shift in diet selection, with L. beneden i switching from B. calyciflorus toCryptomonas sp. already when the two prey species were provided in equal biomasses. Different functional responses were observed for each prey type, indicating somewhat different foraging mechanisms for each prey type. These findings provide experimental evidence on the feeding flexibility of invasive mysid shrimps and potential implications for trophic interactions in invaded ecosystems.
Keywords: Limnomysis benedeni , prey density-dependent foraging, diet shift, invasive species, prey selectivity
Introduction
Diet shift and prey selectivity are among the most important aspects to understand the impact of a predator on community structure and food web functioning (Greene, 1983; Sailley et al., 2015). Prey selection is influenced by multiple variables such as the predator’s ontogenetic stage, prey availability and size, and environmental factors such as temperature (Divoky et al., 2021; Graeb et al., 2006; Greene, 1986). According to the optimal foraging theory, the diet preference of an animal is influenced by maximizing net energy gain and minimizing the cost of obtaining the prey (Gooding & Harley, 2015, Krebs et al., 1977). At the same time, foraging behaviour can be adjusted in response to changes in prey density (Murdoch & Oaten, 1975). Usually, predators choose habitat patches with high prey availability as it can increase the chances of successful predation events (Ioannou et al., 2009; Wellenreuther & Connell, 2002). Besides quantity, diet preference can be also influenced by the nutritional quality of their prey (Mayntz et al., 2005; Schmidt et al., 2012). Along this trade-off, consuming smaller and less nutritious prey may only be beneficial if it is present in large quantities and/or if the higher quality prey is difficult to capture (Langerhans et al., 2021). Omnivores can show higher foraging flexibility and may switch between different diet items depending on their availability. Their ability to exploit more than one trophic level (Pimm & Lawton, 1978) allow them to affect the dynamics of an ecosystem through multiple pathways. Understanding the feeding behaviours of omnivores, especially invasive species, can provide insight into their functional roles, and be used to better predict population and ecosystem dynamics.
The spread of multiple Ponto-Caspian invaders in numerous European brackish and freshwater habitats and more recently in North American Great Lakes received considerable attention in the past decades (Bij de Vaate et al., 2002; Reid & Orlova, 2002; Ricciardi & MacIsaac, 2000). The invasion was initially facilitated via unintentional means such as migration through artificial canals, reservoirs acting as stepping stones on rivers, and passive transport by ships (Jażdżewski, 1980). The expansion was further supported by intentional introduction to improve the fish food resources (Karpevich, 1975). Crustaceans, specifically, amphipods, cladocerans, and mysids are among the most successful Ponto-Caspian invaders (Leppäkoski et al., 2002; Leppäkoski & Olenin, 2001; Ricciardi & Rasmussen, 1998). They can influence native communities by eliminating native species, decreasing functional diversity, and ultimately altering the energy flow in ecosystems (Dick & Platvoet, 2000; Jazdzewski et al., 2004; Ketelaars et al., 1999; Ojaveer et al., 2002).
Mysid shrimps, including species such as Hemimysis anomala ,Limnomysis benedeni , Katamysis warpachowskyi , andParamysis lacustris , have become prominent Ponto-Caspian invaders. These species have successfully established in Europe (Audzijonyte et al., 2008, 2009; Borza et al., 2019; Wittmann, 2008), while H. anomala also invaded the North American Great Lakes (Audzijonyte et al., 2008). Traits such as wide environmental tolerance, omnivory, high annual number of generations, and related ability to get established quickly in new habitats likely contributed to their invasive success (Borza, 2014; Borza et al., 2017). Mysid shrimps are small, generally omnivorous crustaceans found in both marine and freshwater environments (Mauchline, 1980). They are grazers and predators of microorganisms and zooplankton (Viherluoto et al., 2000), and a primary food source for many fish species, birds, and other aquatic animals (Arrhenius & Hansson, 1993; Möllmann et al., 2004; Moran & Fishelson, 1971). They are facultative filter feeders, using their specialized thoracic limbs to filter plankton from the water column (Borza et al., 2023; Wittmann et al., 2014). By feeding on organisms from multiple trophic positions, and representing food for higher trophic levels, they occupy key positions in their ecosystems (Rakauskas, 2019). Therefore, their establishment and changes in abundance can have cascading effects on other organisms in the ecosystem (Kiljunen et al., 2020). To study the functional roles and impacts of invasive Ponto-Caspian mysid shrimps in invaded ecosystems, understanding their feeding interactions with native communities is therefore of high importance. Consequently, there are numerous studies exploring the feeding habits of native mysid shrimp species. For instance, omnivorous nature, role of predator and prey size, ontogenetic diet shift and seasonal migration to different aquatic zones have been explored (Arbačiauskas et al., 2013; Fink et al., 2012; Lesutienė et al., 2007; Penk & Minchin, 2014). However, limited data is available on the feeding ecology, specifically on the diet shift as a function of different prey types and densities in invasive Ponto-Caspian mysid shrimps.
Limnomysis benedeni is one of the most widespread omnivorous invasive mysids in continental Europe, including main river systems like the Danube, Rhine, neighbouring lakes and even some isolated lakes (Audzijonyte et al., 2009; Borza, 2014; Borza et al., 2011; Wittmann, 2007). L. benedeni feeds by actively selecting and capturing small-sized zooplankton, while also passively filtering smaller particles such as algal cells (Fink & Harrod, 2013). It can feed in both the benthic and pelagic zones which can affect the nutrient cycling in the littoral zone (Hanselmann et al., 2013). Earlier studies found that L. benedeni is a selective predator of zooplankton, and also suggested that prey density-dependent diet switching between phytoplankton and zooplankton may have a significant impact on plankton community structure and dynamics (Fink et al., 2012). However, the occurrence and direction of this diet shift were so far not quantified and studied in detail.
We aim to, through this study, yield a better understanding of the impact of phytoplankton and zooplankton density on the diet preference of L. benedeni . To do this, we compared the ingestion rates ofL. benedeni on food mixtures containing a rotifer and a microalga, across increasing algal biomass. We expected the mysids to feed on the larger and highly nutritious zooplankters preferentially, and only switch to phytoplankton when present at sufficiently high biomass, which consequently reduces predation on rotifers.
Materials and Methods
Study organisms and culturing conditions
Two different types of prey, a microalga and a microzooplankton species, were used in the experiment. Cryptomonas sp. (strain 26.80 of the SAG Culture Collection, originally isolated from Lake Windemere, UK) was grown in WC medium (Guillard, 1975) enriched with Vitamin B12 (0.135 g L-1 ) (recipe available at UTEX culture collection of algae, Texas website). The rotifer Brachionus calyciflorus (obtained from AQ4Aquaristics, Braunschweig, Germany) were cultivated in aerated and filtered (by a JBL Cristal Profi e702 external aquarium filter) tap water and were fed byCryptomonas sp. Individuals of the mysid shrimp Limnomysis benedeni were collected from the littoral zone of Lake Balaton, Hungary, in March 2022 by a hand net. They were subsequently reared in lab aquaria at 20℃, under a 16:8 light: dark photoperiod, and gentle aeration. The population was fed with Cryptomonas sp. at saturating concentrations (>1 mg C L-1) three times per week. Three-fourths of the volume (~3-4 litres) of the water in the aquaria was replaced with fresh filtered tap water twice a week.
Grazing experiment
In order to estimate the individual dry mass and carbon mass ofCryptomonas sp. and B. calyciflorus , we measured the axial dimensions of 20 randomly selected individuals of each species. The measurements were obtained using an inverted microscope (Zeiss Axio Vert.A1) after the individuals had been preserved with 10% Lugol’s iodine solution. The biovolume of Cryptomonas sp. was approximated by using the formula for a prolate spheroid (Hillebrand et al., 1999). Afterwards, the carbon content of the algae was estimated using the formula \(Cc\ (in\ pg/cell)=0.109.\ V^{0.991}\) where V is the biovolume in µm3(Weisse et al., 2001). The biovolume of the rotifers was estimated using the formula\(Vr\ (in\ um^{3})=\ \frac{4\Pi.l.w^{2}}{3}\) where l is the length (in µm) and w is the width of the rotifer (in µm) (Bottrell et al., 1976). The dry weight of the rotifers was calculated by assuming it to be 10% of the wet weight, which was calculated using the assumption that 1 mm3 equals 1 mg (Pace & Orcutt Jr., 1981). The dry weight was then converted to carbon content using a conversion factor of 0.48 (Work et al., 2005).
Prior to the grazing experiment, L. benedeni specimens were pre-acclimatised to the experimental conditions (temperature and light) for 48 hours. We kept them in 200 ml glass jars with 160 ml of filtered tap water in two water baths at a constant temperature of 21℃, with gentle aeration and a 16:8 light and dark photoperiod. Each jar contained three individuals of L. benedeni , fed by a mixture ofCryptomonas sp. and B. calyciflorus at saturating densities (both at 1 mg C L-1) during the first 24 hours, followed by exposing them to the experimental conditions (see below) in the next 24 hours. We applied a third water bath with the same treatments containing individuals of L. benedeni used as backup, in order to replace any dead individuals in the experimental jars to keep experimental biomass constant.
After this 48-hours pre-acclimatisation period, L. benedeniindividuals were carefully rinsed in filtered tap water to remove any debris or food attached to them. To quantify grazing rates of L. benedeni on both prey types, we run a grazing experiment consisting of four treatment levels with different algae biomasses (corresponding to 0.1, 0.5, 1, and 1.5 mg C L-1, ranging between ~ 2,000k to 32,000k cells ml-1), while rotifer biomass was kept constant (1 mg C L-1, ~ 3 rotifers ml-1) across them. This setup was replicated four times. We applied two types of controls at each treatment levels: one containing only Cryptomonas sp. to measure the growth rate of algae (applied in two replicates) and the second containing both Cryptomonas sp. and B. calyciflorusto measure the consumption rate of algae by rotifers at different algae biomass treatments (applied in three replicates). The experiment was run for 14 hours under constant dim light (4000 K LED, overall light intensity approx. 3 * 1014 photons /cm2/s). Other experimental conditions (i.e., medium, volume, temperature, aeration) were the same as described above for the pre-acclimatisation period. The few dead individuals (three) of L. benedeni were replaced with individuals from the backup cultures. At the end of the experiment, 1 ml and 10 ml samples were taken from each jar after thorough mixing to estimate the biomasses ofCryptomonas sp. and B. calyciflorus . The samples were fixed with Lugol’s iodine solution and counting was done using a stereomicroscope (Zeiss Stemi 305) using a Sedgwick rafter cell counter (S50, Graticules Optics, UK) for Cryptomonas sp. and a Petri dish with grids for B. calyciflorus .
The specific growth rates, h-1 (k ) of both prey types were calculated using formula suggested by Frost (1972):
\(\ k\ =\frac{ln(\ C2)\ -ln(\ C1)}{t2-t1\ }\ \)
The grazing coefficients, h-1 (g ) of L. benedeni were calculated by
\(g=kprey-\frac{[ln(C*2)-\ ln(C*1)]}{t2-t1}\)
where C1 and C2 are initial and final prey abundances (algal cells or rotifer individuals mL-1 ) in the control jars at timest1 and t2 andC*1 and C*2 are initial and final prey abundances (individuals mL-1) in the jars with predators at times t1 andt2 . The mysid filtration and ingestion rates of mysids were calculated using formulae by Marin et al. (1986) and Nejstgaard et al. (2001).
Data analysis
We ran multiple models, including linear models, generalized linear models, and generalized additive models to test how different algal densities affect the ingestion rates of L. benedeni onCryptomonas sp. and B. calyciflorus . Model selection was then based on the Akaike Information Criterion (AIC), which selected generalized additive models (GAM) to be the relative best-fit models (Burnham & Anderson, 2004). Ingestion rates on algae or rotifers were used as response variables, plotted against treatment levels (i.e., biomass ofCryptomonas sp.) respectively with k=4. We used the ‘anova.gam’ of the ‘mgcv’ R package to test the significance of the smooth term, treatment levels (Wood et al., 2016). Model assumptions (normality, heterogeneity of variances across treatments) were visually assessed via diagnostic plots (histogram of model residuals, residuals versus fitted values, normal Q-Q plots) and no deviations were found. All statistical tests were performed using R studio version 4.1.1 with nlme, mgcv, and ggplot2 packages (José Pinheiro et al., 2022; R Core Team, 2021; Wickham H, 2009; Wood et al., 2016). Besides, to quantify the preference of L. benedeni for different prey items with respect to their relative biomass, we calculated Ivlev’s index (ei) for each prey type across treatments (Ivlev, 1961; Jacobs, 1974). Here, a positive score indicates a preference for a particular food item while a negative score may indicate inaccessibility or avoidance of the food item (Jacobs, 1974). We also calculated and plotted the relative biomasses (mean of three replicates with 95% confidence intervals) ingested from each prey against their relative availability. Here any deviations from the 1:1 line indicate preferential feeding and a change in preference can be used to reveal prey switching (Cuthbert et al., 2018; Hughes & Croy, 1993).
Results
Ingestion rates of L. benedeni on Cryptomonas sp. increased in a slightly sigmoidal pattern with an increase in algae biomass (Fig. 1A). Parallel to this, ingestion rates on B. calyciflorus decreased following a clear sigmoidal pattern, with an abrupt decline occurring between the algae concentrations 0.5 and 1.0 mg C L-1 (Fig. 1A). The models explained 99.8% (adj. R2 = 0.998) and 97.8% (adj. R2 = 0.978) variance for algae and rotifer consumption as a function of algae biomass. Consumption of Cryptomonas sp. or B. calyciflorusby mysids were found to have a strong association with the experimental treatment levels (edf= 2.977 (for algae), p<0.001, edf= 2.948 (for rotifers), p <0.001). The total carbon biomass ingestion rate (i.e., the sum of ingested Cryptomonas sp. and B. calyciflorus carbon biomass) by L. benedeni increased with an increase in algae biomass. The relationship was non-linear, with a steeper increase above 1 mg C L-1 than below (Fig. 1B).
The Ivlev’s electivity index suggested a treatment-specific (i.g. algae biomass dependent) preference for Cryptomonas sp. and B. calyciflorous (ei>0). At lower algae biomass (0.1 mg C L-1 and 0.5 mg C L-1), index values showed a preference for B. calyciflorus (ei = 0.02 and 0.08) vs Cryptomonassp. (ei = -0.09 to -0.6). However, at higher algae biomass (1 mg C L-1 and 1.5 mg C L-1) the pattern shifted to the opposite, with higher ei values for Cryptomonas sp. (0.18 to 0.25) than for B. calyciflorous (ranging from -0.2 to -0.9).
The pattern of the relative contribution of Cryptomonas sp. andB. calyciflorus biomass to the total ingested biomass also confirmed treatment-specific preferential feeding. When rotifers were present in higher biomass than algae (i.e., 90.1% and 66.7% of total available biomass), their share in the ingested biomass was higher than could be expected in the case of non-preferential feeding, while lower in treatments with equal to or higher biomasses of Cryptomonassp. Consequently, the pattern found for Cryptomonas sp. was similar, but mirrored (Fig. 2).